Table of Contents
Walk into almost any retail business right now and you’ll find AI being “used”. Someone’s summarising a document. Someone’s drafting an email. Someone’s asking ChatGPT for a few headline ideas.
And yet… most teams aren’t feeling a real lift in the quality of their work.
That mismatch is the gap I want to talk about here. Because the problem isn’t that retail teams don’t have access to AI tools. Most do.
The problem is that people are using AI with the wrong habits, usually the same habits they’ve built for Google.
The good news? Closing the AI skills gap doesn’t mean becoming technical. It means taking a skill retail professionals already have (like writing a good brief) and pointing it at a new tool.
If you want to close the AI skills gap in retail, this is the lever: better briefs, not better buzzwords.
The AI Skills Gap in Retail Is Wider Than Adoption Stats Suggest
A lot of AI adoption headlines in retail are broadly true. Tools are spreading quickly.
But adoption is not the same thing as capability.
If 89% of retail staff have never received any formal AI coaching (British Chambers of Commerce), then most people are learning by trial and error. A bit of YouTube. A few LinkedIn tips. A colleague who “seems good at it”.
And that kind of self‑taught learning produces a predictable pattern:
- People find one or two low‑risk uses that work (draft an email, summarise a doc)
- They stop there
- The higher‑value uses feel uncertain, so they go untried
Retail Economics (December 2025) suggested around three‑fifths of retail tasks could be AI‑assisted by 2035. But the important part is the caveat. The gains only materialise when you close the skills gap first.
Or put more bluntly:
AI’s impact isn’t additive. It’s multiplicative.
Tool × skill.
If the skill is close to zero, the return is close to zero, even if the tool is brilliant.
For retail professionals who are still building confidence with the basics, I cover the fundamentals in epsiode 5: AI Literacy for Retail Professionals. What follows here is about what comes next; the gap between “I’ve tried it” and “I’m getting real value”.
The search engine habit is the real culprit
Most underwhelming AI outputs aren’t caused by the model. They’re caused by the prompt.
And specifically, they’re caused by a habit people have been practising for twenty years. The search engine habit.
In other words, the AI skills gap in retail is often just “search behaviour” carried into a tool that doesn’t work like search.
Google was designed to tolerate vague inputs. You can type something underspecified and still get a page of links you can interpret yourself.
AI doesn’t work like that.
A vague prompt doesn’t return “no results found”. It returns a confident, fluent answer built on guesswork. That’s what makes it tricky.
The output often sounds fine. It just answers the wrong question.
And in a retail environment, where time is tight and “good enough” is often the default, that’s how poor AI use becomes normalised.
It’s also a familiar pattern if you work in customer experience. The distance between intent and impact can be invisible when the surface looks smooth. (I’ve written about that perception gap in CX in episode 23 here.)
The Jagged Frontier: why teams stick to low‑value AI use cases
There’s another reason people retreat to the basics.
AI capability isn’t a neat line between “easy” and “hard”.
Research led by Ethan Mollick at Harvard with Boston Consulting Group described what they call the Jagged Frontier: AI performance is uneven. Two tasks that look similar to a human can produce very different outcomes from an AI tool.
That unevenness is a big reason the AI skills gap in retail persists. People don’t trust where AI will hold up, so they keep it on the shallow stuff.
So you get this strange experience:
- AI can produce a surprisingly solid strategic document draft
- Then it can fail at something that feels simple (like counting, or pulling the right items from a list)
- Then it’s brilliant again… somewhere else
That unpredictability makes people conservative.
If you can’t predict when AI will perform well, you stick to the uses that feel safe. Summaries. First drafts. Obvious admin.
And that means the uses that would actually change the quality of your work, like synthesis, briefing, decision support, stakeholder messaging, stay on the “maybe it’ll work, maybe it won’t” side of the frontier.
This is also why some retailers are treating AI adoption as a training problem, not a tech deployment. Sainsbury’s framing of colleague empowerment alongside their Microsoft partnership is one example. But it’s still the exception.
What a useful prompt actually looks like (and why it’s basically a good brief)
Here’s the simplest way to think about prompting:
A good prompt is a good brief. It’s not long. Not fancy. Just properly constructed.
And if you’re trying to improve AI outcomes across a team, this is the most teachable, repeatable place to start.
In practice, four things separate a vague prompt from a useful one:
- Who’s asking (role)
- What’s happening (context)
- What “useful” looks like (output criteria)
- What to ignore (boundaries)
Let’s make that concrete.
Imagine you have verbatim customer feedback from a post‑purchase survey and you want patterns.
The vague version: “Summarise the key themes from this customer feedback.”
You’ll get a list of what customers mentioned. It will be tidy. It might even be accurate. But it’ll also be generic. The kind of summary you could have written yourself in twenty minutes.
The constructed version:
- Role and context: “I’m a CX manager at a mid‑market fashion retailer.”
- Situation: “We’ve had a spike in complaints about our online returns process over the last six weeks. The feedback below is from a post‑purchase survey across that period.”
- What useful looks like: “I’m preparing a briefing for our commercial director. I need to understand which issues are most likely to affect repeat purchase, not just overall satisfaction.”
- What to leave out: “Don’t give me a general list of everything. Focus on patterns a commercial audience will care about, things that link experience to business outcomes.”
Same data. Different brief. Completely different result. And nothing about that is technical. It’s just professional clarity.
Closing the AI skills gap. Retail teams already have the core skill
The skills that make AI useful are the same skills that make any retail work better:
- setting context
- defining purpose
- specifying the audience
- being clear on what decision the output needs to support
- knowing what to leave out
This is the practical way to reduce the AI skills gap in retail without turning everyone into “AI people”.
If you work in retail marketing, CX, insight, ops, or commercial roles, you already do this when you brief:
- an agency
- a research partner
- a colleague
- a supplier
The AI skills gap, for most retail teams, isn’t a capability problem. It’s a transfer problem.
Most people haven’t been told that prompting well is structurally the same skill as briefing well, so they keep using AI like a search engine and then conclude that it’s underwhelming.
Bottom line? You don’t need new people. You need to unlock the capability in the people you’ve already got.
And once you see the AI skills gap in retail as a briefing gap, it becomes fixable, quickly.
Key Takeaways
- The AI skills gap in retail is primarily a prompting problem, not a tool problem. Most teams already have access to capable tools.
- The search engine habit doesn’t transfer. Vague prompts produce fluent answers that can miss the brief without looking “wrong”.
- The Jagged Frontier makes AI feel unpredictable, so teams retreat to low‑value use cases. A structured briefing approach reduces that unpredictability.
- Retail professionals already have the core skill: briefing. The opportunity is to transfer it.
Let’s connect – Find me on LinkedIn (https://www.linkedin.com/in/jo-williams-ccxp/)
Never miss an episode – Subscribe to Where’s Your Customer? wherever you get your podcasts and get fresh customer experience insights in your ears weekly.
Found this valuable? Please leave me a review. It helps other retail professionals discover these conversations and honestly means the world to me.
Share the knowledge – If someone in your network could benefit from the ideas we’ve talked about today, please share this episode with them.
