A plain-language guide for event teams
Every event runs on a few tools that each keep their own version of the same people. This guide shows — with real fields and values — how one person becomes several disconnected records, what that quietly costs you, and the four-part idea that puts them back together. No code. If you've ever exported a list and fixed it by hand, you already know the problem; we're just going to name it and solve it.
Worked example: Foire de Paris We follow one man: Marc
Splio (consumers) and HubSpot (businesses) never talk to each other. Klipso feeds both — by hand.
The problem, in real data
Marc Dubois runs a regional food wholesaler. He grabbed a free visitor ticket, then spent two days getting bulk-pricing quotes from exhibitors. Here is what he literally looks like inside your tools after the show. Read down each column: same man, four times, nothing connecting them.
| Field | Klipso (registration) | Splio (newsletter) | Exhibitor A's CSV | Exhibitor B's CSV |
|---|---|---|---|---|
| Record ID | KLP-2027-44821 | splio_88213 | row 47, leads_hallC.xlsx | line 1120, FdP.csv |
| Name | Marc Dubois | Marc Dubois | M. Dubois | Dubois, Marc |
| marc.dubois@gmail.com | marc.dubois@gmail.com | — blank — | mdubois@grossiste-sud.fr | |
| Company | — (registered as individual) | — blank — | Grossiste Sud | GROSSISTE DU SUD |
| Type / role | Visitor (free) | Newsletter subscriber | Lead — "hot" | Prospect |
| Buying intent | — none captured | "premium?" (guessed) | Wants pallet pricing | Bulk buyer |
| Linked to the others? | No | No | No | No |
Why this costs you money
Marc isn't a data curiosity. Because his four records never connect, three concrete things happen, every one of them avoidable.
A wholesaler ready to buy by the pallet gets treated like a tourist — your most generic email lands on your best prospect.
Who pays: your sales pipelineTwo reps chase the same buyer, duplicating effort and sometimes undercutting each other — neither knows he's a repeat customer.
Who pays: your exhibitors' trustNext year's planning has no idea a serious buyer attended on a free ticket, so you can't invite him back as one. The intelligence evaporates.
Who pays: next year's revenueSee it happen
Pick someone, then step through the three stages. Watch one human at the door turn into scattered records — then watch them resolve into a single trustworthy one. The two dials at the top show the show-wide damage and recovery.
The 612,000 → 410,000 figures are illustrative, to show the shape of the effect. Swap in real Foire numbers and they flow straight through.
The fix · part 1 of 4
This is the whole game (its proper name is entity resolution): you compare records and decide, with a confidence level, whether they're the same person or company. Some signals are near-certain; others say "probably — have a human check." Nothing is deleted; you just record that records belong together. Click through Marc's three matches.
The fix · part 2 of 4
Now you build one clean version. The trick: you don't pick a winning record, you pick a winning value for each field, by a simple written rule. Press the button and watch Marc's single record assemble.
The fix · part 3 of 4
A clean record is worthless if any tool can overwrite it. So each field gets one owner — one path allowed to change it. Fire each action and watch what's allowed through and what bounces.
The fix · part 4 of 4
The obvious fix is to wire two tools to update each other. It feels right — but without one owner per field, each sync cycle quietly adds a duplicate and the two tools drift out of agreement. Step through what actually happens to Marc's record.
The shape of the fix
Here's the flyover, with Marc threaded through it. Each tool plugs into a small adapter that knows the rules: which fields that tool is the source of truth for, which just pass through, and which are blocked. In the middle sits a data spine — one clean backbone record per person — and the clean values are then pushed back out to the tools that don't own them, via their API, so everyone shows the right value without the sync loop. Every box is an ordinary part; the cleverness is the order, which is why a small team can build it.
One sentence to remember
Everything here is a concept, not a product — matching, one clean record, an owner per field, one-way edits. Together they form what we call a data spine: a single trustworthy backbone that runs down the middle of the tools you already have and keeps them honest. Building one is well-understood work — and it's what Visual Hive does end to end: auditing your current stack, designing the architecture, implementing it, training and onboarding your team, then hosting and supporting it for the long run. Your tools stay exactly where they are; the spine just runs between them.