Publisher ad forecasts: connect requests to payable impressions and net revenue

An ad request is a potential delivery event, not automatically a payable impression. A publisher's forecast needs a consistent path from requests through recorded delivery to the population accepted for settlement. Apply the commercial rate to that final population, then account for the agreed revenue share and other costs.

Google Ad Manager's metric documentation distinguishes requests, code served, impressions and viewability measures, and contains different counting notes for particular report types and video pods. That is a reason to read the definitions for the actual inventory and report. A convenient relationship from a simple display example should not be imposed on every ad format.

Choose one reporting period, currency, inventory scope and time-zone basis for the comparison. Confirm whether a displayed revenue amount is estimated, finalized, gross or already net of a partner share. Combining incompatible populations can create an apparent monetization change that is actually a reporting mismatch.

Identify internal promotions and any contractual makegood delivery separately where they appear in the measured population. A recorded impression can satisfy a delivery purpose without creating an additional charge in the current period. Apply the relevant commercial treatment to each population instead of assigning the paid campaign's CPM to every counted ad.

Assume a simplified single-ad display population contains ten million requests during the month. Its compatible reporting shows eight million responses with ad code and two million unfilled requests. Suppose 7.2 million impressions are subsequently recorded, and the agreed finalized payable population is 6.9 million after documented settlement adjustments.

These numbers illustrate separate stages; they do not assert a universal conversion rate or explain the cause of every difference. The publisher must investigate its own delivery and settlement evidence. In particular, the 800,000 difference between code responses and recorded impressions should not automatically be classified as the same issue as the 300,000 settlement adjustment.

If the hypothetical agreement pays a gross $4 per thousand payable impressions, the result is 6.9 million divided by 1,000 times $4, or $27,600. Multiplying all ten million requests by the same rate would predict $40,000 and overstate this example's gross revenue by $12,400.

Suppose the arrangement then deducts an illustrative fifteen percent partner share from the $27,600 gross amount. The share is $4,140 and the publisher retains $23,460 before other expenses. These are invented commercial terms, not a statement about any platform's actual revenue-sharing rate.

Using all 7.2 million recorded impressions as the denominator, retained revenue corresponds to approximately $3.26 per thousand impressions. Using ten million requests, it corresponds to approximately $2.35 per thousand requests. Both calculations can be useful, but neither should be labeled simply the CPM without identifying its revenue scope and denominator.

If a report already supplies net revenue, do not deduct the partner share again. Similarly, a settlement adjustment already reflected in payable impressions should not be applied again as a separate percentage haircut unless another distinct contract provision requires it. Save the bridge so the forecast and the final settlement can be reconciled line by line.

Google's documented Active View percentage for viewable impressions uses measurable impressions as its denominator. That differs from assuming all recorded impressions were measurable. Read the exact metric selected, since a distribution over another eligibility population can answer a different question even when the report uses similar words.

For example, suppose six million of the hypothetical recorded impressions are measurable and 4.2 million of those are viewable. The viewability rate is seventy percent. Multiplying all 7.2 million recorded impressions by seventy percent would imply 5.04 million viewable impressions, which does not match the stated measurement evidence.

Do not automatically multiply the already finalized payable population by viewability. Whether viewability changes payment depends on the agreement and the settlement basis. Keep it as an operating quality measure where appropriate, and apply a viewable-impression price only to the population actually governed by that commercial rule.

A projected traffic increase changes the starting opportunity population, while a delivery improvement may change the relationship between responses and recorded impressions. A different sales mix can change the price earned per payable impression. Model those changes separately before multiplying several optimistic percentages together.

Use comparable inventory segments where their behavior differs materially. A blended historical rate can move because the proportion of formats, locations or demand sources changed, even if each segment's rate remained stable. Explain the expected mix and use supported assumptions for the segments that drive most of the revenue.

Include the actual operating costs and commitments affected by the plan, such as contracted serving, content or sales expenses. A higher gross yield does not necessarily increase contribution if obtaining it adds more cost. Evaluate proposed inventory changes against the product's user experience and approved policies as well as the financial forecast.

Retain the original reporting extract and forecast assumptions, then replace provisional amounts with finalized evidence without erasing the comparison. Explain differences in request volume, delivery, accepted payable population, realized price and partner deductions. An unexplained plug to net revenue prevents the next forecast from improving.

Check collection timing separately from the month in which delivery occurred. The expected settlement date, payment terms and any outstanding claim affect the cash forecast even when the delivered inventory is complete. A dashboard's estimated monthly earnings do not establish that the corresponding cash is available now.

The finished revenue model should show which population generated each amount and where provisional judgments remain. That gives the publisher a usable basis for planning and diagnosing changes, without turning requests, viewability, recorded impressions and settled revenue into interchangeable versions of one metric.