Most AI consultants pitching local business are selling chatbots or theoretical AI, with nothing behind it. Here's what I've actually built and run — including the parts that didn't work.
Both of these get judged by the same rule I learned in 1981, before AI existed: garbage in, garbage out. What that means in practice →
A real local IT repair and support business — the one I run alongside OptimisedLead, in public, including the parts that didn't work.
Started as buying old iMacs and MacBooks, refurbishing them, and reselling — moderate success, not a fortune. Pivoted to a repair-and-OS-update service when that made more sense than chasing resale margins.
What I've tested — results differ: leaflets, Google ads, and Facebook ads were all tried and all failed to convert. Magazine ads are more mixed — I still advertise there, just for a smaller return than a few years ago. They're still one of the few ways to land a message directly in local homes, and worth it for brand awareness, but the cost doesn't always match the return the way it used to. SEO is largely dead now too, since local traffic has shifted to Facebook and YouTube reels. Not every channel is a flat failure — some just don't pay what they used to.
What's working, still early: organic short-form video. Underway, unproven at scale, but the first channel that's shown real signs of life after everything else failed.
Why this matters to you: every month, a plain review goes up — what happened, what moved, what didn't — so you can watch the actual struggle to get traffic, not a highlight reel. If you run a local business, this is probably closer to your own experience than anything a polished agency will show you.
"Marcus is very knowledgable in this area and he has helped me no end... I don't know how I would have resolved all my tech issues with[out him]." — Paul Maxwell
"My iMac has been completely transformed by Marcus — has literally saved me from having to buy a whole new device. I would most definitely recommend." — Aidan Phillips
5.0 rating on Google, real client reviews — not curated testimonials.
A client needed multiple wholesale suppliers connected to their retailers — orders, stock, couriers, payments, the lot. This was their product, not mine — I was brought in to build it. A real piece of software, built with heavy AI assistance, shipped, and since sold on to a real buyer, who's onboarding now.
Why they came to me, specifically: not mainly for the code. They already knew, from experience, that I could produce something robust — I've been coding since I was a kid, long before AI, and I know exactly where it genuinely speeds up development and where it doesn't. But that was the smaller half of it. What they actually needed was someone who understood the business: how suppliers operate, how retailers actually place orders, where the friction and the tech-aversion really sit — and someone with a sales background, so I knew what would actually make a sales agent's life easier day to day, not just in theory. Small things like autocomplete and lead-finding come from having done the job, not from a spec document.
They could have gone offshore for cheaper code. What they couldn't get offshore was someone who already spoke both languages — business and technical — without a project manager translating between them. That's a real part of why this took weeks, not years: not just AI speeding up the typing, but not needing six rounds of "that's not what we meant" to get the brief right in the first place.
The real problem: suppliers and retailers in this world are notoriously tech-averse. Ask them to upload a spreadsheet or fill in an account form, and they'll go straight back to phone, pen, and paper. So the design had to remove friction, not add features.
Accounts created automatically. Auto-login links sent out — no password to forget. The retailer view kept dead simple. The supplier view even simpler. All the complexity — data insights, order tracking, courier integration, the ledger — lives in the admin view, out of the way of the people who just need to place an order.
Courier tracking plugged straight in. Gap analysis catches missing orders before they turn into a phone call. Full ledger and communications trail, so nothing falls through the cracks.
The platform's since been sold. Buyer's onboarding now.
Kept deliberately outcome-level — no product name, no code, no architecture — by agreement with the buyer. What's shareable is the back-and-forth that actually got us to a solution.
Phase 1 — working out what problem we were actually solving
Before software came up at all — how the ordering process actually worked, day to day.
Gave a straight answer on why none of them actually fit a wholesale business at this scale.
Not the default answer — the tested one, reasoned through together before committing to it.
Phase 2 — turning that into a real brief
GDPR-compliant handling of customer data, a full ledger, communications built in rather than scattered across email, room to handle real visitor volume without falling over, and properly secure — not just a password on the login page.
Turned those requirements into an actual feature list, not just a wish list.
Back and forth until both sides were confident it was right — the build only started after that, not before.
A few of the things that came out of it, kept deliberately vague: the entire order lifecycle — supplier confirmation through to retailer invoicing and any refund — runs as a single automatic cascade, so nothing needs manually pushing from one stage to the next. Every order status shown to a retailer and to the admin side comes from exactly the same source, so the two views can never quietly disagree about where an order actually stands. And commission owed to sales agents is tracked as a running ledger, updated automatically, rather than reconciled by hand at the end of the month.
Real documentation throughout, iterative AI-assisted development, complexity pushed into the admin layer so everyone else gets simplicity, and a genuine sale at the end.
Not that I'm the finished expert with all the answers. That I've actually built real things, with real constraints, for real people who don't want to think about technology — which is exactly the audience OptimisedLead is built for.