SEO Case Studies: 7 Red Flags Behind Impressive Traffic Claims

Intro

0:00

/210.308866

1×

Broader Discovery

0:00

/194.10932

1×

Traffic Growth Tactic

0:00

/177.023946

1×

Unexplained Average

0:00

/160.776961

1×

An SEO case study can document genuine growth in the metric an agency chooses to display. That observation still leaves three distinct questions for a business owner evaluating a purchase: whether the reported change is accurately measured, whether the work caused it, and whether the result is relevant to the buyer’s own starting position, market, budget, and execution capacity. A real increase in sessions or clicks does not automatically establish the broader sales story that surrounds the chart. The seven checks that follow help separate what the case study actually records from the narrower or wider conclusion the surrounding narrative invites.

The percentage has no starting or ending count

A percentage without its baseline and endpoint is incomplete. Ask for the unit being counted, the starting count, the ending count, and the absolute change. A baseline is the starting measurement against which later change is compared. A move from 50 sessions to 150 sessions is a 200 percent increase and an absolute gain of 100 sessions. A move from 5,000 sessions to 6,000 sessions is a 20 percent increase and an absolute gain of 1,000 sessions. The smaller percentage describes the larger absolute change. Low baselines make large percentage gains easier to produce, but they do not make the growth illegitimate. They change what the number means for a business that already operates at higher volume.

Percentage growth from zero is undefined under the usual formula because division by zero has no finite result. When a site begins with no measurable traffic and later records sessions, the useful description is the absolute count and the period, not a percentage. Google’s general guidance on analyzing data notes that context and the shape of the underlying distribution matter when interpreting change; a single ratio can hide the scale of the counts themselves. Request the actual numbers so you can judge magnitude against your own situation. The next check examines whether the chosen dates conceal a wider pattern.

The dates hide the wider trend

Exact start and end dates for complete reporting periods are required before the comparison can be read. Ask what the same metric did in the weeks or months before the “before” period and after the “after” period. Seasonality, a weak recovery baseline after an earlier drop, a site migration, or a temporary outage can all affect the picture. Google’s guidance on investigating traffic changes lists seasonality, demand shifts, technical problems, migrations, and ranking changes as relevant context; longer series and comparable periods help interpret the movement. Equal duration of the two windows does not eliminate those confounders.

Prior-year comparisons are useful when the records exist, because they show whether the same seasonal pattern appeared before. No universal minimum length applies to every study; a short, carefully documented period can still be informative once the surrounding trend is visible. Older observations require retained historical data when current reporting history is insufficient. The following section asks whether the word “traffic” refers to the same measurement in both periods.

The report never defines what “traffic” measures

“Traffic” is not a single measurement. Three common figures that appear under a traffic headline record different events:

Three measurements that can appear under a traffic headline.

Search Console clicks are the clicks Google records from its search results. GA4 sessions are periods of interaction recorded after the analytics tag collects data on the site or app. Ahrefs Organic traffic is an estimate of monthly Google clicks derived from rankings, search volumes, and estimated click-through rates; daily points on its charts still describe that monthly estimate and should not be summed as recorded visits. Google’s comparison of Search Console and Analytics notes that the two systems use different metrics, so totals need not match even when the same property and dates are examined. The Google-only scope in GA4 uses Session source = google and Session medium = organic.

First-user acquisition reports and session acquisition reports use different scopes and should not be swapped. A first-user view attributes activity to the source of the user’s initial acquisition; a session view attributes each session to its own source. A repaired tracking tag, a change in consent settings, or a gap in collection can alter the reported numbers without any change in actual demand. Eligible GA4 properties can model behavior that was not observed because analytics consent was declined, so first-party reports are not an exact census immune to implementation or consent issues. Estimates remain useful when they are labeled as estimates. Disagreement between a third-party model and first-party records does not, by itself, establish that either side is fabricated. Check the report type, channel filter, property or page scope, geography, and whether collection continued without interruption. The next section looks at whether the queries behind the clicks were branded.

Branded demand is presented as broader discovery

A branded query is a search that includes the brand name, a close variation or misspelling, or a brand-related product or service name. Search Console’s native brand filter is available for eligible domain properties and uses an AI-assisted classification that can make mistakes. Manual query filters omit the anonymized queries that Search Console does not display individually. Do not subtract the visible branded rows from the overall total and then label the remainder an exact non-brand figure; the missing rows prevent that arithmetic from being precise.

Ask which queries changed and how they were classified. Brand labels describe the queries themselves, not whether the people searching are new or returning. Gains in branded demand can be valuable for a business without proving broader discovery of the site among people who had not already heard of it, and without identifying what caused the demand. Attribution here means the rules used to assign an outcome to a particular source or channel. The following check asks whether any single tactic can be credited for the entire change.

One tactic receives credit for everything that changed

A before-and-after chart records that something changed. It does not allocate the increase among the activities that occurred in the same window. Request the dates of implemented content, indexing fixes, redesigns, internal linking changes, outreach, publicity, and any relevant external events. Consider this hypothetical campaign: a site expands its content, repairs an indexing problem, and conducts journalist outreach during the same period. Any of those activities, or some combination, or an external change in demand, could contribute to the observed sessions. Paid advertising clicks are not organic traffic, though advertising can stimulate later searches that appear as organic. When the work includes earned media, managed AI-assisted reporter outreach with tailored pitches—such as the service offered by Backlinker AI—still needs the same date and change log so the contribution of placements can be separated from other concurrent work.

Comparisons of similar periods or similar pages can strengthen the reasoning without proving causality. Google’s general guidance on data analysis distinguishes correlation from causation; a coincident update is not, by itself, a proven cause. Not every useful case study needs to be a randomized experiment. The record can still support a narrower conclusion once the concurrent work is listed. The next section examines whether the traffic figure is being asked to stand in for a business outcome.

Traffic growth stands in for qualified leads or revenue

Traffic growth and qualified lead growth are different observations. A qualified lead is a recorded action that meets a defined business standard of relevance and intent; the definition must be stated. A GA4 key event is any collected event that has been marked as important; the label alone does not establish that the event is a qualified lead or a sale. Counting methods can be set to once per event or once per session, and a change applies only to future data, so continuity of the definition must be checked. Attribution rules and the timing of when an action is recorded also affect the totals.

Consider this hypothetical example with two complete 28-day periods, an unchanged property, GA4 Traffic acquisition scope, Session source = google, and Session medium = organic. The CRM uses the same attribution and qualification rules in both periods. A qualified inquiry here means a genuine request for a relevant service within the served region, deduplicated and qualified within the reporting window. This is an illustrative business definition, not a Google standard.

Hypothetical example: two complete 28-day reporting periods with unchanged definitions.

Equal duration does not remove seasonal effects. Later qualification or sales may change the picture. These small counts describe the illustration only. Session growth and inquiry growth are different observations; do not calculate a conversion rate across differently scoped records or infer statistical significance, failure, profit, ROI, or causality from them. For claims about business results, request the appropriate CRM or transaction records. Missing downstream records mean the business claim remains unsupported, not that the traffic was worthless. Attributed revenue is not necessarily incremental revenue. The final check asks whether a highlighted success or an average represents a typical outcome.

A selected success or unexplained average is presented as typical

A selected success can be real without predicting what a typical client experiences. Ask for the cohort that was included, the sample size, the selection rules and any exclusions, the averaging method, the variation across the group, and how similar the included clients were to the buyer’s starting position, market, budget, and execution resources. A cohort is the defined group of clients or periods over which the average was calculated. No minimum sample size is required for a case study to be informative, and confidential clients need not be disclosed publicly.

Google’s guidance on hiring an SEO recommends asking about prior work, measurement methods, relevant experience, and references. That advice supports examining the evidence, but it does not prescribe a particular framework. An average presented without its method and distribution can describe a selected subset rather than the typical result for a client who begins from a different baseline.

Closing evidence request

Request a permissioned or redacted source report that includes the exact dates, the definitions of every metric, and the filters that were applied. Ask for the timeline of implemented changes, the business records relevant to any outcome that is claimed, and the cohort method if typicality is asserted. Available evidence can support the narrower observation that the reported numbers moved under the stated conditions. It leaves open whether the work caused the movement, whether the result is typical, and whether the same movement would appear for a different starting baseline. Those remaining questions are the ones that determine whether the case study supplies relevant evidence for the purchase.

Sources

1. Backlinker AI, “Homepage,” Backlinker AI. Backlinker AI
2. Ahrefs, “What is Organic Traffic in Ahrefs and how do we calculate it?,” Ahrefs Help Center, September 3, 2026. Ahrefs Help Center
3. Google, “Using Search Console and Google Analytics Data for SEO,” Google Search Central, January 7, 2026. Google Search Central
4. Google, “Introducing the branded queries filter in Search Console,” Google Search Central Blog, November 20, 2025. Google Search Central Blog
5. Google, “Debug Google Search Traffic Drops,” Google Search Central, December 10, 2025. Google Search Central
6. Google, “[GA4] About the Acquisition reports,” Analytics Help. Analytics Help
7. Google, “[GA4] Key event,” Analytics Help. Analytics Help
8. Google, “[GA4] Change the counting method of key events,” Analytics Help. Analytics Help
9. Google, “[GA4] Behavioral modeling for consent mode,” Tag Manager Help. Tag Manager Help
10. Google, general guidance on data analysis, correlation, and causation, August 25, 2025. Google Search Central
11. Ahrefs, “Do search engines use domain authority/domain rating as a ranking factor?,” Ahrefs Help Center, September 3, 2026. Ahrefs Help Center
12. Google, “Do You Need an SEO? Tips for Hiring an SEO,” Google Search Central, June 5, 2026. Google Search Central