MindClick
A supplier sustainability rating for a global hospitality brand. Three measures that had nothing to do with each other, joined into one grade a buyer could act on without leaving the page.
Is this supplier better or worse than the others I could buy from, and by how much?
The problem
Sustainability data does not arrive comparable
A procurement lead choosing between two suppliers is not running an ESG audit. They have a category to fill, a shortlist, and about a minute. They need a comparison.
What exists instead is three unrelated things. A facility reports emissions against whatever baseline year it picked. The same facility self-declares environmental and labour commitments, in prose. Somewhere else entirely, individual products get scored on their own attributes. None of the three share a unit, a scale, or a scope, and two of them are about the factory while the third is about the box.
Nobody can hold that in their head. So the work was not visualisation. It was deciding, defensibly, how three incompatible measures become one number.
The model
Three inputs, one join, one grade
Emissions against a baseline
What the facility actually emits now, measured against the year it started counting. Movement, not absolute size, so a small supplier is not automatically the winner.
Unit: change vs. baseline yearEnvironmental and labour commitments
What the facility says it is committed to, on the environment and on labour. Declared rather than measured, and weighted accordingly.
Unit: commitment coverageProduct scoring, nine dimensions
The item itself, scored across nine attributes. The only tier that answers “is this specific thing any good,” and the one a buyer recognises.
Unit: nine scored dimensionsThe join is the product. Everything downstream is presentation, and everything upstream is somebody else’s data collection. The decisions that matter all live in the middle: which tier dominates, what happens when a tier is missing, and whether a supplier who declines to answer is scored as bad or scored as unknown.
That last one is not a technical question. It is an editorial one, and it changes who wins.
The scale
Five tiers, and one of them is not a score
The rating a buyer sees has five positions. The first is deliberately not a grade.
The supplier did not answer. Reported as absent, never averaged into a grade as if it were a low one.
Answered, and below the floor. A real finding, not a missing one.
Has begun. Enough to be worth a conversation.
Meeting the standard across the tiers that were measured.
Ahead of the category. The bar the rest are compared against.
Keeping no response outside the scale is the single most consequential decision on the page. Fold silence into the low end and every scorecard quietly punishes suppliers for having a slow compliance team rather than a dirty factory, and the buyer cannot tell which is which.
Held separate, silence stays legible. A category that is 25 percent unanswered is telling the buyer something true, and it is telling the client where to go chase paperwork.
Missing is not the same as bad. If the chart cannot say that, the chart is lying.
The page
One supplier, one category, one comparison
The scorecard resolves to a single screen with the same three moves every time. Pick a product category. See this supplier’s distribution across the five tiers. See the category total beside it, as the thing to beat.
Two bars and a table. The bars answer the question at a glance and the table lets someone check the arithmetic, because a procurement lead who cannot check the arithmetic will not stake a contract on it.
Filter to the category. Ratings only mean anything within a category. Comparing a diaper supplier to a flooring supplier is noise.
Supplier against category total. Side by side, same scale, same colours. The comparison is the visual, not a computed delta.
The table underneath. Exact percentages per tier, so the claim can be audited rather than trusted.
Live scorecards carry named suppliers and their ratings, so the built pages are not published here. The structure above is the deliverable; the data belongs to the client.
The system underneath
A design system, because there was never going to be one scorecard
Every supplier in every category needs this page. That is not a design problem, it is a manufacturing problem, and it gets solved once.
Two parts of that system did more work than the rest. The rating colours, because five tiers used consistently everywhere means a reader learns the scale once and never re-reads a legend. And the Prep flow, because the join has to be reproducible: the same three tiers, the same missing-data rule, every refresh, or the ratings stop being comparable across time.
The palette also had to carry sector, since the same index gets applied to more than hospitality. A colour per sector, defined centrally, so a scorecard reads as itself before you have read a word of it.
What I take from it
The hardest decisions are editorial, not technical
Joining three datasets is a Tuesday. Deciding what a non-response means, which tier outranks which, and whether a facility’s promise counts as much as its meter reading, is the actual work, and no amount of tooling makes those calls for you.
Get them right and a buyer trusts a single grade. Get them wrong and you have built a very fast way to be confidently unfair to a supplier.
Have several measures that need to become one number?
This is the join, the missing-data rule, and the page that makes the result defensible. Tell me what has to be compared.
Start the conversation