usdigitalmarketingindex.com SEO
Which verifiable evidence should an SEO case study include to support its results claims?
Updated November 15, 2026
A case study that others can check includes six things: the scope of what was measured, a baseline from before the work, dated observations from a named data source, a record of what was changed and when, an honest account of what else could explain the result, and confirmation that the client permitted publication. A claim such as traffic grew 200 percent is not evidence. It becomes evidence when a reader can see what was counted, over which dates, compared with what, in which tool, and what else was happening at the time.
Defining scope and baseline
State exactly what the numbers describe.
- The site or section: the whole domain, the blog, a set of product pages.
- The metric and its source: clicks from Search Console, sessions from the Organic Search channel in analytics, leads from a CRM. These are different counts and should not be mixed in one claim.
- The filters: country, device, branded queries included or excluded.
- The baseline period, and why it was chosen. For a seasonal business the fair comparison is the same months of the previous year.
A percentage needs its starting figure. Growth from 50 visits to 150 and growth from 50,000 to 150,000 are both 200 percent.
Showing dated observations
Show the data, not a description of it.
- Exports or screenshots from the tool, with the date range and filters visible.
- The absolute numbers behind every percentage.
- A chart that covers the period before the work as well as after, so the reader can see what the line was already doing.
- The date each export was taken.
Pick the comparison periods before looking for the best result. A case study that reports the single best week against the single worst is selecting its evidence.
Attribution and confounders
The hardest part of a case study is the word because. A rise that follows a change was not necessarily caused by it. Present the change log and the other candidates together.
| What the case study should state | Why |
|---|---|
| Each change made, with its date | Lets the reader line changes up against the chart |
| Other marketing in the period: ads, PR, email, offline | Any of these can raise branded search and direct visits |
| Seasonality of the business | A summer peak is not a result |
| Announced Google updates in the period | A sitewide shift may have an outside cause |
| Site events: migration, redesign, tracking changes | Measurement changes can create apparent growth |
| Pages or sections left unchanged | Their trend is the closest thing to a comparison |
Then word the conclusion to match the evidence. Organic clicks to the rewritten pages rose while clicks to the unchanged pages stayed level is a supported statement. The rewrite increased traffic by 200 percent claims a cause that the evidence may not isolate.
If a single change is credited, say why the alternatives were ruled out. If they cannot be ruled out, say so.
Permissions and redactions
Client data belongs to the client.
- Obtain written permission that names what will be published: the client’s name, the figures, the screenshots.
- Where the client must stay anonymous, describe the business type and size, keep the figures, and state that identifying details were removed. Do not invent a name.
- Redact account identifiers, email addresses, and anything else in a screenshot that is not needed.
- Do not alter numbers to protect confidentiality. Index them to a base of 100 and say that you did.
Use this checklist before publishing. Each row gives the input to verify, the action, and the pass or fail condition.
| Input to verify | Action | Pass or fail |
|---|---|---|
| Scope statement | Read the claim and identify site, section, metric, source, filters | Fail if any of the five is missing |
| Baseline | Find the comparison period and its rationale | Fail if absent or chosen after the fact |
| Raw data | Locate the export or screenshot behind each figure | Fail if a figure has no source shown |
| Absolute numbers | Check every percentage has its start and end values | Fail if any percentage stands alone |
| Date range | Confirm dates are visible on each chart | Fail if not |
| Change log | List changes with dates | Fail if the work is described without dates |
| Confounders | Check each row of the table above is addressed | Fail if the period's other events are not mentioned |
| Causal wording | Compare the claim with what the data isolates | Fail if the wording asserts more than the evidence |
| Permission | Find written consent covering what is shown | Fail if absent |
| Redaction | Inspect screenshots for identifiers | Fail if any remain |
A case study that passes is also more persuasive. A chart that goes up is easy to produce. One that shows its baseline and admits what it cannot prove gives a careful reader a reason to believe the rest.
Google’s guidance on choosing SEO help gives buyers the same instinct, warning that “No one can guarantee a #1 ranking on Google” (Google’s guidance on hiring an SEO). A case study should not imply one either. Report what happened for this client, and leave out any suggestion that it will happen for the next.
Deciding in advance what would count as success or failure for a planned test is a related discipline with its own guide: the criteria that determine whether a completed SEO experiment has failed.
The standard to hold: a skeptical reader with access to the same account should be able to reproduce every number in the study.
Add your comment to this answer