Personalisation and search
Recommendations and on-site search that lift conversion and basket size while staying inside the consent and profiling rules that govern them.
Sector programmes · Retail and e-commerce
AI in retail,
profitable and fair.
Independent, standards-led support for retailers and e-commerce businesses putting AI to work: grounded in UK data protection and marketing law, competition expectations, and your own brand standards, free of vendor incentives.
— State of play —
The algorithm is already
on the shop floor.
In retail, AI is not a pilot programme; it is the operating model. Recommendations drive a meaningful share of online revenue, forecasting models decide what gets bought and where it sits, and pricing engines move faster than any trading team. With margins thin and customer acquisition expensive, the gap between retailers who use these tools well and those who rent them blindly is widening every quarter.
The constraints are tightening at the same pace. UK GDPR and PECR govern profiling and targeted marketing, the CMA has algorithmic pricing and online choice architecture firmly in view, and customers punish anything that feels manipulative. The winning position is the boring-sounding one: models you understand, data you are entitled to use, and personalisation your customers would not mind seeing explained.
1 in 4
UK retail pounds is now spent online
~35%
of sales on leading platforms attributed to recommendations
<1 in 4
retailers with AI governance in place
— Use cases —
Where AI is already
earning its keep.
The strongest retail use cases protect margin and trust at the same time, with merchants and marketers still making the calls.
Demand forecasting
Buying, allocation, and markdown decisions informed by models that read seasonality and trends, owned by merchants rather than dictated to them.
Pricing and promotions
Dynamic pricing and promotion planning that respond to the market within fairness and competition-law boundaries set in advance, not discovered in a headline.
Customer service
Assistants that resolve routine contacts on sizing, stock, and delivery around the clock, and know precisely when to hand a frustrated customer to a person.
Supply chain and returns
Stock accuracy, returns triage, and payment fraud screening that recover margin quietly across thousands of daily decisions.
Marketing compliance
Profiling, consent, and targeting practices that satisfy UK GDPR and PECR, audited before the regulator or the press does it for you.
— How the Institute helps —
One institute,
three kinds of help.
Advice, engineering, and professional standards from a single independent body, with no software to sell.
Consultancy
Strategy, governance, and compliance
Independent advisory for trading directors, e-commerce leaders, and boards setting an AI position that protects margin without gambling the brand.
- AI strategy across trading, marketing, and ops
- Profiling and personalisation compliance reviews
- Readiness assessment and a costed roadmap
Solution design
Working systems, handed over
The Institute's engineering practice designs and builds AI systems around your commerce stack and data estate, then hands them to your team.
- Forecasting, service, and content tools
- Integration with your existing platforms
- Knowledge transfer to your own staff
Professional body
Standards, training, and research
As the UK's professional body for AI, the Institute trains and accredits staff, publishes research, and maintains the UK AI Readiness Charter.
- Staff training and accreditation pathways
- The UK AI Readiness Charter, free to sign
- Published research and sector guidance
— Sector considerations —
What we watch
most closely.
01
Privacy
Personalisation has a lawful basis
Profiling customers for recommendations and marketing sits squarely inside UK GDPR and PECR. Every segment, signal, and trigger maps to a lawful basis and a consent state you can actually evidence.
100%
of profiling mapped to a lawful basis
02
Fairness
Pricing must survive scrutiny
Algorithmic and personalised pricing draw attention from the CMA and from customers, who talk to each other. Pricing logic gets fairness boundaries set in advance and someone who can explain it in plain English.
0
prices personalised without disclosure
03
Brand
Scale must not flatten the voice
Generated product copy, images, and campaigns at volume can quietly sand the character off a brand. Tone-of-voice controls and editorial review keep the output sounding like you.
1
brand voice, however much is generated
04
Customers
False positives are real people
Fraud and returns-abuse models inevitably flag honest customers. The cost of a blocked order is a lost customer and a public review, so high-impact declines get a human look first.
100%
of high-impact blocks human-reviewed
05
Resilience
Peak is the only test that counts
A model that degrades quietly in November costs more than one that fails loudly in March. Peak-trading load, fallbacks, and manual overrides are designed in before the golden quarter, not during it.
0
single points of failure at peak
Set your brand's
AI position.
A short call with the retail team. Plain answers, a clear next step, and no software to sell.
— Go deeper
— Consultancy
Independent advice
Standards-led advisory for boards and leaders, with no software to sell.
— Solutions
Need it built?
Working AI systems designed and built by the Institute's engineering practice.
— Sectors
Explore other sectors
The full directory of IoAI sector programmes, grounded in UK regulation.