1. Define the question before opening the CSV
“Revenue fell” is an observation, not a diagnosis. Before looking at the numbers, state the comparison you want to make. For example: “Why did estimated earnings change between two complete seven-day periods?” Avoid comparing a complete week with a partial week, or a month-to-date period with a full prior month.
Write down the reporting time zone, currency, date range, and dimensions included in each export. A change in the report settings can look like a performance change even when user behavior stayed the same.
2. Verify that the columns mean what you think they mean
At minimum, a simple page-RPM calculation needs page views and earnings from the same reporting scope. To calculate click-through rate (CTR), you also need clicks and impressions. If you want to compare pages or locations, the file must contain a page/path or country/region dimension. Do not fill missing columns with guesses.
Keep the original export untouched. Work on a copy, preserve the headers, and check whether the file uses commas, semicolons, quoted text, a decimal comma, or a currency symbol. Review a few records manually after importing a file into a tool.
3. Calculate ratios from totals, not by averaging row ratios
For an aggregate report, calculate page RPM from the total earnings divided by total page views, multiplied by 1,000. Do not take a plain average of per-page RPM values if pages have very different amounts of traffic; the result can misrepresent the overall rate.
CTR = (total clicks ÷ total ad impressions) × 100
eCPM = (total earnings ÷ total ad impressions) × 1,000
These ratios are useful descriptions of the selected report. They are not proof of why a metric changed, and they are not a promise of future earnings.
4. Work through a transparent example
The following figures are illustrative, not a real account or a market benchmark.
| Period | Page views | Estimated earnings | Page RPM |
|---|---|---|---|
| Period A | 100,000 | $420.00 | $4.20 |
| Period B | 120,000 | $456.00 | $3.80 |
Earnings rose by 8.57%, while page views rose by 20%. Page RPM fell by about 9.52%. So “revenue increased” is true, but it is not the whole story. The site attracted more views while the observed earnings per thousand views were lower. The numbers alone do not identify the cause.
A disciplined next step is to split the same comparison by page, country, device or another available dimension—one at a time—and check whether the underlying mix changed. Do not assume that any single dimension caused the result until the data supports that conclusion.
5. Compare like with like
- Use identical date windows, reporting time zone and currency wherever possible.
- Check if one export is filtered to particular pages, countries, ad units or devices.
- Identify missing days, delayed reporting, unusually small sample sizes and rows with empty earnings.
- Keep estimates separate from finalized financial records; a reporting dashboard may update after the first export.
- Do not interpret CTR movement as a reason to ask people to click ads or otherwise manipulate ad activity. Follow the ad network’s invalid-traffic rules.
6. Use a scenario model only after the audit
A forecast should start with observed data and explicit assumptions. If a period contains 14 days, monthly baseline views can be approximated as total views divided by 14 and multiplied by 30. Multiply that result by observed page RPM and divide by 1,000 to estimate a baseline revenue scenario. A low/high range can change traffic and RPM assumptions, but every such adjustment should be visible.
The baseline assumes the selected period is representative. Seasonality, geography, consent rates, viewability, ad formats, ad demand, changes to site layout and policy status can alter the result. A scenario is a planning tool, not a prediction.
7. Keep an audit log
For each comparison, record the question, data source, date coverage, currency, filters, formulas, missing columns, notable differences and follow-up checks. When you revise a conclusion, note what evidence changed it. This reduces the risk of attributing a movement to a layout change or traffic source merely because it happened around the same time.
Sources and related tools
- Google AdSense Help: Page RPM — definition and formula for page RPM.
- Google AdSense Help: Impression RPM — how impression-based RPM is defined.
- Google AdSense Help: Click-through rate (CTR) — meaning of CTR and interpretation context.
- Google AdSense Help: Make sure your site's pages are ready — original content and user-experience guidance for site owners.
These sources define reporting metrics and site-readiness principles; they do not validate this tool’s input file or guarantee approval or revenue.