Blog
Notes from the work.
Solutions, delivered.
What we learn fixing real systems: the causes we found, how we proved them, and what we would check first on your site.
A site that updates itself, and checks every page before it does
A new affiliate site publishes fresh data twice a day and new content every morning, with no one at the keyboard. How the automation works, and the one rule that makes it safe to leave alone.
Decide what success means before the test starts
One way of wording a title matches 28,413 searches a month; the other, 24. Before we test which wins, we write down how we will judge it. Why that order matters.
Auditing the pages nobody reads: category and section pages
Section pages hold a site together, and they are rarely reviewed. We audited 300 of them: 4,575 findings, 255 rewritten, 81 cut to half their length.
Check a warning against a known answer before a manager sees it
Pipeline warnings that nobody has checked are just noise with confidence. So before launch we built a test account where we knew every right answer, and the product had to match all 23.
Put the offer where the product is
A new affiliate site led with the market’s biggest topics, but its partner sold something narrower. We researched the partner in a day and rebuilt the home page the next. Why matching offer and page matters more than traffic.
101 releases in 10 weeks, without breaking what earns
Speed and safety are usually traded off. On our current engagement we shipped 101 reviewed releases in 10 weeks on a site that earns every day. How.
AI content that does not make things up
430 data-led articles in two weeks, in two languages, with no invented numbers. The trick is not a better prompt: it is a pipeline where the AI writes and the data decides.
Before you buy a website: six questions due diligence should answer
Traffic screenshots and revenue spreadsheets are where a site sale starts, not where it should end. Six technical questions to answer before money changes hands.
Redesign without losing what ranks: compare every page before launch
A redesign is the most common way to lose traffic you already earned. How we rebuilt a site, cut phone load time to 1.9 s, and compared 3,050 pages before anyone saw it.
On Google the same day: how new articles get found fast
An article that went live at 09:18 was seen 547 times in search that day. New pages averaged position 4.8 within 13 days. What made the difference.
Less site, more traffic: why we cut 85% of the pages
15,587 pages brought one visit between them in 25 days. Why a smaller site can earn more, and how to cut without losing anything you might need back.
From zero automated checks to 918 in under nine weeks
Sixteen weeks of a product had no automated checks. By its September release it had 918, and a gate that blocks any release when one fails. What we built, and the two times the safety net itself needed fixing.
The clean-up job that was deleting page-one articles
An automated clean-up removed articles two days after publishing, including 670 ranking on Google’s first page. How we found it, and the audit we now run on every site.
One system, many customers: proving their data stays apart
A product that holds several companies’ sales calls must never let one see another’s. Saying so is easy. Here is how we checked it on the live system, and what we built so the answer stays true after every change.
How one person produces thousands of lessons with AI
SuperEscuela has 733 topics, 2,786 lessons and 13,705 practice questions, written by an AI pipeline against Costa Rica’s national curriculum. How the pipeline works, and where the AI is kept out on purpose.
Your investors will check your numbers
Traction is the story of the next round, and investors increasingly check it. Two real cases of numbers that looked fine and were not, and the checks that would have caught them before a board deck did.
Write the rulebook before the first page
Before building a new affiliate site, we audited 26 sites in its market against 13 rules. None of the affiliates met all of them. Why the rules came first, and how they became part of the build.
A confident wrong number is worse than a missing one
A close rate that read 100% when the truth was 50%, feeding seven reports a sales manager would act on. How it happened, why we fixed it at the source, and what to check in your own dashboards.
Just funded? Instrument before you accelerate
New funding means pressure to show traction fast. The numbers you will be judged on are only as good as the measurement under them. What to set up in the first weeks.
Building a school platform with an AI pair programmer
97 code changes, 74 of them written together with an AI coding assistant. How we built SuperEscuela’s student app and teacher records system, and the rules that kept the AI useful and the data safe.
10% of searches bring half the visits
We sorted 431,227 searches by intent to buy. A small slice brought about half the visits. Why intent, not volume, should decide what you write and fix first.
How we date a traffic drop to within four days
“Traffic fell” is not a diagnosis. How we line up search data, releases and server logs until every step of a decline has a date and a cause.
Build on your own accounts from day one
Who holds the repository, the cloud account and the AI keys decides how easy it is to change partners, hire a team or raise the next round. Why we build on the client’s own accounts from the first day.
More than half of your “visitors” may be bots
On one site, 55% of the visits analytics reported were bots, and real engagement was 22.4%, not 8.9%. Why it matters for every decision you make, and how to separate people from bots.
Let customers connect their own CRM in one click
Self-serve software cannot ask every customer to copy an access key and book a call with their admin. How we built a one-click, read-only CRM connection, and why “what you did not share” deserves a screen of its own.
Your security policy may be hiding your revenue
A site’s own security settings were silently blocking its analytics and ad tracking, while one bot out-crawled Google. Two security findings from the first month of an engagement.
Test the paths that have never run
A billing error would have given top-tier customers a fifth of what they paid for. It was waiting in a feature nobody could test yet. How one round of hands-on review found it, and 30 other problems, before the first invoice.
Proven, likely or unknown: a weekly report your CEO can trust
Most status reports mix facts, hopes and guesses in the same tone. Three labels fix that. How we write the weekly report on every engagement.
Built for the oldest phone: offline-first, no passwords, no frameworks
SuperEscuela is a free school for Spanish-speaking children who may have an old phone and a weak signal. The technical choices that follow from that, and why “light” is a feature, not a compromise.
Ingestion, normalization, trust: feeding practitioner data to an AI assistant
An AI assistant for legal and healthcare practitioners is only as good as the records it can read. What we learned owning the integration and data layer of such a platform, and why calling an endpoint is the easy part.
The right model, not the cheapest: what a few cents per lesson buys
SuperEscuela started writing lessons with the smallest, cheapest AI model. We measured the quality against what children deserve, and moved to a larger one. Why the saving was never worth it, in numbers.
Senior team or first hires? The first 90 days after a round
After a raise, the instinct is to hire. Hiring is right, but it is slow, and the first months decide what the next round sees. When a senior team that ships from week one makes more sense, and how it should hand over.
Fourteen websites, one platform: lessons from consolidation
Many organisations end up running a dozen sites that drifted apart. What we have learned merging them into one platform, and shrinking the code while everything stays live.
How to trust an AI score
An AI that rates leads, deals or documents is only useful if people can check it. The six things we build around every model before its scores reach a sales team.
Scoring sales calls with AI, against your own scorecard
A sales-coaching client had a scorecard proven by hand. We taught a product to score real calls against it, with evidence quoted from every call. What made the scores worth trusting.
What the first eight weeks of an AI product look like
A sales-coaching expert had a method proven by hand, a clickable mockup and an 8-week plan. The core product was working in eight weeks, as planned. Week by week, what that took, and how risks were handled along the way.
Want us to look
at your site?
Tell us where traffic, revenue or your numbers stopped making sense. We will tell you what we would check first.
Prefer to write directly? enable JavaScript to see the address