Three cases, and what is missing from them
Most agency case studies are fiction with a chart. These are short because only what can be verified is here — where a number is not backed by a real report, it is left out and said so.
Six Months To Day 4
| Industry | e-commerce |
|---|---|
| Problem | A store had been open for six months with no sales. |
| What was done | An audit, a rebuild of the store, a photo and video shoot, and ads. |
| Result | The first sale came on day four after launch. |
| Services used | shopify redesign · meta ads management · website audit |
| Figures verified | No — revenue figures are not published because they are not ours to verify |
Audit Conversion
| Industry | Prodigy LAB's own agency practice |
|---|---|
| Problem | We needed to know whether a paid audit pays for itself as a first step. |
| What was done | Eight paid audits were carried through to presenting the findings on a call. |
| Result | Four of the eight turned into projects — a 50% conversion. |
| Services used | website audit |
| Figures verified | Yes — from the studio's own records |
Adaptive Site
| Industry | photography — the studio's own site |
|---|---|
| Problem | A portfolio site shows everyone the same thing. Someone who arrives from an ad about sports photography lands on a general showcase and leaves to look for what they came for. |
| What was done | We built the site on our own worker and connected a decision layer. It reads intent from the landing page, the ad's tags and what the visitor opened during the session, then picks a pre-approved variant of the block. Rules answer first; the model only refines, and only inside the prepared list — it cannot write copy of its own. If the layer is unavailable, the site falls back to the rules. |
| Result | Live since 12 October 2026. A visitor with no ad tags who opens the wedding work is shown the wedding variant, and the decision is marked as the model's own rather than a rule's. No conversion figures: the layer has been running for days, not months. |
| Services used | website development · ai chatbot · ai automation |
| Figures verified | No — revenue figures are not published because they are not ours to verify |
What the cases have in common
The owner misdiagnosed it
In both, the thing about to be bought was not the thing that was broken. The audit is what caught the difference.
The fix was smaller than expected
Neither needed everything rebuilt. Finding the actual constraint is most of the work.
The proof came fast or not at all
When the constraint is the right one, the change shows up in days, not quarters. When nothing moves in a month, the diagnosis was wrong and it gets said out loud.
Straight answers
Why are there so few numbers on this page?
Because only the numbers that come out of a real report are published, and most client data is not ours to publish.
A case study with invented percentages is worse than no case study: it is the same thing every agency does, and it teaches a prospective client to distrust everything else on the site. Where a figure is missing here, it is missing on purpose.
Can I talk to a past client?
Ask on a call and it will be arranged where the client agrees to it.
References are given directly rather than published, because clients did not sign up to become marketing material.
What does a typical engagement look like?
Audit first, then a fixed-price fix, then usually ongoing advertising or the bot.
The pattern in the cases below is the same: the thing the owner wanted to buy was not the thing that was broken.
Your situation is not on this page
It never is — that is the point of an audit rather than a template. The $495 full audit looks at your analytics, your ads and your funnel and says what is actually costing you sales. It is credited in full toward any project you order within 30 days.
See what the audit covers