View-Through Conversions on Meta: How to Value Impressions

A post-impression conversion is advertising credit under a reporting rule. It is not automatically an incremental sale. Compare click-only and click-plus-view reporting with identical scope, then use a valid control experiment when the budget decision requires causal evidence.

Check what the account can report

Consult Meta’s attribution-setting help from the actual account and record the selected event, click window, eligible view/engagement windows and report date. Use available reporting-comparison controls rather than changing delivery settings merely to inspect another view. If a comparison is unavailable, document the exported report definitions. Do not assume a 28-day view option or a universal default.

Aggregate video-view metrics do not establish a customer-level ledger joining percentage watched, impression time and eventual order. Verify what the permitted export actually contains before building weights. Avoid trying to recover private viewer identity from unrelated datasets. No arbitrary engagement discount is used here.

Two comparable reports

Illustrative calculation, not observed results: credited ROAS is 4.0 versus 5.0. The difference is 20 outcomes and USD 1,000 of credit under these stated report definitions. Confirm deduplication, attribution priority and whether engagement-through credit is included before naming that difference “view-only.” Never add comparison-window columns unless the source explicitly establishes they are disjoint.

Neither number is a lower or upper bound on the true incremental effect. Click credit can include people who would have bought without an ad; missing exposures can hide influence. Keep platform credit, reconciled total net sales and causal lift as separate rows.

Why a lift percentage cannot value a view-through share

Suppose, hypothetically, a valid randomized study estimates 15% relative conversion lift and an attribution report assigns 20% of credited purchases to views. Dividing 15 by 20 gives 0.75 arithmetically but says nothing about the share of view-through purchases that is incremental. The percentages use different denominators and the study tests the campaign, including clicks and views, not only the view-credit subset.

A synthetic comparable study has 1,000 assigned users per group: 115 convert under the advertising strategy and 100 in control. The rate difference is 1.5 percentage points; relative lift is 15%. A simple large-sample 95% interval for the rate difference is about −1.21 to 4.21 percentage points, so these inputs do not establish positive lift. The NIST two-proportion reference explains the statistical comparison; validate independence and the actual study method.

Limit the business decision

1. Record the exact experiment scope and prevent control exposure through the campaigns being tested. Assess spillover and permitted event capture.
2. Include refunds, variable costs, response lag and uncertainty before comparing incremental contribution with spend.
3. If the interval includes zero or commercial loss, report an inconclusive result and choose a smaller affordable follow-up rather than inventing a view-credit multiplier.
4. Prepare the message and a chosen spend ceiling as a NovaStorm campaign draft, then review the evidence before publication. Reporting comparisons and causal experiments remain in the analytics process.