The return nobody planned for
Most retailers treat returns as a policy decision. By the time the item reaches the dock it has become a capacity problem, and policy cannot help.
Home Insights
Field notes.
Practical notes from more than thirty years inside UK retail — what holds up in production, what quietly falls over, and the questions worth asking before anyone signs anything.
Your data is fine. Your customer record isn't.
Personalisation programmes rarely fail on the model. They fail earlier, on the unglamorous question of whether you can tell that two records are the same person.
Availability is a promise, not a number
The stock figure in your system and the stock a customer can actually buy are different things. The gap between them is where trust gets spent.
Replatforming rarely fixes what is actually broken
Three years, a new stack, the same complaints. Usually because the pain was never in the platform to begin with.
What we'll cover
The subjects we keep being asked about
These come up in nearly every first conversation, and they run through the notes above.
Where AI actually pays off
How to spot the processes worth automating, and the tell-tale signs of the ones that aren't.
Why pilots stall
The predictable reasons promising proofs of concept never reach production, and how to avoid them.
Agents, honestly
What autonomous systems can genuinely be trusted with today, and what still needs a person in the loop.
Getting your data ready
The minimum a business needs in place before an AI project is worth starting at all.
Buying vs building
When an off-the-shelf tool is the right answer, and when it will cost you more than a custom build.
Measuring what changed
Setting success measures before the build, so you can tell afterwards whether it worked.
Rather just ask us directly?
You don't need to wait for us to write it up. Bring the question and we'll answer it against your actual situation.