A buyer who attends your webinar, asks two pricing questions, stays past the offer, and clicks the CTA link is the hottest lead you'll see all quarter. Your CRM scores them the same as someone who downloaded an ebook three months ago and opened two emails. That is the gap webinar intent scoring closes.
What is webinar intent scoring?
Webinar intent scoring is the practice of assigning each attendee a 0-100 score that reflects how close they are to a purchase, based on what they actually did during a live webinar, workshop, or demo. The score combines in-session behavior (watch-time, chat, Q&A, clicks, polls) and cross-event behavior (repeat attendance) into a single ranked value you can use to prioritize follow-up, segment outreach, and route hot leads to sales.
It's a sibling of lead scoring, not the same thing. Generic lead scoring (as HubSpot's guide lays it out) weights firmographic and lifecycle signals: company size, role, email engagement, page visits. Webinar intent scoring weights in-event behavior, which is qualitatively richer and time-bound to a specific window. This guide is the companion to our broader webinar analytics framework; that post covers which metrics predict pipeline at the program level, this one covers how to score each attendee inside an event.
Why webinar intent scoring is different from generic lead scoring
Three structural differences matter:
- Event-bound. Webinar signals are concentrated in a 60-minute window. Signal density is far higher than a quarter of email engagement. A single session can produce more usable behavioral data than the previous six months of marketing touches combined. ON24's 2024 Webinar Benchmarks Report puts average B2B engagement at 51 minutes per attendee, which is a behavioral signal of a kind no email open or page visit can match.
- Behavior-rich. Attendees type, ask, click, stay, leave, react. Each action is unsolicited and costly. They're not filling out a form because you asked. They're typing in chat because something landed.
- Time-decaying. The signal expires fast. Intent captured in a webinar at 2pm is most actionable at 3pm and significantly less actionable by Friday. Generic lead scoring doesn't price time decay because the underlying signals don't decay the same way.
These differences mean a generic CRM score and a webinar intent score should not be averaged or substituted. They're complementary. A high webinar score plus a high CRM score is a sales conversation today. A high webinar score and a low CRM score is a net new opportunity you didn't know you had.
The 6 signals inside a webinar intent score
The signals below are ranked roughly by predictive weight. Each is a direct or near-direct read of buyer behavior.
Signal 1: Watch-time depth past the offer
Did the attendee stay in the session beyond the moment you made the offer or revealed pricing?
This is the single strongest behavioral signal in a webinar. Most people who aren't going to buy leave when friction appears: the ask, the price, the CTA. The people who stay past that point have already decided you might be a fit and are sticking around to learn how, not whether.
This signal is roughly binary: stayed or didn't. The point allocation reflects that.
Signal 2: Question-intent ratio
How many of the attendee's chat or Q&A messages were buying-shaped questions versus general or educational ones?
A buying-shaped question is someone mentally trying on your solution: "How does this integrate with our CRM?", "Is there a plan for teams under 10?", "What's the implementation timeline?", "Do you support healthcare compliance?"
A general question is someone exploring the topic: "Can you say more about that?", "Where can I read more?", "What was that chart at minute 20?"
Both kinds of questions add some points. Buying-shaped questions add a lot more.
Signal 3: In-session link clicks
Did the attendee click any link you shared live during the session?
A live click is a costly action. The attendee had to leave the video to do it. They did anyway. That's active intent expressed in real time. Clicks on the offer link, pricing page, or case study links are weighted higher than general resource clicks.
Signal 4: Chat engagement density
How many chat messages did this attendee send relative to the session length?
Density beats raw count. A single substantive message at minute 47 is worth more than five "hellos" in the first two minutes. Score on substance and total participation, not message count alone.
Signal 5: Poll and reaction behavior
Did the attendee respond to polls? Use reactions? Especially polls tied to buying readiness, budget, timeline, team size.
Polls are an underused scoring surface because most hosts use them for engagement theater rather than intent capture. Zoom's native polls in webinars and the matching poll report give you per-attendee responses; ask a poll that segments your audience by purchase readiness and the responses become high-signal data instead of warm-up filler.
Signal 6: Repeat attendance across events
Has this person attended more than one of your webinars in the last 90 days?
Repeat attendance is the strongest cross-event signal you can capture. Someone who shows up twice has demonstrated active interest at a level that single-event signals can't match. Most platforms don't surface this. Most scoring systems don't account for it. It belongs in the score.
How to weight the signals: a 0-100 rubric
A defensible default allocation. Tune to your audience and offer over time.
| Signal | Max points | Why this weight |
|---|---|---|
| Watch-time depth past the offer | 25 | Strongest single binary signal |
| Question-intent ratio | 25 | Highest-leverage active signal; rewards substance |
| In-session link clicks | 15 | Costly real-time action |
| Chat engagement density | 15 | Sustained attention proxy |
| Poll and reaction behavior | 10 | Useful when polls are intent-shaped, weak when they're not |
| Repeat attendance across events | 10 | Strongest cross-event signal |
| Total | 100 |
You can adjust the weights to match your funnel. B2B SaaS programs tend to weight in-session clicks and question-intent higher. Coach and creator programs tend to weight watch-time past offer and repeat attendance higher because chat tends to be sparser. Don't add more than six signals. The score loses interpretability fast.
A worked example
Maria runs a 12-person SaaS team. She attends your 60-minute webinar on B2B activation playbooks.
What she does:
- Joins on time, stays for the full 60 minutes including the offer at minute 47
- Asks two questions in chat: "Does this work for self-serve products?" (general) and "Is there a plan that includes team training?" (buying-shaped)
- Clicks the pricing link you shared at minute 50
- Responds to your poll on "where is your team in the activation journey"
- Did not attend a previous event
Scoring:
- Watch-time depth past offer: 25/25 (stayed full session, well past the offer)
- Question-intent ratio: 18/25 (two questions, one buying-shaped)
- In-session link clicks: 15/15 (clicked the pricing link, the highest-value link)
- Chat engagement density: 8/15 (two messages in 60 minutes, moderate)
- Poll behavior: 7/10 (responded to a meaningful poll)
- Repeat attendance: 0/10 (first event)
Total: 73/100. Hot.
Maria is in the top tier. Her follow-up email shouldn't lead with the replay link. It should reference her question about team training, point at the relevant plan, and invite a 15-minute call.
Same event, different attendee:
Tom joins, watches for 22 minutes, leaves before the offer, types nothing, clicks nothing, did not attend a previous event.
Scoring: 0+0+0+0+0+0 = 0/100. Cold.
Tom's follow-up should be short, low-pressure, with a replay link and a single question.
Different scores, different emails. That's the point, and our webinar follow-up email templates cover the per-tier copy you'd actually send.
Manual vs. automated webinar intent scoring
You can do this by hand. For a few events, that's the right call: it forces you to learn what each signal looks like in your specific audience. The math falls apart at scale.
| Manual | Automated | |
|---|---|---|
| Watch-time past offer | Timestamp the offer, export attendance CSV, filter | Computed per attendee |
| Question-intent classification | Read every chat message, tag each as buying-shaped or not | Computed from chat text |
| In-session link clicks | UTM-tag every link, pull clicks from your click tool, match to attendees | Computed |
| Chat density | Count messages per attendee from chat export | Computed |
| Poll behavior | Pull poll responses from Zoom's webinar reports, weight by question | Computed |
| Repeat attendance | Compare attendee lists across events in a spreadsheet | Computed |
| Time to score per event | 3 to 6 hours for 100 to 200 attendees | Minutes |
| Time to act on score | Same evening or next morning if you're disciplined | Within an hour of the recording releasing |
Manual scoring breaks on question-intent ratio first. A 60-minute webinar with 200 attendees produces around 400 chat messages on average. Tagging each one as buying-shaped or not takes hours. Most programs that try manual scoring stop using it by the third event.
This is the gap Sponja closes. It reads every Zoom session the moment the recording releases, scores each attendee 0 to 100 on these six signals, and writes a personalized follow-up draft per intent segment. The output lands in your existing email tool (HubSpot, Kit, Mailchimp, ActiveCampaign) tagged by score band. You review and send.
What to do with the score
Once you have a 0-100 score per attendee, segment by band:
- Hot (70 to 100): sales-ready or close to it. Direct ask, references their specific behavior in the session. Skip the replay link. Offer a call or a specific next step. Route to sales if you have a team.
- Warm (40 to 69): active interest, not at the ask. Resource that addresses what they engaged with. Soft transition toward the offer. Two-touch follow-up over the next 5 days.
- Cold (0 to 39): present but not engaged. Short replay email with a single question. No pressure. They're a nurture lead, not a follow-up lead.
The segments should drive different emails. If every band gets the same email, you've collapsed the score back into a single broadcast and undone the work.
Common mistakes when building a webinar intent score
- Including the host's own activity in chat density. Strip the host and panelist messages before computing.
- Treating "yes" reactions as scoring signals. They're noise. Reactions tied to specific buying-readiness prompts can be scoring signals. Generic thumbs-ups are not.
- Letting the score age past 7 days. The signal decays fast. A 90-day-old score is closer to noise than to data.
- Averaging webinar scores with CRM scores. They measure different things. Carry both.
- Tuning the weights every event. Set the weights, run 5 to 10 events, then look at the data. Tuning each event chases noise.
If you run webinars and your CRM lead score is the only number you carry into follow-up, you're leaving the highest-intent signal in your funnel on the table. Webinar intent scoring closes the gap. Sponja computes it automatically for every Zoom session. Try it free on your next event at app.sponja.ai.
