What breaks for clients - and how it gets fixed.
Written from real orders: totals that lie, imports that get rejected, AI code that dies the moment it meets live data. To the point, no sales talk.
By Ben Ison · Python / AI Engineer
Read now
The form said thanks. Did anyone get the message?
The form showed a thank-you and the email never arrived. How to tell in a few minutes whether leads are dropping, and how to count the ones already gone.
Read →ChatGPT wrote your script. Your real data broke it.
It ran perfectly on ten sample rows and fell over on your actual file - or worse, did not fall over and handed you a number.
Read →The Import Said Success. Which Rows Are Missing?
The importer reported success and quietly skipped rows. How to find which ones, on a copy, before you touch the live catalog.
Read →Why your Excel or Google Sheets total is wrong: 7 real causes, ranked
A 60-second triage that names the cause, then the fix for each of the seven - including the two that never show an error.
Read →Coming next
Invoices from a spreadsheet row: the boring build that works
One template, one row, one click - and the guard formulas that stop a broken invoice reaching a client.
In progress →Google Sheets as a small CRM: where the limit really is
How far a spreadsheet carries your customer records, and when it's honest to say "time to move".
In progress →404 after a migration: finding the cause before repairing anything
A read-only way to tell a redirect problem from a plugin problem from a server problem.
In progress →Same problem, but in your file?
Describe what's broken. I'll restate it in my own words, give a plan and an honest deadline.