Fixed on this site
My own case study numbers did not add up
Two of eight showed a total that the breakdown under it could not reach. On a site that sells checking, that was the first thing to fix.
On this page
On 20 September I audited my own site the way I audit anyone else's: from the outside, with no special access, checking every page against itself.
The worst finding was not technical. Two of my eight case studies printed a number of findings, and directly under it a breakdown by severity that did not add up to that number. The other six matched exactly, which is what made the two stand out.
Before and after
This site, audited 20 September 2026
- Before
- 2 of 8 case studies did not reconcile. One said 14 findings above a breakdown that counted 2. The other showed a dash above a breakdown of 17.
- After, checked 21 September
- 8 of 8 reconcile. The first reads 14, broken down as 2, 8 and 4. The second reads 17, broken down as 3, 5, 6 and 3.
- Stops it coming back
- The site now refuses to publish if any case study's breakdown does not add up to its total.
What went wrong: a template that knew four words
Each case study shows its findings split by severity. The template that turns my case study data into pages knew four labels: Critical, High, Medium and Low.
One engagement had been rated on a different scale. Its own report used Critical, Warning and Info: 2, 8 and 4, which makes 14. The template recognised Critical, found nothing it knew in the other two, and dropped them without a word. So the page said 14 findings above a bar that accounted for 2.
The second case was simpler. The total had been left empty in the data, so the page printed a dash above a breakdown that added up to 17.
Neither is the kind of mistake you spot by reading the page. You only see it when you add the numbers up, which is exactly what a careful reader does before trusting anything else on it.
The fix, and why it is not a relabel
The quick fix would have been to rename Warning to High and Info to Low, so the template understood them. I did not do that. Renaming a severity changes what that audit found, and the report the client received says Warning, not High.
So the page now shows the report's own labels: Critical 2, Warning 8, Info 4. The empty total now reads 17, the sum of its own breakdown.
Then the part that matters more than either fix. Before anything goes live, the site now checks that every case study's severity counts add up to its stated total. If one does not, nothing is published. The first time that check ran, it caught this exact problem.
Why this mattered more than any technical finding
This is my judgement, not a Google rule. Google does not add up the numbers on your case studies. People do.
Someone deciding whether to hire an SEO consultant is doing, on a small scale, what I do in an audit: checking whether the claims hold up. A total that its own breakdown cannot reach tells that reader nobody checked. On a site whose whole pitch is checking, that costs more than any missing tag.
Check my fix yourself
- Open the B2B telecom case study and find Findings near the top. It says 14. B2B telecom provider
- Just below it, Findings by severity reads C 2, W 8, I 4. That adds up to 14.
- Do the same on the design-led e-commerce case study: a total of 17, and 3, 5, 6 and 3 under it. Design-led e-commerce brand
Check your own site, no code needed
- Write down every number your site states about itself: clients, projects, years in business, results, team size.
- For each one, open every page where it could appear and press Ctrl+F (Cmd+F on a Mac) to find it. The homepage, About page, case studies and PDFs often each have their own copy.
- Wherever a total has a breakdown under it, add the breakdown up.
- Check your downloads too. A PDF or slide deck made months ago still says whatever it said then.
What to ask your developer
- Is each of these numbers stored once and shown everywhere, or typed separately on each page?
- Can the site refuse to publish a page where a breakdown does not add up to its total?
What this does not show
This check proves the numbers on a page agree with each other. It cannot prove the audits behind them were right. Each case study links its own report for that.
A number that is wrong the same way everywhere passes it.
Want your site checked the same way?
Send a URL. Three to five verified findings come back within two working days, free, each with a way to check it yourself.