SEO case study: how to read one, and how to write one that earns trust

An SEO case study is a written account of what a site looked like before a piece of work, what was changed, and what happened after. A credible one contains six things: the starting position with dates, the specific changes made, the time period covered, the brand and non brand split, the business outcome rather than traffic alone, and a named client or at least a verifiable live URL. A case study missing those is a marketing asset rather than evidence. Whether you are reading one to choose an agency or writing one to win clients, those six items are the whole test.

This matters because SEO results are easy to present misleadingly without saying anything untrue. A percentage increase from a tiny base looks enormous. A chart with no axis labels can show anything. Traffic that rose because the client ran television advertising will sit happily in an agency's portfolio. None of that is lying, and all of it is useless for predicting what the agency can do for you.

How to read a case study you are shown

The claim	The question that tests it
"Traffic up 300 per cent"	From what number to what number, and over how long?
"First page for 200 keywords"	Which keywords, with what search volume, and how many bring enquiries?
"Organic revenue doubled"	Brand or non brand? What else changed in the business that period?
"We fixed their technical SEO"	Which specific problems, and what did the before and after crawl show?
"Results in 30 days"	Which metric moved in thirty days, and is it still there a year later?
"Built 500 backlinks"	From which sites, and which of them send real referral traffic?
"Beat a national competitor"	On which queries, in which location, measured how?

Ask for the live URL of the site. Then check it yourself: look at the pages the case study credits, see whether they are still live, and look at the site's own content to judge whether the work described is the work you can see. Ten minutes of that reveals more than any slide.

Six questions to ask the agency directly

1. "What was happening to this site before you arrived?" A site recovering from a technical disaster produces spectacular charts that say little about steady growth work.
2. "What else changed in that period?" A redesign, a new product, paid media, press coverage, a competitor closing, or a seasonal peak. Honest agencies volunteer these.
3. "What was the brand versus non brand split?" The non brand numbers are the ones the SEO work earned.
4. "Can I speak to that client?" One reference call outweighs a portfolio.
5. "What did you try that did not work?" Any real engagement has an answer, and the answer tells you how they handle it.
6. "Which of these results would you expect for a site like mine?" Listen for the differences they identify between that client and you. A firm that says you will get the same outcome is not thinking about your situation.

Warnings worth taking seriously

Be careful with a case study that guarantees you the same outcome. Google's own guidance has been consistent for years that nobody can guarantee a number one ranking, and a past result for one site in one market is not a commitment for yours. Be careful with charts that have no dates or no axis values, with keyword counts presented as the headline result, and with before and after screenshots from different tools or different locations, since results differ by device and place. And be careful with anonymised case studies where the industry and the country are both withheld, because nothing in them can be checked.

None of this means case studies are worthless. A specific, dated, checkable case study from an agency in your sector is one of the most useful things you can read before signing a contract. The point is to ask for that version.

How to write an SEO case study that earns trust

If you are the agency or the in house team, this structure works and takes an hour once you have the data:

• Client and context, in two sentences. Industry, market, size, and what they sell. Get written permission to name them, and if you cannot, name the sector and link the live site anyway.
• The starting point, with dates and numbers. Non brand clicks, conversions, indexed pages, and the specific problems found, with the date of the audit.
• The diagnosis. Why the site was underperforming, in plain language. This is the part readers judge your competence on.
• What you did, in order. Specific and checkable: the pages rewritten with their URLs, the redirects fixed, the templates changed, the links earned with the live links shown.
• What else was happening. Other changes in the same period, stated plainly. This is counterintuitive and it is the single strongest trust signal in the document, because it shows you measure honestly.
• Results, with the period and the comparison. Year on year where seasonality matters, with non brand split out, and the business metric the client cares about: enquiries, bookings, orders, revenue.
• What you learned and what came next. Including anything you tried that did not work and how you reallocated the effort.
• A client quote that mentions a specific thing, not a compliment.

Use screenshots from Search Console or analytics with the dates visible and the axis labelled. Blur revenue figures if the client prefers, and keep the percentages and the shape of the curve.

Measurement choices that make a case study defensible

• Compare year on year, not month on month, in any seasonal category.
• Use a control group where you can: page groups you did not touch, over the same period. This is the strongest evidence available in SEO and almost nobody shows it.
• Report at query cluster level, so a reader can see which intents grew.
• Annotate the chart with launches, migrations and search updates.
• Keep the raw exports. If someone asks for the underlying data, being able to send it ends the conversation in your favour.

This page covers reading and writing SEO case studies as evidence. For choosing an agency overall, including contract terms, account ownership and reporting cadence, those decisions deserve their own guide.

What gets asked about SEO case study

What makes an SEO case study credible?

Dates, a starting baseline, the specific changes made, the brand and non brand split, a business outcome, and a site or client a reader can verify. Those six items are what separate evidence from promotion.

Why do agencies anonymise case studies?

Often for a genuine confidentiality agreement, especially in finance, health and competitive retail. Ask for a reference call instead, which gives you the verification the written document cannot.

Is a traffic increase a good result?

Only if it brought the right people. Traffic that does not convert can come from informational pages far from the buying decision. Ask what happened to enquiries, orders or revenue over the same period.

How long should the results period be?

Long enough to cover a full seasonal cycle for that business, and ideally long enough to show the result held. A six month window in a seasonal category tells you less than a twelve month comparison.

Can I write a case study with no client permission?

Write it without naming the client: give the sector, market and size, describe the work precisely, and show the metrics as percentages and shapes. Never publish a client name, a logo or a quote without written agreement.

What single number should the headline use?

The business outcome the client was buying: enquiries, qualified leads, bookings or revenue from non brand organic search. Traffic and rankings belong inside the document as the explanation.

One more result worth documenting now

A good case study states the starting point, the change and the outcome, with the measurement method beside each. Add a line the old template never had: what an assistant says about the client's category before and after the work. Ask the same buyer question in ChatGPT, Gemini and Perplexity, and save the answers with the date.

Those saved answers are evidence a prospect can check for themselves in a minute. AI Knows Us keeps that kind of record, listing the companies each assistant named and the pages it cited, which makes the before and after comparison something you can put in front of a sceptical reader.