How Wigan Restaurants Can Use AI to Analyse Sales Data and Optimise Their Menu

You run a restaurant near Wigan Market with 42 dishes on the menu. You suspect five of them are dragging down the kitchen and eating into margins, but you don't have the time to sit down and work through the numbers. Most restaurant owners are the same: the data is there in the EPOS system, but turning it into useful decisions feels like a job for someone else. AI makes that job yours, and it takes an afternoon rather than a week.

Understand Menu Engineering Before You Start

Menu engineering is a method for classifying dishes by two variables: how popular they are (how often they sell) and how profitable they are (contribution margin, not just selling price). Every dish on your menu falls into one of four categories:

Stars: high popularity, high margin. These pay the bills.
Ploughhorses: high popularity, low margin. People love them but they're not making you much.
Puzzles: low popularity, high margin. If you could sell more of these, they'd be your best dishes.
Dogs: low popularity, low margin. These need to go.

Once you know which dish is which, you know where to focus.

Export Your Sales Data and Feed It to ChatGPT

Most EPOS systems let you export sales data as a CSV or spreadsheet. Export the last three months of dish-level sales: dish name, number sold, selling price, and food cost (or estimated margin if that's all you have).

Paste the data into ChatGPT and use a prompt like this:

"Here is three months of sales data from my restaurant. Each row has the dish name, number sold, selling price, and estimated food cost. Please categorise each dish as a Star, Ploughhorse, Puzzle, or Dog based on popularity and margin. Then suggest which dishes to promote, which to reprice, and which to remove."

ChatGPT will work through the data and give you a categorised list with reasoning. It won't be perfect, especially if your cost data is rough, but it gives you a clear starting point for a conversation you can have with your head chef or supplier.

Use AI to Rewrite Menu Descriptions

The way a dish is described on your menu affects what people order. "Chicken breast with roasted vegetables" sells less than "Free-range chicken, charred leeks, roasted heritage carrots, thyme butter." The second version justifies a higher price and sounds worth ordering.

ChatGPT can rewrite your menu descriptions in seconds. Give it your current description, the key ingredients, and a brief on your restaurant's tone (relaxed, upscale, traditional, modern British), and ask for three alternatives. Pick the one that fits.

For Puzzles, a better description can shift ordering behaviour without changing the dish or the price. If your high-margin lamb shank is sitting at the bottom of the menu with a flat description, a rewrite and a better position on the page might be all it needs.

Plan Seasonal Menu Changes With AI Support

Changing your menu seasonally keeps it fresh and lets you take advantage of cheaper, better-quality ingredients. ChatGPT can help you plan what to add: give it your current menu style, your rough food cost target, and the season, and ask it to suggest five new dishes that fit.

It can also help you plan transitions: which dishes to keep year-round, which to retire, and how to introduce new items without confusing regular customers. Use it to draft the text for a "new season menu" social post or email to your regulars too.

For a Wigan restaurant competing with chains, a seasonal menu that uses local suppliers and changes quarterly is a genuine point of difference worth marketing.

Track the Impact of Menu Changes

Once you've made changes, track whether they've worked. Set a baseline for the dishes you repositioned or repriced and check the numbers again after six weeks. You don't need complex software: a spreadsheet with weekly dish sales totals is enough to see trends.

ChatGPT can help you interpret the numbers. Paste in your before and after figures and ask: "Has removing the duck breast and promoting the braised short rib improved overall margin? What does the data suggest?" It'll give you a plain-English read on whether your changes are working.