Case Study

Why AI can't agree on your green fees: the pricing consistency problem

In our full panel test of St Mellion Estate, one finding sat quietly inside the Content Quality category rather than headlining the report: green fee figures for the same course were quoted as both £60 and £80 depending on which third party source you checked. Not a typo, not an outdated page, two genuinely different figures circulating for the same product. It's a small looking finding with an outsized effect on AI recommendation, and it's worth its own explanation because it's one of the most common, least understood gaps we see.

Why a model can't just pick the right number

When an AI assistant is asked "how much does it cost to play at St Mellion," it isn't visiting the club's live booking engine and reading a real time price. It's drawing on whatever text it has indexed about the club, which includes the club's own site, third party golf directories, review platforms, and press coverage. If two of those sources disagree, the model has no reliable way to know which one is current or correct. Its options are to guess, to hedge with a vague range that satisfies neither figure, or to omit price entirely rather than risk stating something wrong.

Why omission is the worst outcome

Of those three options, a hedge is annoying but survivable. A wrong guess damages trust in a way that's hard to recover, since the next person who calls to book at the quoted price and finds it wrong blames the club, not the AI. Omission is the outcome we see most often in practice, and it's the quietest kind of damage: the club simply doesn't get mentioned in answers about price, even though it would have won the comparison outright on facilities and reputation. A prospect asking "what's the best value golf resort in the South West" never sees a club whose own pricing data contradicts itself across sources.

Where the inconsistency actually comes from

The fix isn't chasing down every third party listing and asking for a correction, though that's worth doing where you can. The fix is making your own site the unambiguous, most current, most citable source, in plain text, so that when a model has to choose between competing figures, yours is the one that's freshest, clearest, and easiest to lift.

What a clean fix looks like

Publish one canonical current rate, or an honest banded range if pricing genuinely varies by day or season, as plain text on the page a prospect and a crawler would both land on, not buried inside a booking tool. State clearly what the figure includes and excludes, and update it the moment the actual rate changes rather than leaving last season's number live. Pair that with LocalBusiness or GolfCourse schema carrying a matching priceRange field, so the structured data and the plain text agree with each other and with reality.

The wider point

This finding didn't come from a golf specific quirk. Pricing inconsistency between a business's own site and the third party sources AI models are drawing on is one of the most common gaps we find across every category we test, and it's rarely the business's fault in the sense of doing anything wrong. It's simply never been anyone's job to notice that the internet disagrees with itself about your own prices, until an AI assistant started reading all of it at once and had to decide which version to trust.

Find out if your pricing data agrees with itself.

Our AI Recommendation Score checks content consistency across categories, then hands you a ranked fix list. Start with a free AI Snapshot.

Get my free AI Snapshot