For small businesses & startups

Software trouble, in plain English

No tech words needed. If something in your business keeps going wrong and nobody can tell you why, you’re in the right place.

Most small businesses run on software someone else set up: a website, an online shop, a booking system, and a few spreadsheets and apps that are meant to agree with each other. When one of them goes wrong quietly, you’re usually the last to find out.

Who I help

You don’t need a tech team. That’s the point.

A few people, and nobody whose job is the software

Shops, salons, clinics, trades, cafés, studios and charities. Your setup was put together by a freelancer, an agency, a friend, or someone who has since left, and it mostly works.

Built quickly, and now real customers rely on it

Maybe you built it with AI tools or a no-code builder, or a developer built it and moved on. You can’t read the code yourself, so when something goes wrong, it’s hard to tell where.

Common problems

Does this sound like your week?

Selling online

  • You sold something that was out of stock, and found out when the customer emailed.
  • The price on your website doesn’t match the price in your shop or spreadsheet.
  • A discount code kept working after the sale ended.
  • Some order confirmation emails never arrive, and nobody knows which ones.

Bookings & appointments

  • Two customers turned up for the same slot.
  • Reminder messages stopped going out, and you only noticed when no-shows went up.
  • You changed your opening hours, and the booking page still offers the old ones.

Money & paperwork

  • Invoices sometimes go out with the wrong total or the wrong VAT.
  • The numbers in your accounts don’t match the numbers in your shop.
  • A payment shows as paid in one place and unpaid in another.
  • A weekly report quietly stopped updating, weeks ago.
  • Your spreadsheet now feeds your Making Tax Digital updates, and you’re not sure the figures it sends are the right ones.

AI tools & new apps

  • An automation linking two of your apps stopped running, and nothing told you.
  • Your AI chatbot tells customers things that aren’t true, like an old price or a refund policy you don’t have.
  • Some people can’t sign up, and the screen only says “Something went wrong”.
  • Every update fixes one thing and upsets another.
  • The person who built it has moved on, and everyone is nervous to touch it.

None of this means anyone did a bad job. It’s what happens when systems that were set up separately are expected to agree, and nothing checks that they still do.

Three made-up examples

What it looks like in practice

These businesses are invented. The problems are real ones, the kind that turn up in small businesses all the time.

Example 1 · made up · online shop

The candle shop selling stock it didn’t have

What they noticed

They sell on their own website, on Etsy and at weekend markets. Every few weeks, someone ordered a candle online that had already sold out somewhere else. Refunds, apologies, and a couple of unhappy reviews.

What was really going on

Stock counts were shared between the three through an automation, set up with an AI tool, that ran each night. When it failed, nothing said so, and the website kept selling yesterday’s stock.

What the review gave them

A short note in plain English: where the automation could fail, why nobody was told, and the first thing to change. £200 fixed.

What happened next

They chose a First Fix: the owner now gets an email when the automation hasn’t run, and an item pauses itself if its stock count is more than a day old. £450 fixed.

Example 2 · made up · clinic

The clinic with two patients in one appointment

What they noticed

Now and then, two patients arrived for the same appointment. Reception blamed the website, and the website company blamed the phones.

What was really going on

Bookings came in online and by phone, straight into the staff calendar. Each one checked itself, but neither checked the other, so a slot could be taken twice within minutes.

What the review gave them

Findings that showed exactly when a double booking could happen, and one recommendation: let one diary be in charge, and have the other ask it first. £200 fixed.

What happened next

Their booking provider already had a setting for this. They switched it on themselves, following the steps in the findings, and needed nothing more from me.

Example 3 · made up · startup

The startup losing sign-ups without knowing

What they noticed

A small team launched an app built quickly with AI tools. Sign-ups were lower than they expected, but nothing looked wrong.

What was really going on

The sign-up form turned away some real names, like O’Brien or names with accents, and only said “Something went wrong”. Those people just left, and the team was never told.

What the review gave them

I read the code the AI tools had written, so they didn’t have to. Findings ranked by how many customers each problem was likely to affect, starting with the sign-up form. £200 fixed.

What happened next

Their own developer fixed the form from the findings. Every name gets through now, and if something else stops a sign-up, the message says what to change.

Step by step

How working with me goes

  1. Tell me what keeps going wrong

    In your own words, by email or the contact form. Three short questions, nothing to download, and “I don’t know” is a fine answer. You’ll hear back within one working day.

  2. We pick one part to look at

    Together we choose the part most likely to be causing it, like the booking system or the stock update. The price is £200 fixed, and nothing starts until you say yes.

  3. I look, but I don’t touch

    I read how it’s set up without changing anything, so your business carries on as normal. Within 5 working days you get findings written for you, not for a programmer, plus one follow-up.

  4. You decide what happens next

    Fix it yourself, pass the findings to whoever looks after your software, or ask me for a First Fix at £450 fixed. I’d do it in a copy you control, never in the system your customers are using.

Jargon, translated

Words you’ll see on this site

Read-only
I look at how something is set up. I don’t change anything.
Findings
A short written list of what’s going wrong, most important first.
Slice
One part of your setup, like your bookings or your stock update, rather than everything at once.
Live system
The one your customers are using right now. I never work in it directly.
Guardrail
A check that stops a wrong number or a missing detail before it causes trouble.
Silent failure
When something goes wrong and nothing tells you. It’s the main thing I look for.

Why trust me with it

Careful hands, plain answers

  • You deal with me directly, from the first email to the findings. No account managers and no hand-offs.
  • Every price is fixed and agreed before any work starts, so there are no surprise bills.
  • I lead a tool in the European Space Agency’s open-source Pyxel project, and engineers there named my work in a published research paper.
  • Registered with the ICO, the UK’s data protection regulator. Check the public register.

Who this isn’t for: if something is down right now and you need someone within the hour, I don’t do emergency call-outs. And if you already have a tech team with someone on call around the clock, you probably have this covered.

The first step

Ready when you are

Describe what keeps going wrong, in your own words. If a review isn’t the right fit, I’ll say so.

Starter Reliability Review

£200 fixed

  • One part of your setup, chosen together
  • I look, but nothing is changed
  • Findings in plain English, plus one follow-up

Within 5 working days of your yes · nothing in your system is touched

Starter inquiries answered within one working day.

Tell me what keeps going wrong → See a sample findings report →
Doby Baxter

Doby Baxter

I read every message myself, and I’ll answer in the same plain words you use.