After two days at Hospitality Tech Expo, I lost count of how many times I saw the letters AI.
What I heard far less about was the quality of the information being fed into it.
AI is being presented as the answer to almost every operational challenge in hospitality. Better forecasting. Faster reporting. Smarter purchasing. Improved labour planning. More accurate margins. Less administration.
All of those things are possible.
But AI cannot rescue an operation built on inaccurate recipes, poor stock discipline, fragmented systems, unclear ownership and teams that do not trust the data.
It may simply produce incorrect answers faster.
The industry is already paying more for technology
Hospitality operators are not short of software.
Most businesses now rely on a collection of systems covering sales, payments, stock, purchasing, labour, reservations, finance and reporting. Each one may solve a particular problem, but the overall result is often a patchwork of platforms that do not communicate properly.
That fragmentation has become one of the industry’s most persistent technology frustrations. Hospitality Tech360 has described data integration as the single biggest barrier to digital progress in hospitality, with operators managing multiple systems across POS, reservations, labour, inventory and CRM.
At the same time, the cost of that technology has risen sharply. Research reported by Restaurant Online found that technology and software costs for pubs and bars increased by 167% between 2015 and 2025.
If the industry is paying considerably more for technology, it should expect measurable operational value in return.
Another dashboard is not value. Another login is not value. A report that arrives after the opportunity to act has passed is not value.
Technology creates value when it helps somebody make a better decision, take action sooner or remove work that should not need to be done manually.
More revenue is not solving the problem
The financial pressure makes this more urgent.
IGD reported that 23% of hospitality businesses were operating at a loss. It also cited research showing that combined profits among the UK’s 100 largest restaurant groups had fallen by 44% year on year, from £365 million to £204 million, even as revenue rose from £12.9 billion to £13.3 billion.
That is the uncomfortable reality behind the excitement about AI.
The industry does not simply need more sales, more data or more software. It needs tighter control over the relationship between revenue and profit.
AI could play a valuable role in that. It could identify unusual stock movements, highlight recipe cost changes, flag margin risks, spot purchasing patterns and help managers focus on the exceptions that need attention.
But it can only work with the information it receives.
If a product has the wrong pack size, the calculation will be wrong.
If a recipe does not reflect what the kitchen actually produces, the theoretical margin will be wrong.
If purchases, transfers and waste are recorded inconsistently, the stock position will be wrong.
If sales, stock, workforce and finance systems all describe the business differently, the answer may sound intelligent while being operationally useless.
The foundations useful AI actually needs
Before asking what AI can do for a hospitality business, I think operators should ask whether the business is ready to give it reliable information.
That requires seven foundations.
1. Accurate product information
Products, pack sizes, units, prices and supplier details need to be correct and consistently maintained. A system cannot calculate an accurate cost if a case, bottle, kilogram and individual unit are being confused.
2. Recipes that reflect reality
A recipe in a system should represent what the kitchen genuinely produces, including realistic yields, portions and preparation losses. A beautifully costed recipe that nobody follows is not a control. It is fiction.
3. Consistent stock movements
Purchases, transfers, returns, waste, staff food and complimentary items need to be recorded in a consistent way. Missing movements do not disappear. They reappear later as unexplained variance.
4. Connected operational systems
Sales, stock, purchasing, workforce and finance systems need to exchange reliable information. If teams are repeatedly downloading spreadsheets, rekeying figures or manually reconciling different versions of the truth, the operation is not ready for more complexity.
5. Clear ownership
Someone must be responsible for maintaining products, recipes, supplier information and processes. If everybody assumes somebody else is looking after the data, nobody is.
6. Teams that understand and trust the process
People are more likely to follow a process when they understand why it matters and can see that the information produced is useful. When teams stop trusting the numbers, they create workarounds. Those workarounds make the data even less reliable.
7. AI that leads to action
Operators do not need another screen full of information to monitor. Useful AI should identify what has changed, explain why it matters and direct attention towards the action that will make a difference.
The goal should not be to produce more analysis. It should be to reduce the distance between information and action.
AI readiness is operational readiness
The most useful question may not be, “Which AI tool should we buy?”
It may be, “What would this tool be learning from?”
If the answer is inconsistent product data, theoretical recipes, missing stock movements and systems that do not agree, the priority is clear.
Fix the foundations first.
This is not an argument against AI. Used well, AI could give hospitality teams faster access to insight, reduce administration and help operators spot risks before they become expensive problems.
But intelligence cannot compensate for a process nobody owns or data nobody trusts.
Hospitality does not need technology that makes weak operations look more sophisticated.
It needs technology that makes strong operations easier to run.
Leave a Reply