- Published
- Nov 3, 2026
- Reading time
- 11 min read
LevelPlay vs Meta Revenue: Reconcile the Report and Price Basis
Compare LevelPlay and Meta Audience Network revenue by report family, impression population and valuation. Keep estimates separate from payment evidence.

In this guide (9 sections)
When LevelPlay and Meta Audience Network show different revenue, first identify which LevelPlay surface produced the number. Real Time Pivot, Performance reports, a Reporting API response and an SDK revenue ledger are different evidence. Then compare impression populations and price bases separately. A mediation estimate and Meta's estimated revenue are not interchangeable with a payment statement.
A publisher question about Meta bidding through Unity LevelPlay asks whether a large revenue difference reflects historical CPM estimates or reporting delay, and which number represents earnings. The original also reports different impression totals. It was visibly marked three days old when reviewed on 3 November 2026. Replies suggest several causes, but the thread does not establish its exact LevelPlay report or a verified diagnosis.
The useful first answer is conditional: Unity explicitly documents a Meta bidding exception in Real Time Pivot's revenue calculation. Verify that report before applying the exception. Neither its existence nor the size of a discrepancy proves that it explains this publisher's result.
Give each number a report identity#
Save original exports and the configuration beside them. “LevelPlay says” is too broad for a reproducible investigation.
| Evidence | Preserve with the amount | Question it can answer |
|---|---|---|
| Real Time Pivot | Data cube, measures, dimensions, filters and extraction time | What did this real-time reporting view estimate? |
| Performance report | App, network, format, period and breakdowns | What does this selected performance population contain? |
| Reporting API response | Endpoint/version, parameters, pages and response time | What did this particular filtered query return? |
| AdInfo or impression-level ledger | Listener event, SDK version, source, currency and precision | What value accompanied this SDK observation? |
| Meta report | Exact metric, property/placement, interval, timezone and export time | What does Meta report for the selected scope? |
| Financial record | Covered period, statement or transaction status and currency | What financial amount and status are actually recorded? |
Do not replace an original export with a spreadsheet that contains only the two headline values. Keep transformations repeatable and label any unavailable field. An amount with no report name or date boundary remains an unresolved input, even if the number looks plausible.
For a callback that is being forwarded into Firebase or another analytics tool, use the AdMob/Firebase revenue collection investigation to examine ownership, units and repeated forwarding. Here, the task begins with the reporting and valuation contract, rather than assuming another event should be logged.
Identify which LevelPlay report you are comparing
Real Time Pivot, performance reports, API output and SDK records are different evidence. Discuss the report surface and price basis involved.
Discuss measurement scopeA short inquiry to scope relevant measurement options; financial adjustments are not promised.
Apply the Meta exception to the documented surface#
Unity's Real Time Pivot reference differentiates bidding-price revenue from revenue based on an instance rate or historical CPM. It explicitly includes Meta Audience Network bidding in the latter treatment.
That is a reporting exception, not a statement that Meta stopped bidding in auctions. It also does not establish how every LevelPlay report calculates every row. Record the selected data cube and retrieve the corresponding documentation before generalizing across surfaces.
The same page describes an average accuracy comparison for its prediction. An average documentation statement is not a contractual tolerance for your app, a settlement guarantee or a reason to dismiss a larger unexplained gap. It does not identify the cause of the original question's discrepancy.
Performance reports have their own configuration, including filters and exported breakdowns. Preserve those selections when moving from a pivot view to a performance export. A fresh export from another view is a second observation, not a correction that silently replaces the first.
If you cannot identify the original view, ask the report owner for its name, configuration and export. Do not change instance rates or live mediation settings merely to make two totals agree. A reporting investigation should first describe the existing setup.
Match the population before comparing revenue#
Create a private mapping between LevelPlay's app, network instance and ad unit and Meta's property and placement. Preserve identifiers privately; public documentation does not need customer, app or campaign names.
Confirm each pair actually covers the same inventory. Check platform, format, country, test status, app versions where available and any selected bidding filter. If a report includes another network or all apps, separate that scope before interpreting its price.
Meta's performance metric definitions distinguish an ad returned to a request from an ad appearing on screen. They also explain that other platforms can count differently. A filled ad that is cached and never displayed is not automatically a Meta impression.
Keep requests, fills, displays, recorded impressions and revenue observations as separate stages. Do not decide that LevelPlay counts a fill merely because Meta warns that some mediation systems do so. Inspect the documented event and actual implementation for the selected path.
Use the impression-reconciliation evidence workflow to name counting boundaries while retaining the app-specific metric definitions. If there is a count difference, localize it by compatible daily and placement breakdowns. An all-format total can conceal one problematic version or placement. Conversely, a very small slice may have limited or unavailable data. Retain missing rows and unsupported dimensions instead of silently converting them into zeros.
No aggregate count difference identifies invalid traffic by itself. Meta documents exclusions in its metric definitions, but that does not establish how many of this publisher's impressions were excluded or whether the difference has that cause. An evidence-based count bridge needs actual records for each allocated component.
Preserve API filters, dates and pagination#
The current LevelPlay Reporting API reference specifies UTC start and end dates. It also documents app, network, format, platform and bidding filters, breakdown selections and paginated results.
Retain the exact request parameters with a redacted export. Keep credentials and access tokens out of the packet. Check that every required page was retrieved, and preserve response errors rather than treating an incomplete response as the full report.
Use the source's supported network name and identifier mapping. A local label such as “Meta revenue” does not establish that the query selected the intended bidding rows. Similarly, an API default covering all sources can produce a valid response for the wrong population.
Do not assume a dashboard uses UTC because the API does. Record each report's actual clock and interval boundaries. Where reports use different timezones, obtain compatible daily or finer detail if available; changing the column heading cannot move events into a matching interval.
Record extraction times as well as event dates. To investigate revisions, save another observation of the same historical interval with the same filters. A later chart covering a different week cannot demonstrate that yesterday's revenue estimate matured or that a missing amount was restored.
Use SDK values with their event and precision#
Unity's AdInfo listener documentation describes revenue in USD and a precision field. RATE reflects a manual instance price, CPM reflects historical instance performance, and BID refers to an auction-supplied value under supported coverage. The page explicitly says exact bid values are not supported for Meta and Google bidders.
Keep the returned precision with the amount. A bidding instance label does not turn every supplied value into an exact clearing price. Do not multiply a Meta estimate by an invented revenue-share factor or relabel a CPM value as BID to reconcile it.
There is also an event boundary. The AdInfo reference says the object represents the last successfully loaded ad. Loading evidence alone is not proof that an impression occurred. By contrast, the documented impression-level revenue SDK flow supplies its callback on impression success. Establish which event your custom ledger consumes before calling its row count impressions.
This distinction can reveal a faulty comparison without proving a production defect. Request a bounded, authorized trace for the installed SDK and adapter versions. If development testing is required, use the documented network test setup and label test data; do not create paid clicks or synthetic revenue to improve agreement.
Work the count and price differences separately#
Hypothetical teaching example: two consistently scoped USD reports contain these values. They are invented examples, not the original publisher's results or an acceptable-discrepancy benchmark.
| Observation | Impressions | Reported revenue | Derived eCPM |
|---|---|---|---|
| Mediation reporting view | 12,000 | $24.00 | $2.00 |
| Network estimated-revenue view | 10,000 | $15.00 | $1.50 |
The count difference is 2,000 impressions: 16.67% of the mediation count or 20% of the network count. Name the denominator. The revenue difference is $9.00, or 37.5% of the mediation amount. Those percentages answer different questions.
A purely arithmetic bridge can separate two effects using the mediation eCPM as a reference:
- Removing 2,000 impressions at $2.00 per thousand changes $24.00 to $20.00: a $4.00 count component.
- Applying $1.50 per thousand to the remaining 10,000 impressions changes $20.00 to $15.00: a $5.00 price component.
- The components total $9.00, matching the headline difference.
This is an ordered decomposition of the two totals, not a causal allocation. It does not prove which impressions match, which price is correct, or that $4.00 was filtered for invalid traffic. A different reference order gives different components. Use matched evidence to establish causes; use arithmetic to show why a count-only explanation may be incomplete.
For aggregate eCPM across compatible rows, divide total revenue by total impressions and multiply by 1,000. Do not take an unweighted average of placement eCPMs or combine currencies before an explicit conversion.
Keep estimated earnings and payment evidence separate#
Meta's metric reference distinguishes estimated revenue from the later financial result and describes possible adjustments and timing differences. It names a Pacific reporting clock and a local-time payout basis. Preserve the displayed clock and the dates' applicable offset; do not turn the page's shorthand into a fixed UTC offset for every date or guess which local timezone an account uses.
The practical consequence is narrower than “trust this dashboard for cash.” A metric explicitly labeled estimated remains an estimate. For a payment question, obtain the relevant financial statement or transaction record, its covered period and actual status. Do not promise that a daily reporting total is the amount a bank will receive.
Likewise, do not borrow payment rules from AdMob, Facebook creator monetization, Stars or subscriptions. They are separate products. A large difference does not establish a refund, withheld payment, fraud finding or entitlement to compensation. Preserve any stated adjustment reason and keep unexplained amounts unresolved.
Finish with a reproducible comparison packet#
Before asking a reporting or implementation owner to investigate, check that the packet contains:
- The exact LevelPlay report family and its original export.
- Meta's named metric and matching property/placement scope.
- Date boundaries, timezones, currency, filters and extraction times.
- Complete API pages and any failed requests, where applicable.
- Separate impression-count and valuation comparisons.
- SDK event and precision evidence only where a custom ledger is involved.
- A dated repeat observation for any claimed revision.
- Financial records kept separate from operational estimates.
For campaigns already measured with AdProof, retain a supported event or aggregate export with matching dates, timezone and filters alongside these independently obtained reports. The anonymized FMCG in-app count-once case illustrates delivery-count reconciliation; it does not establish a LevelPlay integration or a Meta payout diagnosis. Discuss the available measurement evidence before assuming this revenue path is independently covered.
Frequently asked questions#
Does LevelPlay use historical CPM for Meta bidding revenue?
Unity explicitly documents that treatment for Meta Audience Network bidding in Real Time Pivot. Identify the actual report first. Do not assume the same calculation applies to every dashboard, API result or custom ledger.
Does a bidding label mean AdInfo revenue is an exact bid?
No. Preserve the returned precision and the SDK's supported source coverage. The cited AdInfo documentation excludes exact bid values for Meta and Google bidders. A network label is not a substitute for that precision contract.
Can different impression totals explain the entire revenue gap?
Only if compatible counting and valuation evidence supports that conclusion. Compare the count population and derived eCPM separately. An arithmetic bridge describes the totals; it does not identify filtered traffic or establish a causal allocation.
Which dashboard proves the amount I will be paid?
An estimated-revenue report does not prove a bank payment. Keep its period and scope alongside the relevant financial statement or transaction status. Preserve adjustments and unanswered questions without inventing a standard delay or a guaranteed settlement amount.
Identify which LevelPlay report you are comparing
Real Time Pivot, performance reports, API output and SDK records are different evidence. Discuss the report surface and price basis involved.
Discuss measurement scopeA short inquiry to scope relevant measurement options; financial adjustments are not promised.
