We asked one client, five different ways across discovery, what makes a lead worth a salesperson's time. We never got the same answer twice.
They weren't dodging. They're an industrial manufacturer selling through three regional teams, and the good deals in their history arrived from every direction: a large OEM one quarter, a two-person engineering shop the next. Their honest answer to "what does a good lead look like" was "we know it when we talk to them."
That's the week we stopped trying to build them a lead score for go-live. The same week, a software client launching in December asked for a full scoring model in phase one. We gave both the same advice. Launch with a gate and a clock. Score in month three.
I understand why marketing leaders want it on day one. A score is what makes a new CRM feel like a system. It gives sales an order to call in and gives marketing a number to report on. It's on every implementation checklist you'll find, and if you ran a score in your last portal, leaving it out of the new one feels like going backwards.
The mechanics are easy, too. HubSpot will let you build a score in an afternoon. That's the problem. The tool makes it painless to encode whatever you believe about your buyers before you've checked any of it against a single closed deal.
Here's what the manufacturer's marketing lead sent us when we asked for scoring inputs. Reason for contact. Customer type, with OEM weighted highest. Application fit, read from a free-text field. Industry, from enrichment. Job title, nice to have.
Walk through it. Reason for contact is a routing field: a sales request goes to sales, a support request goes to support, and no score changes that. Customer type: if someone picks OEM and asks to talk to sales, they're going to sales anyway. Application fit lives in a text box, and nobody can score a paragraph without custom code or an AI agent reading submissions from a form that has never been live. Industry only exists if the enrichment vendor returns a value. Title isn't a required field, so most records won't have one.
Strip out everything the score can't actually see and you're left with two dropdowns. Those two dropdowns already decide where the lead goes. The score would be decoration on top of the routing.
The portal had never taken an inbound lead. There was no data to score against, only opinions about what the data would say.
What we recommended instead, for both clients.
A qualification gate. Each of the manufacturer's regions wrote down, in plain language, what has to be true before a lead becomes sales qualified. Those three definitions were clear and agreed, and we could build every one of them. That's the handoff. It doesn't need a score.
A response clock. For the software client we split inbound into two queues: demo requests go to an SDR immediately, and everything else raises an alert if nobody has touched it within 24 hours. That's the whole phase one "scoring model", and it's the one piece that changes rep behaviour on day one.
Then let the portal run. Two months of real submissions tells you which fields arrive populated, which enrichment values actually show up, and which form answers sit on the deals that close. Build the matrix from fields you can prove exist. We've built scoring this way for other business units in the same group, and the rule we kept coming back to was short: no data, no scoring. The model has to be fed by specific fields you actually capture.
If leadership still insists on a score at launch, the fallback is to score only on what the form reliably captures (customer type, request type, a short industry list) and to say out loud that it's a placeholder until the data arrives.
Scoring in month three is less exciting than scoring at go-live. It's also the version that gets used, because every weight in it points at a deal somebody can name.
If you launched HubSpot with a score and it still decides who sales calls first, I'd like to hear how you built it. And if you launched with one that everyone quietly ignores, I'd like to hear that too.
This came out of a client project. The story behind it, and what to check in your own portal, goes to subscribers on Thursdays. Get Thursday's note.