A few months into building OmniWatchGuard (automated website change monitoring — price, stock, SEO, GDPR pages, etc.), I hit a moment where I realized I couldn't honestly answer "is this claim on my site true?" without going and checking the code first.
It wasn't deliberate dishonesty. It's the normal failure mode of solo-founder velocity: you write the marketing copy for a feature while planning it, ship other things first, and the copy just... stays. Months later it reads as a promise you're not keeping.
So I did a full pass. Process:
- Listed every file that generates customer-facing content — FAQ, pricing schema, comparison tables, JSON-LD, the pricing calculator.
- For every functional claim, went and found (or didn't find) the actual implementation.
- Bucketed each into: true, partially true (implemented but with unstated limits), or false.
- For anything false or partial, either fixed the copy or built the feature to make it true.
What I found:
- AggregateRatingSchema and Testimonials had a hardcoded 4.9 rating, "200+ reviews," "500+ companies," and testimonial quotes from customers who didn't exist. Replaced with real numbers computed from the reviews table — the section now hides itself if there are zero real reviews yet.
- Login-page monitoring, RBAC, and SSO were listed as available features in the FAQ and comparison table. They weren't built. Fixed six files to say "Roadmap 2026" instead.
- The homepage comparison table had an actual rendering bug — it showed a green checkmark for every single feature regardless of the real data behind it. That one wasn't even copy drift, it was a bug quietly lying to visitors.
For login-page monitoring specifically, I chose to build it rather than just correct the copy — added an encrypted cookie/header column, had the crawler inject it on both fetch paths, added session-expiry detection. Tested it end-to-end against my own dashboard before flipping the copy back to "available."
Full writeup with the before/after table: https://omniwatchguard.com/en/blog/product-transparency-audit
Anyone else done something like this? Curious what you found when you actually checked.