---
updatedAt: 2026-07-26T13:07:51.000Z
agentTools:
  projectIndex: https://triplewhale.readme.io/llms.txt
---

# Ads Experiments Geolift Table (BETA)

ads_experiments_geolift_table

The Ads Experiments Geolift table stores geo-based incrementality (GeoLift) experiments and their measured results. One row per experiment cell.

> 📋 Table Guide
>
> **What You Can Answer With This Table**
>
> * What incrementality tests have I run and what did they find? — use `experiment_name`, `experiment_status`, `start_date`, `end_date`, `result_revenue_lift`, `result_revenue_lift_pct`, `result_revenue_iroas`, `result_revenue_p_value`
> * Which geolift experiments are currently running and when do they end? — use `experiment_name`, `experiment_status`, `start_date`, `end_date`
> * What is the incremental ROAS of the `campaigns` I tested? — use `experiment_name`, `campaigns`, `result_revenue_iroas`, `result_revenue_effect`, `result_revenue_p_value`
>
> **Before You Query**
>
> * This table is not time-series. No event date filter is needed.
> * **Grain:** One row per experiment cell (`cell_id`); `experiment_id` identifies the experiment and `experiment_set_id` groups experiments designed together.
> * **Inverse tests:** Triple Whale runs inverse tests where ad spend is reduced in the treatment geos, but all result columns are already re-oriented so positive lift means the ads add value.
> * **Sign convention:** All lift / iROAS / iCPA / effect columns are already oriented as the incremental contribution of advertising: positive = ads add value. Do not flip signs. The Compass UI shows absolute values with a separate effect-direction label, so a signed value here may appear unsigned there.
> * **Percent scales:** `spend_reduction_pct` is percentage points 0-100 (21.7 = 21.7%), while result lift percentage columns are fractions (0.05 = 5%). Multiply lift-percent columns by 100 to present them as percentages. The MDE columns are efficiency thresholds (minimum detectable iROAS for revenue MDEs; spend-per-incremental-conversion for conversions MDEs), not lift fractions.
> * **Holdout spend:** `spend_reduction` / `spend_reduction_pct` describe only the holdout (treatment) geos: the absolute column is the holdout geos' spend over the test duration, and the pct column is the holdout geos' share of the tested `campaigns`' total spend.
> * **Confidence values:** Result p-value columns hold a confidence percentage clamped to 1-99 (higher = more confident), not a raw p-value.
> * **Incomplete results:** Result columns are NULL (effect empty) until `experiment_status` is completed; filter to completed experiments when reporting results.
> * **Nested fields:** `campaigns` and `accounts` are nested (repeated) fields; use ARRAY JOIN to expand them into rows.
> * **Platform breakdown:** Per-sales-platform results other than Shopify (Amazon, TikTok Shop, Walmart, and similar) live only in the `extra_platform_results` JSON column. Shopify has dedicated columns such as `result_shopify_revenue_lift` and `result_shopify_conversions_lift`. Platform-level lifts are floored at 0 and normalized so `result_shopify_revenue_lift` plus the revenue entries in `extra_platform_results` sums exactly to `result_revenue_lift` (same for conversions).
> * **Scope:** Recommended-but-not-launched test setups are excluded; only real experiments (draft through completed) appear.
>
> **When to Use a Different Table**
>
> * Use [MMM Attributions table](https://triplewhale.readme.io/docs/mmm-attributions-table) when you need always-on modeled channel contribution rather than a one-off measured test. Geolift rows are point-in-time experiments; ongoing model-based channel contribution lives in MMM Attributions.
>
> *[View the full Triple Whale Data Ontology →](https://triplewhale.readme.io/docs/triple-whale-data-ontology)*

***

# Dimensions

Dimensions are immutable properties that can be used for grouping data.

<Table align={["left","left","left","left"]}>
  <thead>
    <tr>
      <th>
        Title
      </th>

      <th>
        ID
      </th>

      <th>
        Type
      </th>

      <th>
        Description
      </th>
    </tr>
  </thead>

  <tbody>
    <tr>
      <td>
        Ad Accounts
      </td>

      <td>
        `accounts`
      </td>

      <td>
        record repeated
      </td>

      <td>
        The ad accounts participating in the experiment. An array of objects with keys `channel` and `ad_account_id`.
      </td>
    </tr>

    <tr>
      <td>
        Channel
      </td>

      <td>
        `accounts.channel`
      </td>

      <td>
        string
      </td>

      <td>
        The ad channel of the account.

        Example values: `facebook-ads`, `google-ads`, `tiktok-ads`
      </td>
    </tr>

    <tr>
      <td>
        Ad Account ID
      </td>

      <td>
        `accounts.ad_account_id`
      </td>

      <td>
        string
      </td>

      <td>
        The identifier of the ad account on its channel.
      </td>
    </tr>

    <tr>
      <td>
        Campaign Subcategory ID
      </td>

      <td>
        `campaign_subcategory_id`
      </td>

      <td>
        string
      </td>

      <td>
        The Compass subcategory the tested campaigns belong to. `NULL` when the experiment is not tied to a subcategory.

        Example value: `9b47d2c1-6e0f-4a83-b5d9-2c71e8f4a605`
      </td>
    </tr>

    <tr>
      <td>
        Campaigns
      </td>

      <td>
        `campaigns`
      </td>

      <td>
        record repeated
      </td>

      <td>
        The campaigns whose spend is reduced in the treatment geos during the test. An array of objects with keys `campaign_id`, `campaign_name`, and `channel`.
      </td>
    </tr>

    <tr>
      <td>
        Campaign ID
      </td>

      <td>
        `campaigns.campaign_id`
      </td>

      <td>
        string
      </td>

      <td>
        The identifier of the campaign on its channel.
      </td>
    </tr>

    <tr>
      <td>
        Campaign Name
      </td>

      <td>
        `campaigns.campaign_name`
      </td>

      <td>
        string
      </td>

      <td>
        The display name of the campaign.
      </td>
    </tr>

    <tr>
      <td>
        Channel
      </td>

      <td>
        `campaigns.channel`
      </td>

      <td>
        string
      </td>

      <td>
        The ad channel of the campaign.

        Example values: `facebook-ads`, `google-ads`, `tiktok-ads`
      </td>
    </tr>

    <tr>
      <td>
        Cell ID
      </td>

      <td>
        `cell_id`
      </td>

      <td>
        string
      </td>

      <td>
        The identifier of the experiment cell (the treatment group of geos). Each row is one cell; most experiments have a single cell.

        Example value: `3b8e2f6a-91c4-4d7e-a5b0-6c2d8f4e1a92`
      </td>
    </tr>

    <tr>
      <td>
        Cooldown End Date
      </td>

      <td>
        `cooldown_period_end_date`
      </td>

      <td>
        timestamp
      </td>

      <td>
        The end of the post-treatment cooldown period. Formatted according to the <a href="https://www.iso.org/iso-8601-date-and-time-format.html" target="_blank" rel="noopener">ISO 8601 international standard</a>. `NULL` when the experiment has no cooldown.

        Example value: `2026-02-09 20:00:00`
      </td>
    </tr>

    <tr>
      <td>
        Cooldown Start Date
      </td>

      <td>
        `cooldown_period_start_date`
      </td>

      <td>
        timestamp
      </td>

      <td>
        The start of the post-treatment cooldown period, during which delayed effects are still measured. Formatted according to the <a href="https://www.iso.org/iso-8601-date-and-time-format.html" target="_blank" rel="noopener">ISO 8601 international standard</a>. `NULL` when the experiment has no cooldown.

        Example value: `2026-02-02 20:00:00`
      </td>
    </tr>

    <tr>
      <td>
        Created At
      </td>

      <td>
        `created_at`
      </td>

      <td>
        timestamp
      </td>

      <td>
        The time at which the experiment was created. Formatted according to the <a href="https://www.iso.org/iso-8601-date-and-time-format.html" target="_blank" rel="noopener">ISO 8601 international standard</a>.

        Example value: `2026-01-12 08:30:00`
      </td>
    </tr>

    <tr>
      <td>
        Currency
      </td>

      <td>
        `currency`
      </td>

      <td>
        string
      </td>

      <td>
        The currency of the monetary columns (spend reduction, revenue lifts).

        Example values: `USD`, `EUR`
      </td>
    </tr>

    <tr>
      <td>
        End Date
      </td>

      <td>
        `end_date`
      </td>

      <td>
        timestamp
      </td>

      <td>
        The date the spend reduction (treatment) ends. Formatted according to the <a href="https://www.iso.org/iso-8601-date-and-time-format.html" target="_blank" rel="noopener">ISO 8601 international standard</a>.

        Example value: `2026-02-02 20:00:00`
      </td>
    </tr>

    <tr>
      <td>
        Experiment Description
      </td>

      <td>
        `experiment_description`
      </td>

      <td>
        string
      </td>

      <td>
        The free-text description of what the experiment tests.
      </td>
    </tr>

    <tr>
      <td>
        Experiment ID
      </td>

      <td>
        `experiment_id`
      </td>

      <td>
        string
      </td>

      <td>
        The unique identifier of the geolift experiment.

        Example value: `7e3a9b21-c4d5f6a8`
      </td>
    </tr>

    <tr>
      <td>
        Experiment Name
      </td>

      <td>
        `experiment_name`
      </td>

      <td>
        string
      </td>

      <td>
        The human-readable name of the experiment.

        Example values: `Meta Prospecting Geolift Test`, `Brand Search Incrementality Test Feb 2026`, `TikTok || Retargeting Test || July 2026`
      </td>
    </tr>

    <tr>
      <td>
        Experiment Set ID
      </td>

      <td>
        `experiment_set_id`
      </td>

      <td>
        string
      </td>

      <td>
        The identifier of the experiment set the experiment belongs to. Sets group related experiments that were designed together.

        Example value: `a1f2c3d4`
      </td>
    </tr>

    <tr>
      <td>
        Experiment Status
      </td>

      <td>
        `experiment_status`
      </td>

      <td>
        string
      </td>

      <td>
        The lifecycle status of the experiment. Results (`result_*` columns) are populated once the experiment completes.

        Possible values: `draft`, `scheduled`, `running`, `cooldown`, `completed`, `cancelled`
      </td>
    </tr>

    <tr>
      <td>
        Experiment Type
      </td>

      <td>
        `experiment_type`
      </td>

      <td>
        string
      </td>

      <td>
        The type of the experiment.

        Example value: `geolift`
      </td>
    </tr>

    <tr>
      <td>
        Extra Platform Results
      </td>

      <td>
        `extra_platform_results`
      </td>

      <td>
        string
      </td>

      <td>
        A JSON-encoded string holding an array of per-sales-platform results for every platform other than Shopify (e.g. Amazon, TikTok Shop, Walmart); this is the only place per-platform results other than Shopify live.

        Each element has keys `platform`, `kpi` (`revenue` or `acquisitions`), `incremental_lift`, `lift_pct`, and optionally `iroas` or `icpa`.

        Platform values are floored at `0` (a platform with no positive contribution is reported as `0`) and normalized together with the `result_shopify_*` columns so that the Shopify lift plus these entries' lifts sum exactly to the experiment-level `result_revenue_lift` / `result_conversions_lift`.

        Per-platform effect, confidence and intervals are not provided; use the top-level result columns for experiment-level statistics. The string is `[]` (never `NULL`) when there are no extra platforms.

        Example value: `[]`
      </td>
    </tr>

    <tr>
      <td>
        Full End Date
      </td>

      <td>
        `full_end_date`
      </td>

      <td>
        timestamp
      </td>

      <td>
        The date the experiment fully ends, including the cooldown period when there is one. Formatted according to the <a href="https://www.iso.org/iso-8601-date-and-time-format.html" target="_blank" rel="noopener">ISO 8601 international standard</a>.

        Example value: `2026-02-09 20:00:00`
      </td>
    </tr>

    <tr>
      <td>
        Objective ID
      </td>

      <td>
        `objective_id`
      </td>

      <td>
        string
      </td>

      <td>
        The identifier of the experiment objective the cell was designed to measure.

        Example value: `5d21c7e9-40b3-4f8a-9c66-1e7a2b9d0f34`
      </td>
    </tr>

    <tr>
      <td>
        Pre-Experiment Confidence
      </td>

      <td>
        `pre_experiment_confidence`
      </td>

      <td>
        string
      </td>

      <td>
        The design-time assessment of how reliable the experiment is expected to be, based on the geo split and historical data quality.

        Possible values: `low`, `medium`, `high`
      </td>
    </tr>

    <tr>
      <td>
        Primary Metric
      </td>

      <td>
        `primary_metric`
      </td>

      <td>
        string
      </td>

      <td>
        The main business outcome the experiment was designed to measure.

        Example values: `Revenue`, `New Customer Revenue`, `acquisitions`
      </td>
    </tr>

    <tr>
      <td>
        Recommendation ID
      </td>

      <td>
        `recommendation_id`
      </td>

      <td>
        string
      </td>

      <td>
        The identifier of the recommended test setup this experiment was created from. Empty when the experiment was created manually.

        Example value: `4f6a2b1c-8_93e0ab`
      </td>
    </tr>

    <tr>
      <td>
        Conversions Effect Direction
      </td>

      <td>
        `result_conversions_effect`
      </td>

      <td>
        string
      </td>

      <td>
        The measured direction of the ads' effect on conversions: `positive` means the ads add value, `negative` means they detract, `no_effect` means no statistically meaningful effect was found. Empty until the experiment completes.

        Possible values: `positive`, `negative`, `no_effect`
      </td>
    </tr>

    <tr>
      <td>
        New Customer Conversions Effect Direction
      </td>

      <td>
        `result_new_customer_conversions_effect`
      </td>

      <td>
        string
      </td>

      <td>
        The measured direction of the ads' effect on new customer conversions: `positive` means the ads add value, `negative` means they detract, `no_effect` means no statistically meaningful effect was found. Empty until the experiment completes.

        Possible values: `positive`, `negative`, `no_effect`
      </td>
    </tr>

    <tr>
      <td>
        New Customer Revenue Effect Direction
      </td>

      <td>
        `result_new_customer_revenue_effect`
      </td>

      <td>
        string
      </td>

      <td>
        The measured direction of the ads' effect on new customer revenue: `positive` means the ads add value, `negative` means they detract, `no_effect` means no statistically meaningful effect was found. Empty until the experiment completes.

        Possible values: `positive`, `negative`, `no_effect`
      </td>
    </tr>

    <tr>
      <td>
        Returning Customer Conversions Effect Direction
      </td>

      <td>
        `result_returning_customer_conversions_effect`
      </td>

      <td>
        string
      </td>

      <td>
        The measured direction of the ads' effect on returning customer conversions: `positive` means the ads add value, `negative` means they detract, `no_effect` means no statistically meaningful effect was found. Empty until the experiment completes.

        Possible values: `positive`, `negative`, `no_effect`
      </td>
    </tr>

    <tr>
      <td>
        Returning Customer Revenue Effect Direction
      </td>

      <td>
        `result_returning_customer_revenue_effect`
      </td>

      <td>
        string
      </td>

      <td>
        The measured direction of the ads' effect on returning customer revenue: `positive` means the ads add value, `negative` means they detract, `no_effect` means no statistically meaningful effect was found. Empty until the experiment completes.

        Possible values: `positive`, `negative`, `no_effect`
      </td>
    </tr>

    <tr>
      <td>
        Revenue Effect Direction
      </td>

      <td>
        `result_revenue_effect`
      </td>

      <td>
        string
      </td>

      <td>
        The measured direction of the ads' effect on revenue: `positive` means the ads add value, `negative` means they detract, `no_effect` means no statistically meaningful effect was found. Empty until the experiment completes.

        Possible values: `positive`, `negative`, `no_effect`
      </td>
    </tr>

    <tr>
      <td>
        Shopify Conversions Effect Direction
      </td>

      <td>
        `result_shopify_conversions_effect`
      </td>

      <td>
        string
      </td>

      <td>
        The measured direction of the ads' effect on shopify conversions: `positive` means the ads add value, `negative` means they detract, `no_effect` means no statistically meaningful effect was found. Empty until the experiment completes.

        Possible values: `positive`, `negative`, `no_effect`
      </td>
    </tr>

    <tr>
      <td>
        Shopify Revenue Effect Direction
      </td>

      <td>
        `result_shopify_revenue_effect`
      </td>

      <td>
        string
      </td>

      <td>
        The measured direction of the ads' effect on shopify revenue: `positive` means the ads add value, `negative` means they detract, `no_effect` means no statistically meaningful effect was found. Empty until the experiment completes.

        Possible values: `positive`, `negative`, `no_effect`
      </td>
    </tr>

    <tr>
      <td>
        Start Date
      </td>

      <td>
        `start_date`
      </td>

      <td>
        timestamp
      </td>

      <td>
        The date the spend reduction (treatment) starts. Formatted according to the <a href="https://www.iso.org/iso-8601-date-and-time-format.html" target="_blank" rel="noopener">ISO 8601 international standard</a>.

        Example value: `2026-01-12 20:00:00`
      </td>
    </tr>

    <tr>
      <td>
        Updated At
      </td>

      <td>
        `updated_at`
      </td>

      <td>
        timestamp
      </td>

      <td>
        The time at which the experiment row was last updated. Formatted according to the <a href="https://www.iso.org/iso-8601-date-and-time-format.html" target="_blank" rel="noopener">ISO 8601 international standard</a>.

        Example value: `2026-01-12 08:30:00`
      </td>
    </tr>
  </tbody>
</Table>

# Measures

Measures are numeric fields that can be aggregated and/or combined to calculate new metrics.

<Table align={["left","left","left","left"]}>
  <thead>
    <tr>
      <th>
        Title
      </th>

      <th>
        ID
      </th>

      <th>
        Type
      </th>

      <th>
        Description
      </th>
    </tr>
  </thead>

  <tbody>
    <tr>
      <td>
        Conversions MDE
      </td>

      <td>
        `conversions_mde`
      </td>

      <td>
        numeric
      </td>

      <td>
        The minimum detectable cost per incremental conversion for Conversions: holdout spend divided by the minimum number of incremental conversions this experiment's design can reliably detect (an iCPA-style threshold). Set at design time. A completed test whose true effect is below this threshold will typically report `no_effect`.
      </td>
    </tr>

    <tr>
      <td>
        Estimated KPI Impact %
      </td>

      <td>
        `estimated_kpi_impact_pct`
      </td>

      <td>
        numeric
      </td>

      <td>
        The design-time estimate of the experiment's impact on the primary metric. Rarely populated; `0` for most experiments, so do not rely on it.
      </td>
    </tr>

    <tr>
      <td>
        New Customer Conversions MDE
      </td>

      <td>
        `new_customer_conversions_mde`
      </td>

      <td>
        numeric
      </td>

      <td>
        The minimum detectable cost per incremental conversion for New Customer Conversions: holdout spend divided by the minimum number of incremental conversions this experiment's design can reliably detect (an iCPA-style threshold). Set at design time. A completed test whose true effect is below this threshold will typically report `no_effect`.
      </td>
    </tr>

    <tr>
      <td>
        New Customer Revenue MDE
      </td>

      <td>
        `new_customer_revenue_mde`
      </td>

      <td>
        numeric
      </td>

      <td>
        The minimum detectable iROAS for New Customer Revenue: the smallest incremental return on ad spend this experiment's design can reliably detect (minimum required incremental new customer revenue implied by the power analysis, divided by holdout spend), floored at `0.15`. Set at design time. A completed test whose true effect is below this threshold will typically report `no_effect`.
      </td>
    </tr>

    <tr>
      <td>
        Conversions First Click CPA
      </td>

      <td>
        `result_conversions_first_click_cpa`
      </td>

      <td>
        numeric
      </td>

      <td>
        The conversions CPA that the First Click attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iCPA can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Conversions Incremental CPA
      </td>

      <td>
        `result_conversions_icpa`
      </td>

      <td>
        numeric
      </td>

      <td>
        The incremental cost per acquisition for conversions: spend on the tested ads per incremental conversion. Compare with the attribution-based CPA columns to see how attribution over- or under-states incremental value. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Conversions Last Click CPA
      </td>

      <td>
        `result_conversions_last_click_cpa`
      </td>

      <td>
        numeric
      </td>

      <td>
        The conversions CPA that the Last Click attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iCPA can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Incremental Conversions Lift
      </td>

      <td>
        `result_conversions_lift`
      </td>

      <td>
        numeric
      </td>

      <td>
        The absolute incremental conversions attributed to the tested ads over the experiment window. Positive means the ads add value. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Incremental Conversions Lift Lower Bound
      </td>

      <td>
        `result_conversions_lift_lower_bound`
      </td>

      <td>
        numeric
      </td>

      <td>
        The lower bound of the confidence interval around the incremental conversions lift. `NULL` when no interval is available.
      </td>
    </tr>

    <tr>
      <td>
        Incremental Conversions Lift %
      </td>

      <td>
        `result_conversions_lift_pct`
      </td>

      <td>
        numeric
      </td>

      <td>
        The incremental conversions lift as a fraction of the baseline (e.g. `0.05` = 5% lift). Positive means the ads add value. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Incremental Conversions Lift Upper Bound
      </td>

      <td>
        `result_conversions_lift_upper_bound`
      </td>

      <td>
        numeric
      </td>

      <td>
        The upper bound of the confidence interval around the incremental conversions lift. `NULL` when no interval is available.
      </td>
    </tr>

    <tr>
      <td>
        Conversions Linear All CPA
      </td>

      <td>
        `result_conversions_linear_all_cpa`
      </td>

      <td>
        numeric
      </td>

      <td>
        The conversions CPA that the Linear All attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iCPA can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Conversions Linear Paid CPA
      </td>

      <td>
        `result_conversions_linear_paid_cpa`
      </td>

      <td>
        numeric
      </td>

      <td>
        The conversions CPA that the Linear Paid attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iCPA can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Conversions Confidence
      </td>

      <td>
        `result_conversions_p_value`
      </td>

      <td>
        numeric
      </td>

      <td>
        The statistical confidence in the measured conversions effect, expressed as a percentage clamped to `1`–`99` (higher = more confident; this is NOT a raw p-value). `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Conversions Triple Attribution CPA
      </td>

      <td>
        `result_conversions_triple_attribution_cpa`
      </td>

      <td>
        numeric
      </td>

      <td>
        The conversions CPA that the <a href="https://kb.triplewhale.com/en/articles/5960333-understanding-and-utilizing-attribution-models#h_c07dd1dffd" target="_blank" rel="noopener">Triple Attribution</a> attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iCPA can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Conversions Triple Attribution + Views CPA
      </td>

      <td>
        `result_conversions_triple_attribution_plus_views_cpa`
      </td>

      <td>
        numeric
      </td>

      <td>
        The conversions CPA that the <a href="https://kb.triplewhale.com/en/articles/5960333-understanding-and-utilizing-attribution-models#h_c07dd1dffd" target="_blank" rel="noopener">Triple Attribution</a> + Views attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iCPA can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        New Customer Conversions First Click CPA
      </td>

      <td>
        `result_new_customer_conversions_first_click_cpa`
      </td>

      <td>
        numeric
      </td>

      <td>
        The new customer conversions CPA that the First Click attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iCPA can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        New Customer Conversions Incremental CPA
      </td>

      <td>
        `result_new_customer_conversions_icpa`
      </td>

      <td>
        numeric
      </td>

      <td>
        The incremental cost per acquisition for new customer conversions: spend on the tested ads per incremental conversion. Compare with the attribution-based CPA columns to see how attribution over- or under-states incremental value. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        New Customer Conversions Last Click CPA
      </td>

      <td>
        `result_new_customer_conversions_last_click_cpa`
      </td>

      <td>
        numeric
      </td>

      <td>
        The new customer conversions CPA that the Last Click attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iCPA can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Incremental New Customer Conversions Lift
      </td>

      <td>
        `result_new_customer_conversions_lift`
      </td>

      <td>
        numeric
      </td>

      <td>
        The absolute incremental new customer conversions attributed to the tested ads over the experiment window. Positive means the ads add value. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Incremental New Customer Conversions Lift Lower Bound
      </td>

      <td>
        `result_new_customer_conversions_lift_lower_bound`
      </td>

      <td>
        numeric
      </td>

      <td>
        The lower bound of the confidence interval around the incremental new customer conversions lift. `NULL` when no interval is available.
      </td>
    </tr>

    <tr>
      <td>
        Incremental New Customer Conversions Lift %
      </td>

      <td>
        `result_new_customer_conversions_lift_pct`
      </td>

      <td>
        numeric
      </td>

      <td>
        The incremental new customer conversions lift as a fraction of the baseline (e.g. `0.05` = 5% lift). Positive means the ads add value. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Incremental New Customer Conversions Lift Upper Bound
      </td>

      <td>
        `result_new_customer_conversions_lift_upper_bound`
      </td>

      <td>
        numeric
      </td>

      <td>
        The upper bound of the confidence interval around the incremental new customer conversions lift. `NULL` when no interval is available.
      </td>
    </tr>

    <tr>
      <td>
        New Customer Conversions Linear All CPA
      </td>

      <td>
        `result_new_customer_conversions_linear_all_cpa`
      </td>

      <td>
        numeric
      </td>

      <td>
        The new customer conversions CPA that the Linear All attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iCPA can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        New Customer Conversions Linear Paid CPA
      </td>

      <td>
        `result_new_customer_conversions_linear_paid_cpa`
      </td>

      <td>
        numeric
      </td>

      <td>
        The new customer conversions CPA that the Linear Paid attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iCPA can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        New Customer Conversions Confidence
      </td>

      <td>
        `result_new_customer_conversions_p_value`
      </td>

      <td>
        numeric
      </td>

      <td>
        The statistical confidence in the measured new customer conversions effect, expressed as a percentage clamped to `1`–`99` (higher = more confident; this is NOT a raw p-value). `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        New Customer Conversions Triple Attribution CPA
      </td>

      <td>
        `result_new_customer_conversions_triple_attribution_cpa`
      </td>

      <td>
        numeric
      </td>

      <td>
        The new customer conversions CPA that the <a href="https://kb.triplewhale.com/en/articles/5960333-understanding-and-utilizing-attribution-models#h_c07dd1dffd" target="_blank" rel="noopener">Triple Attribution</a> attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iCPA can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        New Customer Conversions Triple Attribution + Views CPA
      </td>

      <td>
        `result_new_customer_conversions_triple_attribution_plus_views_cpa`
      </td>

      <td>
        numeric
      </td>

      <td>
        The new customer conversions CPA that the <a href="https://kb.triplewhale.com/en/articles/5960333-understanding-and-utilizing-attribution-models#h_c07dd1dffd" target="_blank" rel="noopener">Triple Attribution</a> + Views attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iCPA can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        New Customer Revenue First Click ROAS
      </td>

      <td>
        `result_new_customer_revenue_first_click_roas`
      </td>

      <td>
        numeric
      </td>

      <td>
        The new customer revenue ROAS that the First Click attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iROAS can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        New Customer Revenue Incremental ROAS
      </td>

      <td>
        `result_new_customer_revenue_iroas`
      </td>

      <td>
        numeric
      </td>

      <td>
        The incremental return on ad spend for new customer revenue: incremental new customer revenue attributed to the tested ads per dollar of spend. Compare with the attribution-based ROAS columns to see how attribution over- or under-states incremental value. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        New Customer Revenue Last Click ROAS
      </td>

      <td>
        `result_new_customer_revenue_last_click_roas`
      </td>

      <td>
        numeric
      </td>

      <td>
        The new customer revenue ROAS that the Last Click attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iROAS can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Incremental New Customer Revenue Lift
      </td>

      <td>
        `result_new_customer_revenue_lift`
      </td>

      <td>
        numeric
      </td>

      <td>
        The absolute incremental new customer revenue attributed to the tested ads over the experiment window. Positive means the ads add value. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Incremental New Customer Revenue Lift Lower Bound
      </td>

      <td>
        `result_new_customer_revenue_lift_lower_bound`
      </td>

      <td>
        numeric
      </td>

      <td>
        The lower bound of the confidence interval around the incremental new customer revenue lift. `NULL` when no interval is available.
      </td>
    </tr>

    <tr>
      <td>
        Incremental New Customer Revenue Lift %
      </td>

      <td>
        `result_new_customer_revenue_lift_pct`
      </td>

      <td>
        numeric
      </td>

      <td>
        The incremental new customer revenue lift as a fraction of the baseline (e.g. `0.05` = 5% lift). Positive means the ads add value. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Incremental New Customer Revenue Lift Upper Bound
      </td>

      <td>
        `result_new_customer_revenue_lift_upper_bound`
      </td>

      <td>
        numeric
      </td>

      <td>
        The upper bound of the confidence interval around the incremental new customer revenue lift. `NULL` when no interval is available.
      </td>
    </tr>

    <tr>
      <td>
        New Customer Revenue Linear All ROAS
      </td>

      <td>
        `result_new_customer_revenue_linear_all_roas`
      </td>

      <td>
        numeric
      </td>

      <td>
        The new customer revenue ROAS that the Linear All attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iROAS can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        New Customer Revenue Linear Paid ROAS
      </td>

      <td>
        `result_new_customer_revenue_linear_paid_roas`
      </td>

      <td>
        numeric
      </td>

      <td>
        The new customer revenue ROAS that the Linear Paid attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iROAS can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        New Customer Revenue Confidence
      </td>

      <td>
        `result_new_customer_revenue_p_value`
      </td>

      <td>
        numeric
      </td>

      <td>
        The statistical confidence in the measured new customer revenue effect, expressed as a percentage clamped to `1`–`99` (higher = more confident; this is NOT a raw p-value). `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        New Customer Revenue Triple Attribution + Views ROAS
      </td>

      <td>
        `result_new_customer_revenue_triple_attribution_plus_views_roas`
      </td>

      <td>
        numeric
      </td>

      <td>
        The new customer revenue ROAS that the <a href="https://kb.triplewhale.com/en/articles/5960333-understanding-and-utilizing-attribution-models#h_c07dd1dffd" target="_blank" rel="noopener">Triple Attribution</a> + Views attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iROAS can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        New Customer Revenue Triple Attribution ROAS
      </td>

      <td>
        `result_new_customer_revenue_triple_attribution_roas`
      </td>

      <td>
        numeric
      </td>

      <td>
        The new customer revenue ROAS that the <a href="https://kb.triplewhale.com/en/articles/5960333-understanding-and-utilizing-attribution-models#h_c07dd1dffd" target="_blank" rel="noopener">Triple Attribution</a> attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iROAS can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Returning Customer Conversions First Click CPA
      </td>

      <td>
        `result_returning_customer_conversions_first_click_cpa`
      </td>

      <td>
        numeric
      </td>

      <td>
        The returning customer conversions CPA that the First Click attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iCPA can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Returning Customer Conversions Incremental CPA
      </td>

      <td>
        `result_returning_customer_conversions_icpa`
      </td>

      <td>
        numeric
      </td>

      <td>
        The incremental cost per acquisition for returning customer conversions: spend on the tested ads per incremental conversion. Compare with the attribution-based CPA columns to see how attribution over- or under-states incremental value. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Returning Customer Conversions Last Click CPA
      </td>

      <td>
        `result_returning_customer_conversions_last_click_cpa`
      </td>

      <td>
        numeric
      </td>

      <td>
        The returning customer conversions CPA that the Last Click attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iCPA can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Incremental Returning Customer Conversions Lift
      </td>

      <td>
        `result_returning_customer_conversions_lift`
      </td>

      <td>
        numeric
      </td>

      <td>
        The absolute incremental returning customer conversions attributed to the tested ads over the experiment window. Positive means the ads add value. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Incremental Returning Customer Conversions Lift Lower Bound
      </td>

      <td>
        `result_returning_customer_conversions_lift_lower_bound`
      </td>

      <td>
        numeric
      </td>

      <td>
        The lower bound of the confidence interval around the incremental returning customer conversions lift. `NULL` when no interval is available.
      </td>
    </tr>

    <tr>
      <td>
        Incremental Returning Customer Conversions Lift %
      </td>

      <td>
        `result_returning_customer_conversions_lift_pct`
      </td>

      <td>
        numeric
      </td>

      <td>
        The incremental returning customer conversions lift as a fraction of the baseline (e.g. `0.05` = 5% lift). Positive means the ads add value. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Incremental Returning Customer Conversions Lift Upper Bound
      </td>

      <td>
        `result_returning_customer_conversions_lift_upper_bound`
      </td>

      <td>
        numeric
      </td>

      <td>
        The upper bound of the confidence interval around the incremental returning customer conversions lift. `NULL` when no interval is available.
      </td>
    </tr>

    <tr>
      <td>
        Returning Customer Conversions Linear All CPA
      </td>

      <td>
        `result_returning_customer_conversions_linear_all_cpa`
      </td>

      <td>
        numeric
      </td>

      <td>
        The returning customer conversions CPA that the Linear All attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iCPA can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Returning Customer Conversions Linear Paid CPA
      </td>

      <td>
        `result_returning_customer_conversions_linear_paid_cpa`
      </td>

      <td>
        numeric
      </td>

      <td>
        The returning customer conversions CPA that the Linear Paid attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iCPA can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Returning Customer Conversions Confidence
      </td>

      <td>
        `result_returning_customer_conversions_p_value`
      </td>

      <td>
        numeric
      </td>

      <td>
        The statistical confidence in the measured returning customer conversions effect, expressed as a percentage clamped to `1`–`99` (higher = more confident; this is NOT a raw p-value). `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Returning Customer Conversions Triple Attribution CPA
      </td>

      <td>
        `result_returning_customer_conversions_triple_attribution_cpa`
      </td>

      <td>
        numeric
      </td>

      <td>
        The returning customer conversions CPA that the <a href="https://kb.triplewhale.com/en/articles/5960333-understanding-and-utilizing-attribution-models#h_c07dd1dffd" target="_blank" rel="noopener">Triple Attribution</a> attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iCPA can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Returning Customer Conversions Triple Attribution + Views CPA
      </td>

      <td>
        `result_returning_customer_conversions_triple_attribution_plus_views_cpa`
      </td>

      <td>
        numeric
      </td>

      <td>
        The returning customer conversions CPA that the <a href="https://kb.triplewhale.com/en/articles/5960333-understanding-and-utilizing-attribution-models#h_c07dd1dffd" target="_blank" rel="noopener">Triple Attribution</a> + Views attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iCPA can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Returning Customer Revenue First Click ROAS
      </td>

      <td>
        `result_returning_customer_revenue_first_click_roas`
      </td>

      <td>
        numeric
      </td>

      <td>
        The returning customer revenue ROAS that the First Click attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iROAS can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Returning Customer Revenue Incremental ROAS
      </td>

      <td>
        `result_returning_customer_revenue_iroas`
      </td>

      <td>
        numeric
      </td>

      <td>
        The incremental return on ad spend for returning customer revenue: incremental returning customer revenue attributed to the tested ads per dollar of spend. Compare with the attribution-based ROAS columns to see how attribution over- or under-states incremental value. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Returning Customer Revenue Last Click ROAS
      </td>

      <td>
        `result_returning_customer_revenue_last_click_roas`
      </td>

      <td>
        numeric
      </td>

      <td>
        The returning customer revenue ROAS that the Last Click attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iROAS can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Incremental Returning Customer Revenue Lift
      </td>

      <td>
        `result_returning_customer_revenue_lift`
      </td>

      <td>
        numeric
      </td>

      <td>
        The absolute incremental returning customer revenue attributed to the tested ads over the experiment window. Positive means the ads add value. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Incremental Returning Customer Revenue Lift Lower Bound
      </td>

      <td>
        `result_returning_customer_revenue_lift_lower_bound`
      </td>

      <td>
        numeric
      </td>

      <td>
        The lower bound of the confidence interval around the incremental returning customer revenue lift. `NULL` when no interval is available.
      </td>
    </tr>

    <tr>
      <td>
        Incremental Returning Customer Revenue Lift %
      </td>

      <td>
        `result_returning_customer_revenue_lift_pct`
      </td>

      <td>
        numeric
      </td>

      <td>
        The incremental returning customer revenue lift as a fraction of the baseline (e.g. `0.05` = 5% lift). Positive means the ads add value. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Incremental Returning Customer Revenue Lift Upper Bound
      </td>

      <td>
        `result_returning_customer_revenue_lift_upper_bound`
      </td>

      <td>
        numeric
      </td>

      <td>
        The upper bound of the confidence interval around the incremental returning customer revenue lift. `NULL` when no interval is available.
      </td>
    </tr>

    <tr>
      <td>
        Returning Customer Revenue Linear All ROAS
      </td>

      <td>
        `result_returning_customer_revenue_linear_all_roas`
      </td>

      <td>
        numeric
      </td>

      <td>
        The returning customer revenue ROAS that the Linear All attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iROAS can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Returning Customer Revenue Linear Paid ROAS
      </td>

      <td>
        `result_returning_customer_revenue_linear_paid_roas`
      </td>

      <td>
        numeric
      </td>

      <td>
        The returning customer revenue ROAS that the Linear Paid attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iROAS can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Returning Customer Revenue Confidence
      </td>

      <td>
        `result_returning_customer_revenue_p_value`
      </td>

      <td>
        numeric
      </td>

      <td>
        The statistical confidence in the measured returning customer revenue effect, expressed as a percentage clamped to `1`–`99` (higher = more confident; this is NOT a raw p-value). `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Returning Customer Revenue Triple Attribution + Views ROAS
      </td>

      <td>
        `result_returning_customer_revenue_triple_attribution_plus_views_roas`
      </td>

      <td>
        numeric
      </td>

      <td>
        The returning customer revenue ROAS that the <a href="https://kb.triplewhale.com/en/articles/5960333-understanding-and-utilizing-attribution-models#h_c07dd1dffd" target="_blank" rel="noopener">Triple Attribution</a> + Views attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iROAS can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Returning Customer Revenue Triple Attribution ROAS
      </td>

      <td>
        `result_returning_customer_revenue_triple_attribution_roas`
      </td>

      <td>
        numeric
      </td>

      <td>
        The returning customer revenue ROAS that the <a href="https://kb.triplewhale.com/en/articles/5960333-understanding-and-utilizing-attribution-models#h_c07dd1dffd" target="_blank" rel="noopener">Triple Attribution</a> attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iROAS can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Revenue First Click ROAS
      </td>

      <td>
        `result_revenue_first_click_roas`
      </td>

      <td>
        numeric
      </td>

      <td>
        The revenue ROAS that the First Click attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iROAS can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Revenue Incremental ROAS
      </td>

      <td>
        `result_revenue_iroas`
      </td>

      <td>
        numeric
      </td>

      <td>
        The incremental return on ad spend for revenue: incremental revenue attributed to the tested ads per dollar of spend. Compare with the attribution-based ROAS columns to see how attribution over- or under-states incremental value. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Revenue Last Click ROAS
      </td>

      <td>
        `result_revenue_last_click_roas`
      </td>

      <td>
        numeric
      </td>

      <td>
        The revenue ROAS that the Last Click attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iROAS can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Incremental Revenue Lift
      </td>

      <td>
        `result_revenue_lift`
      </td>

      <td>
        numeric
      </td>

      <td>
        The absolute incremental revenue attributed to the tested ads over the experiment window. Positive means the ads add value. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Incremental Revenue Lift Lower Bound
      </td>

      <td>
        `result_revenue_lift_lower_bound`
      </td>

      <td>
        numeric
      </td>

      <td>
        The lower bound of the confidence interval around the incremental revenue lift. `NULL` when no interval is available.
      </td>
    </tr>

    <tr>
      <td>
        Incremental Revenue Lift %
      </td>

      <td>
        `result_revenue_lift_pct`
      </td>

      <td>
        numeric
      </td>

      <td>
        The incremental revenue lift as a fraction of the baseline (e.g. `0.05` = 5% lift). Positive means the ads add value. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Incremental Revenue Lift Upper Bound
      </td>

      <td>
        `result_revenue_lift_upper_bound`
      </td>

      <td>
        numeric
      </td>

      <td>
        The upper bound of the confidence interval around the incremental revenue lift. `NULL` when no interval is available.
      </td>
    </tr>

    <tr>
      <td>
        Revenue Linear All ROAS
      </td>

      <td>
        `result_revenue_linear_all_roas`
      </td>

      <td>
        numeric
      </td>

      <td>
        The revenue ROAS that the Linear All attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iROAS can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Revenue Linear Paid ROAS
      </td>

      <td>
        `result_revenue_linear_paid_roas`
      </td>

      <td>
        numeric
      </td>

      <td>
        The revenue ROAS that the Linear Paid attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iROAS can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Revenue Confidence
      </td>

      <td>
        `result_revenue_p_value`
      </td>

      <td>
        numeric
      </td>

      <td>
        The statistical confidence in the measured revenue effect, expressed as a percentage clamped to `1`–`99` (higher = more confident; this is NOT a raw p-value). `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Revenue Triple Attribution + Views ROAS
      </td>

      <td>
        `result_revenue_triple_attribution_plus_views_roas`
      </td>

      <td>
        numeric
      </td>

      <td>
        The revenue ROAS that the <a href="https://kb.triplewhale.com/en/articles/5960333-understanding-and-utilizing-attribution-models#h_c07dd1dffd" target="_blank" rel="noopener">Triple Attribution</a> + Views attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iROAS can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Revenue Triple Attribution ROAS
      </td>

      <td>
        `result_revenue_triple_attribution_roas`
      </td>

      <td>
        numeric
      </td>

      <td>
        The revenue ROAS that the <a href="https://kb.triplewhale.com/en/articles/5960333-understanding-and-utilizing-attribution-models#h_c07dd1dffd" target="_blank" rel="noopener">Triple Attribution</a> attribution model reported for the tested campaigns over the experiment window. This is a plain attribution ratio (always positive), included so the experiment's incremental iROAS can be compared against what attribution reports. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Shopify Conversions Incremental CPA
      </td>

      <td>
        `result_shopify_conversions_icpa`
      </td>

      <td>
        numeric
      </td>

      <td>
        The incremental cost per acquisition for shopify conversions: spend on the tested ads per incremental conversion. Compare with the attribution-based CPA columns to see how attribution over- or under-states incremental value. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Incremental Shopify Conversions Lift
      </td>

      <td>
        `result_shopify_conversions_lift`
      </td>

      <td>
        numeric
      </td>

      <td>
        The absolute incremental shopify conversions attributed to the tested ads over the experiment window. Positive means the ads add value. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Incremental Shopify Conversions Lift Lower Bound
      </td>

      <td>
        `result_shopify_conversions_lift_lower_bound`
      </td>

      <td>
        numeric
      </td>

      <td>
        The lower bound of the confidence interval around the incremental shopify conversions lift. `NULL` when no interval is available.
      </td>
    </tr>

    <tr>
      <td>
        Incremental Shopify Conversions Lift %
      </td>

      <td>
        `result_shopify_conversions_lift_pct`
      </td>

      <td>
        numeric
      </td>

      <td>
        The incremental shopify conversions lift as a fraction of the baseline (e.g. `0.05` = 5% lift). Positive means the ads add value. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Incremental Shopify Conversions Lift Upper Bound
      </td>

      <td>
        `result_shopify_conversions_lift_upper_bound`
      </td>

      <td>
        numeric
      </td>

      <td>
        The upper bound of the confidence interval around the incremental shopify conversions lift. `NULL` when no interval is available.
      </td>
    </tr>

    <tr>
      <td>
        Shopify Conversions Confidence
      </td>

      <td>
        `result_shopify_conversions_p_value`
      </td>

      <td>
        numeric
      </td>

      <td>
        The statistical confidence in the measured shopify conversions effect, expressed as a percentage clamped to `1`–`99` (higher = more confident; this is NOT a raw p-value). `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Shopify Revenue Incremental ROAS
      </td>

      <td>
        `result_shopify_revenue_iroas`
      </td>

      <td>
        numeric
      </td>

      <td>
        The incremental return on ad spend for shopify revenue: incremental shopify revenue attributed to the tested ads per dollar of spend. Compare with the attribution-based ROAS columns to see how attribution over- or under-states incremental value. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Incremental Shopify Revenue Lift
      </td>

      <td>
        `result_shopify_revenue_lift`
      </td>

      <td>
        numeric
      </td>

      <td>
        The absolute incremental shopify revenue attributed to the tested ads over the experiment window. Positive means the ads add value. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Incremental Shopify Revenue Lift Lower Bound
      </td>

      <td>
        `result_shopify_revenue_lift_lower_bound`
      </td>

      <td>
        numeric
      </td>

      <td>
        The lower bound of the confidence interval around the incremental shopify revenue lift. `NULL` when no interval is available.
      </td>
    </tr>

    <tr>
      <td>
        Incremental Shopify Revenue Lift %
      </td>

      <td>
        `result_shopify_revenue_lift_pct`
      </td>

      <td>
        numeric
      </td>

      <td>
        The incremental shopify revenue lift as a fraction of the baseline (e.g. `0.05` = 5% lift). Positive means the ads add value. `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Incremental Shopify Revenue Lift Upper Bound
      </td>

      <td>
        `result_shopify_revenue_lift_upper_bound`
      </td>

      <td>
        numeric
      </td>

      <td>
        The upper bound of the confidence interval around the incremental shopify revenue lift. `NULL` when no interval is available.
      </td>
    </tr>

    <tr>
      <td>
        Shopify Revenue Confidence
      </td>

      <td>
        `result_shopify_revenue_p_value`
      </td>

      <td>
        numeric
      </td>

      <td>
        The statistical confidence in the measured shopify revenue effect, expressed as a percentage clamped to `1`–`99` (higher = more confident; this is NOT a raw p-value). `NULL` until the experiment completes.
      </td>
    </tr>

    <tr>
      <td>
        Returning Customer Conversions MDE
      </td>

      <td>
        `returning_customer_conversions_mde`
      </td>

      <td>
        numeric
      </td>

      <td>
        The minimum detectable cost per incremental conversion for Returning Customer Conversions: holdout spend divided by the minimum number of incremental conversions this experiment's design can reliably detect (an iCPA-style threshold). Set at design time. A completed test whose true effect is below this threshold will typically report `no_effect`.
      </td>
    </tr>

    <tr>
      <td>
        Returning Customer Revenue MDE
      </td>

      <td>
        `returning_customer_revenue_mde`
      </td>

      <td>
        numeric
      </td>

      <td>
        The minimum detectable iROAS for Returning Customer Revenue: the smallest incremental return on ad spend this experiment's design can reliably detect (minimum required incremental returning customer revenue implied by the power analysis, divided by holdout spend), floored at `0.15`. Set at design time. A completed test whose true effect is below this threshold will typically report `no_effect`.
      </td>
    </tr>

    <tr>
      <td>
        Revenue MDE
      </td>

      <td>
        `revenue_mde`
      </td>

      <td>
        numeric
      </td>

      <td>
        The minimum detectable iROAS for Revenue: the smallest incremental return on ad spend this experiment's design can reliably detect (minimum required incremental revenue implied by the power analysis, divided by holdout spend), floored at `0.15`. Set at design time. A completed test whose true effect is below this threshold will typically report `no_effect`.
      </td>
    </tr>

    <tr>
      <td>
        Spend Reduction
      </td>

      <td>
        `spend_reduction`
      </td>

      <td>
        numeric
      </td>

      <td>
        The tested campaigns' spend in the holdout (treatment) geos over the test duration, which is the spend that is removed during the experiment, in the shop's currency. Computed as the holdout geos' daily spend times the test duration; NOT the campaigns' total spend.
      </td>
    </tr>

    <tr>
      <td>
        Spend Reduction %
      </td>

      <td>
        `spend_reduction_pct`
      </td>

      <td>
        numeric
      </td>

      <td>
        The holdout (treatment) geos' share of the tested campaigns' total spend, in percentage points `0`–`100` (e.g. `21.7` = 21.7%). Note this scale differs from the `result_*_lift_pct` columns, which are fractions.
      </td>
    </tr>
  </tbody>
</Table>

<HTMLBlock>{`
<style>
  .content-toc {
    flex: 0 0 220px !important;
}
  .rm-Guides .content-body {
    max-width: calc(100% - 140px) !important;
    width: 960px !important;
}
</style>
`}</HTMLBlock>