Marketing agency reports: the numbers to check against your own accounts

You can trust the numbers in a marketing agency's report only as far as they match the values you pull again from the source accounts with the same property, date range, filters, and metric definitions. Without view access to the source accounts you can't do this check, so agree on permission levels with the agency at the contract stage.

If a bar chart showing growth in search traffic has no footnote saying which account and date range the values came from, you can't decide which value to follow when it doesn't match the number you get by pulling the same month again in Search Console.

1. How marketing agency reports relate to the source accounts

You can pull the impressions, clicks, sessions, conversions, and ad conversions in an agency report again from Search Console, GA4, and the ad managers, but there is no source account for visibility in AI answers.

The tracking metrics PION has published for its managed GEO service also divide into those with a source account and those without. Among them, AI tool referrer traffic in GA4, Google and Bing indexing status, and Core Web Vitals can be checked directly by the client if the client has view access to GA4, Search Console, and Bing Webmaster Tools.

2. Pulling impressions and clicks again in Search Console

The benchmark is the value you get by pulling one impressions-and-clicks item from the agency report again in the Performance report with the same property, date range, search type, and dimension tab.

1. Pick one item in the report that lists both impressions and clicks. If it's a site-wide value, compare it with the chart totals; if it's a per-query value, compare it with that query's row in the Queries tab.
2. Open the same property as the agency report (whether it's a Domain property or a URL-prefix property), and set the date range, search type, device and country filters, and dimension tab to match the report.
3. If the two values don't match, ask the agency which condition differs: the property, the date range boundaries, the filters, or the device breakdown.

Performance report overview and default settings states that "data in the chart is aggregated by property" and that "chart totals may differ from table totals." Why the Queries tab and Pages tab totals don't match is explained in How to use Google Search Console: What the metrics mean.

Search Console data is "generally published within 2-3 days" (About Search Console data), so pull last month's values after 3 days have passed since the end of the month. According to the same document, "each day's data is labeled according to local time in California." GA4 dates follow the property's time zone, so check the report's clicks against the Search Console values and its sessions against the GA4 values separately.

If the agency lists top-ranking queries as a result but Search Console shows 0 clicks for those queries, ask it to include click counts from the next report on. The causes of ranking high with no clicks are covered in Why you rank high on Google but get no clicks.

3. Pulling sessions and conversions again in GA4

In GA4, query with the same property, date range, and key event name as the agency report, and for conversions by channel, match whether they were pulled with an event-scoped or a session-scoped dimension before comparing.

The Select attribution settings help states: "Changing the reporting attribution model applies to both historical and future data." This change is reflected in conversions for event-scoped dimensions such as default channel group, source, and medium, but not in session-scoped dimensions such as session source. If the report doesn't name the model, don't change the property settings; ask the agency which model it used.

For conversions in event-scoped dimensions, attribution credit "can change for up to 12 days after the key event is recorded" (GA4 data freshness). Compare last month's conversions by channel only between values pulled after 12 days have passed since the end of the month.

According to About Analytics sessions, "by default, a session ends (times out) after 30 minutes of user inactivity." This timeout can be changed in the tag settings of the web data stream. In GA4 properties, "traffic from known bots and spiders is automatically excluded" (Known bot-traffic exclusion). If session counts differ within the same property, ask the agency whether it changed the session timeout and whether it applied any additional filters to the report.

4. Attribution windows and later changes in ad manager metrics

Ad conversion counts change with the window for counting conversions after a click and with later adjustments, so the agency report must state each platform's counting window and the date the data was pulled for the client to check it against the source. According to Google Ads' About conversion windows, "for Search and Display campaigns, if you don't customize the conversion window, a 30-day default window applies to conversions." If you change the window, it "applies to all conversions that occur afterward." So in a month when the window was changed midway, the conversion count mixes conversions counted under the window before the change and under the window after it.

In Google Ads, because of retroactive removal of invalid traffic, late-reported conversions, and month-end adjustments, "reporting metrics are sometimes updated more than a day after the event occurred" (About data freshness).

According to Kakao Moment's How can I check the conversion report? help, 1-day conversions are "conversion metrics collected during the 24 hours after an ad click," and 7-day conversions are "conversion metrics collected during the 7 days after an ad click (including conversion metrics collected during the 24 hours after the click)." On the source screen, select the same window as the report's conversion column.

Conversions in Naver Ads reports are provided "only for conversion actions that occur after an ad visit" (Naver Ads web conversion tracking guide, Version 20260609_01). The online store's order count also includes orders that did not come through ads, so check the report's Naver Ads purchase conversions against the ad report.

5. Why reports and source data differ, by type

Numbers in a report and in the source differ because of property and date range boundaries, filters and aggregation units, attribution windows and models, settings changes, and later adjustments.

6. Questions to ask the agency after checking

Once you've finished checking, ask the agency about its query conditions, its history of settings changes, and how it calculates the numbers that have no source account.

1. Ask when, in which property and on which screen, and with what date range, filters, and definitions each value was pulled; if there's no answer, ask the agency to note the query conditions next to the numbers starting next month.
2. If the key event names, attribution model, conversion window, or session timeout have changed since last month, get the date of the change and the previous value.
3. Ask what data the agency used to calculate the ranking scores, estimated traffic, and visibility in AI answers from its own tools.

In its GEO agency service overview, PION states that comparing the effects of managed GEO requires at least 3 months of data. To check your company's readiness for AI answers separately before receiving agency reports, answer the 8 questions in the AI answer visibility basics check. The result is a self-assessment score tallied from your own answers, so don't mix it with the numbers in agency reports.