Case studies

SEO and AI search case studies

4
records of real work
2
clients named, with their permission
1
tool we built and costed in public
Talk about your project ↗

Four records of real work: a named branding agency's website rebuild, an audit of a flashcard app whose founder fixed every priority within two weeks, an anonymous audit of a services business, and the rank tracker we built for ourselves. Each one separates what we recommended, what was built and what has actually been measured.

GetBorvis / Organic visibility
Roadmap progress Sample view
04stages
MeasurePass
DiagnoseReview
PrioritizeGap
TechnicalContentAuthorityAI
How these are written

Most agency case studies show a number and imply they caused it. These separate four things that usually get collapsed into one, because only the last of them is evidence.

01

Baseline

What was true before we touched anything, recorded with a date. Without this there is nothing to compare against and any later number is a story.

02

Recommendation

What we advised and the mechanism we expected it to work through. Written down in advance so it can be wrong in public.

03

Implementation

What was actually built, by whom, and what was not. Often the client's team ships it, and saying so matters when the result is good.

04

Measured result

The same measurement re-run after a real interval. Where it has not been re-run, the record says so instead of borrowing a number from somewhere else.

When clients are named

Anonymous by default. Named only with permission.

A named logo is worth more to us than an anonymous one, and we still do not publish a client name, fee, or Search Console export without written permission. Most of the time we do not ask, because the work was commissioned privately and the client did not sign up to be our marketing. When we do ask, the client reads the full draft first. We Digital Hub did, and approved theirs on 2 October 2026. Notum Learn's founder agreed to be named and quoted on 14 October 2026.

What this means for you as a reader is that you are getting the method and the numbers we are allowed to show, and a name only where the client has said yes. What it means if you hire us is that your data gets the same treatment.

One consequence you can check: we hold work back. There is a piece of research about the same client that is finished and unpublished, because their approval covered the case study, not everything we measured. It stays unpublished until we either ask them about it or re-run it on an unrelated site.

SEO + AI search foundation6 min read

We Digital Hub: preparing a branding agency’s new website for search.

A branding and design agency in Howrah was rebuilding its website. Twelve months of its own Search Console data shaped every page title, a migration plan protected 185 project pages, and its team shipped the changes within days.

Live, credited by the clientRead the record ↗
SEO audit5 min read

Notum Learn: an SEO audit of a flashcard app, and what its founder fixed.

A broken upgrade link, subject pages showing the wrong decks, flashcard markup one property short and placeholder text on a public page. We sent four fixes on 30 September. Two weeks later, all four were live.

All four fixes live, founder quotedRead the record ↗
SEO + AI search audit8 min read

Seen, not chosen.

A services business was showing up in more searches every month in the market it wanted, and getting no more visits. The pages that should sell ranked too low, the offer was hard to read, and a few defects got in the way. Client not named.

Audit completeRead the record ↗
Tooling8 min read

We built our own geo-grid rank tracker. The searches in a 49-point scan cost nothing.

The commercial geo-grid tools are priced for agencies. This is the test that decided whether the idea worked at all, the field mask decision that is the whole economics of it, and the undocumented daily limit we found by hitting it.

First-party buildRead the record ↗
Why so cautious

A recommendation is not a result.

Search work takes months to surface, and a website launch changes many things at once. Attributing a traffic movement to one intervention, in a period when content, design and campaigns all changed, is a claim the evidence rarely supports.

There is also a timing problem nobody in this industry likes to mention. Site-level quality changes usually take months to register. Google says they do not always wait for a core update, but large movements often coincide with one, so a fair before-and-after needs a long window with any updates inside it marked. A case study published six weeks after implementation is usually measuring noise.

Where we cannot separate our work from everything else that happened, we say so on the page rather than smoothing it over. That is why the records above describe what we did and who built it, rather than showing you a chart.

The research behind the method ↗
Talk to us

Want this level of detail on your site?

Send the URL and what you sell. You will get the first issue worth fixing, with the reason behind it.

Agency rebuilding a client’s site? See the launch handoff, from $1,200. Building something else? See how we partner.

  1. 01
    The one issue holding the site backNot a list of everything a crawler can flag.
  2. 02
    The evidence behind itWhat we saw, where, and the source we checked it against.
  3. 03
    A next step sized to the problemSometimes a small fix. Sometimes no project at all.
Where every engagement startsRequest an initial review
  • Reviewed personally by our team
  • A reply within one business day, usually
  • By email. A call only if you want one