AI Tagging and Suggestion Ranking
How items get tagged to a member’s wellness factors, and how the ranker then decides which of the matches a member actually sees.
Automated Tagging
Every item in the catalogue — a wholefood, an activity, a supplement, a sleep therapy, a mind therapy — needs to know two things: which of a member’s wellness factors it helps with, and which factors mean it should be held back. Those two lists are its tags.
What the tool does
The AI Automated Tagger reads the current tagging rules for a wellness factor, checks them against the real catalogue, and proposes changes — items that should be added, items that no longer belong, or wording that should be clearer. Nothing is applied automatically.
How to use it
1. Open the tool and press Verify. 2. Review each proposal it comes back with. 3. Accept the ones you agree with, skip the rest.
What happens when you accept
The moment you accept anything, the tool backs up the current rulebook first, then rewrites it and reruns tagging across the live catalogue. The backup happens automatically, before any change is made, so anything accepted by mistake can always be undone.
This step doesn’t change anything a member sees directly. It only changes which tags each catalogue item carries. Those tags are the raw material the Ranker (Part 2 below) then works from.
Tool: AI Automated Tagger
Ranking Suggestions List
Once a category has been tagged (Part 1), and those tags are matched against one member’s own wellness factors, that category’s page builds a full list of everything that matches. The Ranker then decides, out of that list, how many items the member actually sees, and which ones.
The budget
Every category has its own cap — a total number of items it’s allowed to suggest. Nothing beyond that number gets shown, no matter how many things technically match.
The scoring
Every item earns points based on how many of the member’s wellness factors it serves, and how strong each of those matches is — a stronger match earns more points than a weak one. An item that helps with three of a member’s factors will outscore one that only weakly helps with one. The highest-scoring items win the available places first.
The floor
Every factor is guaranteed a small minimum number of items, even a weak factor. This stops one strong factor from crowding out every weaker one completely.
The ceiling (Food only, for now)
Food also has two extra limits: a cap on near-identical items (so the list can’t fill up with five different cuts of beef), and a wider cap on one broad category (so Meat & Poultry can’t own the whole list either). Activities, Sleep Therapy and Mind Therapy each have one broad category but no narrow grouping, so only the wider cap applies to them. Supplements have neither.
What’s admin-only
The numbers behind all of this — the budget, the minimums and maximums, and the points each wellness factor is worth — can only be changed by an admin, on the Suggestion Ranker tool. A member never sees or changes these numbers themselves.
What’s member-only (Food only)
Food is the only category where the member also picks their own Diet and Region preferences. Those nudge the score slightly in the member’s favour, but never override a genuine health match. An admin can simulate what a Diet or Region choice would do on the Ranker tool, but can’t apply it to anyone’s actual list — that stays the member’s own choice, not a scored match.
Turning it on
Each category has its own on/off switch on the Ranker tool. Off means nothing changes for members yet, even if all the numbers are filled in — nobody sees a different list until an admin switches that category on.
Tool: Suggestion Ranker