Harloop
Back to blog

Your Lead Score Doesn't Know Anyone Watched the Demo

July 6, 2026 · chinmay


Your lead scoring model gives ten points for a pricing page visit and zero for watching the product demo twice in one week. That's backwards, and most RevOps teams have never questioned it because the video never made it into the formula in the first place.

I've sat in enough lead scoring reviews to know the pattern. Marketing ops opens the Salesforce or Pardot rules engine, and the point allocations are all page-based: pricing page visit, ten points; case study download, fifteen; demo request form, fifty. Video shows up, if it shows up at all, as a single flat rule: "watched a video, plus five." Not which video. Not how much of it. Not whether they watched it once or came back three times to rewatch the pricing section. That's the equivalent of scoring a codebase by whether it compiles and ignoring the test coverage.

The Signal Sitting Outside the Formula

Here's what actually correlates with intent, in what I've watched across mid-market B2B accounts: a lead who watches past 80 percent of a two-minute technical demo converts at a meaningfully higher rate than one who clicks play and bails at ten seconds. Completion percentage is a stronger predictor than the binary "did they watch." Rewatch is stronger still. Someone who comes back to replay the implementation section a second time is telling you something a page-visit rule never will.

Vidyard's own scoring guidance for video view data makes the same case from the other direction: they recommend assigning point values by watch percentage and letting engagement, not just presence, drive the score. The idea isn't new. What's missing at most companies is the plumbing. The video platform knows the watch percentage. The CRM owns the score. Nobody wired the two together, so the richest behavioral signal in the funnel sits in a video analytics dashboard nobody on the sales team opens.

Wiring the Score, Not Just Watching It

This is a plumbing problem before it's a strategy problem. A webhook from the video platform, on a "video viewed past threshold" event, needs to write a field update to the lead or contact record: watch percentage, rewatch count, which specific asset. Then the scoring formula, whether it lives in Salesforce's flow builder or Pardot's scoring rules, needs a rule keyed to that field instead of a flat "watched a video" checkbox. The formula change itself is small. The reason most teams never make it is that the watch data and the score live in two systems that were never introduced to each other.

Two edge cases kill most first attempts. Anonymous viewers, someone watching before they've filled out a form, have no contact record to attach the event to, so the webhook has nowhere to write until identification happens; you either buffer the event and reconcile it after form-fill or you accept the gap. And GDPR-region traffic needs the same consent gate your form tracking already respects. Bolt video engagement onto a compliant identification flow instead of routing around it.

This is exactly why harloop's attribution stack treats watch percentage and rewatch as first-class fields, not a footnote metric, and why the webhook is built to hand that data to whatever CRM owns the scoring formula. The system that shows the personalized clip is the same system that reports back what happened, so the score updates without someone manually exporting a CSV of video analytics into Salesforce once a quarter. The same closed-loop logic applies once video is wired into HubSpot's timeline: the mechanism changes by CRM, the principle doesn't.

I'd take this further than most RevOps leaders are comfortable with: if your scoring model can't distinguish a 90 percent completion from a 10 percent bounce, drop the video rule entirely rather than keep a vanity checkbox. A wrong signal in the formula is worse than no signal, because it tells sales to call someone who bailed in the first ten seconds with the same urgency as someone who rewatched the pricing slide three times. Personalization without a matching attribution layer is theater, and a lead score built on a checkbox instead of a percentage is the same failure wearing a different hat.

Pull your current scoring rules and find the video line item. If it says anything other than a watch percentage or a completion threshold, it's not measuring intent. It's measuring whether the play button got clicked, and that was never the question worth asking.

More posts