Two leads fill out the same demo form on the same day. One found you by searching "best session recording tool for agencies — pricing". The other arrived from "what is a heatmap". If your lead scoring treats them identically, your sales team is flipping a coin on who to call first — and statistically, they'll waste half their morning on the researcher while the buyer goes cold.
SEO intent signals — the search queries and organic landing paths that brought a lead to your site — are the earliest scoring information you will ever have about a lead, available before they've clicked a single internal link. This guide builds a complete, practical scoring model around them: a 100-point framework with exact point tables, three worked examples calculated line by line, thresholds, decay rules, and how to calibrate the whole thing against your real customers.
What are SEO intent signals?
Search intent is the purpose behind a query, and SEO practice groups it into four classes. Each class predicts a very different likelihood of buying:
| Intent class | The searcher wants to… | Example queries | Buying likelihood |
|---|---|---|---|
| Transactional | Act now — buy, sign up, start | "clicktics free trial", "buy heatmap software", "[product] discount" | Highest |
| Commercial investigation | Compare before acting | "best session recording tools", "hotjar vs clarity", "[product] pricing" | High |
| Navigational / branded | Reach a specific site | "clicktics login", "clicktics blog" | High if new, low if existing user |
| Informational | Learn something | "what is session replay", "how to read a heatmap" | Low (today) |
The classes aren't just SEO taxonomy — they map almost perfectly onto funnel stages. A transactional searcher is at the decision stage before your homepage even loads. An informational searcher is at awareness, no matter how enthusiastically they fill in your newsletter form. Scoring leads without this context means scoring blind.
One honest technical note before the model
Search engines stopped passing the exact keyword with each visitor years ago (the famous "not provided"), so you can't read a specific lead's query off their session directly. In practice you reconstruct intent from two proxies, and the model below uses both:
- The organic landing page. Pages rank for predictable query clusters — someone landing organically on your "pricing" or a "best X tools" comparison arrived via commercial queries; someone landing on "what is X" arrived via informational ones. Landing page is a reliable intent fingerprint.
- Search Console query data. Google Search Console reports which queries drive clicks to each page. Connect it (Clicktics has a built-in Search Console integration) and you know each landing page's actual query mix — turning the proxy into measured probability.
The 100-point model: four components
The model allocates 100 possible points across four groups, plus negative signals. The weights reflect a simple principle: intent gets you called, behavior gets you called first, fit gets you called by the right rep.
Component 1: Entry intent — up to 30 points
| Signal (organic landing page / query class) | Points |
|---|---|
| Landed on pricing, free-trial, or demo page from search (transactional) | 30 |
| Landed on a comparison / "best tools" / "alternatives" page (commercial) | 22 |
| Branded query, first-time visitor (navigational, new) | 18 |
| Landed on a solution/feature page (mixed commercial) | 14 |
| Landed on a how-to / tutorial post (late informational) | 8 |
| Landed on a "what is X" post (early informational) | 4 |
Component 2: On-site behavior — up to 35 points
Intent says why they came; behavior says what happened next. These points come straight from session data:
| Behavior signal | Points |
|---|---|
| Viewed pricing page (from anywhere) | 10 |
| Returned within 7 days | 8 |
| Engaged 3+ minutes or 4+ pages in a session | 6 |
| Started the signup/trial funnel (even if abandoned) | 6 |
| Opened live chat or replied to a chat greeting | 5 |
Component 3: Fit — up to 20 points
| Fit signal | Points |
|---|---|
| Work email domain (not gmail/outlook/yahoo) | 8 |
| Role/company field matches your ICP (e.g. agency, marketer) | 7 |
| Geography in your served markets | 5 |
Component 4: Momentum — up to 15 points
| Momentum signal | Points |
|---|---|
| Second visit arrived via a higher-intent query than the first (intent escalation) | 10 |
| Visited 2+ commercial pages across separate sessions | 5 |
Momentum is the most predictive — and most ignored — SEO signal of all: a lead who found you last week via "what is session replay" and came back today via "clicktics pricing" just told you their funnel stage changed. That escalation pattern is worth more than any single visit.
Negative signals — subtract points
| Signal | Points |
|---|---|
| Landed on careers, support, or documentation pages | −15 |
| Existing-customer branded queries ("login", "cancel", "refund") | −20 |
| Bounced under 10 seconds with no interaction | −10 |
| Inactivity decay: per 30 days with no visit | −10 |
Thresholds: what the totals mean
| Score | Grade | Action |
|---|---|---|
| 75 – 100 | Hot (SQL) | Sales outreach same day |
| 50 – 74 | Warm (MQL) | Fast follow-up sequence; sales-ready on next signal |
| 25 – 49 | Nurture | Email education track; watch for intent escalation |
| 0 – 24 | Monitor | No action; re-score on next visit |
Three worked examples, calculated line by line
Lead A: the comparison shopper
Priya lands from Google on your "best session recording tools" comparison post, reads it, clicks through to pricing, and starts — but abandons — the trial form using her agency email.
| Signal | Points |
|---|---|
| Organic landing on comparison page (commercial) | +22 |
| Viewed pricing page | +10 |
| Engaged 4+ pages | +6 |
| Started trial funnel (abandoned) | +6 |
| Work email domain | +8 |
| Role matches ICP (agency) | +7 |
| Total | 59 → Warm (MQL) |
The play: a same-week follow-up referencing the trial she abandoned. If she returns once more — any +8 return-visit or escalation signal pushes her past 65 — she's a same-day sales call.
Lead B: the researcher
Marco arrives from "what is session replay", reads one guide for two minutes, and subscribes to the newsletter with a Gmail address.
| Signal | Points |
|---|---|
| Organic landing on "what is X" post (informational) | +4 |
| No pricing view, no funnel start, under 3 min | +0 |
| Personal email domain | +0 |
| Total | 4 → Monitor |
The play: nothing — and that's the model working. Marco cost your sales team zero minutes. If he reappears next month via a comparison query, intent escalation (+10) plus the new entry score re-grades him automatically.
Lead C: the escalator
Dana first visited ten days ago via "how to reduce cart abandonment" (a how-to post), and returns today via the branded query "clicktics pricing", views pricing for three minutes, and opens chat to ask about the agency plan — from a company domain, in your market.
| Signal | Points |
|---|---|
| Latest entry: branded/transactional landing on pricing | +30 |
| Intent escalation across visits (informational → transactional) | +10 |
| Returned within 7 days | +8 |
| Viewed pricing + engaged 3 min | +10 +6 |
| Opened live chat | +5 |
| Work email + ICP role + geography | +8 +7 +5 |
| Cap at 100 | 89 → Hot (SQL) |
The play: answer that chat personally, now. Dana's ten-day journey from how-to reader to pricing-page chatter is the textbook SEO-intent conversion path — and the chat message is the close.
And one decay calculation
Suppose Priya (59 points) goes quiet. After 30 days: 59 − 10 = 49 → she slides from MQL to Nurture automatically. After 60 days: 39. If she returns in week nine via a pricing query, the new entry intent (+30) and return scoring rebuild her grade in one visit — the decay guarantees your "hot" list is never stale, and the re-scoring guarantees nobody deserving stays buried.
Calibrating the model against reality
The point values above are a sane starting grid, not physics. Calibrate them in three steps:
- Backtest. Score your last 40–50 actual customers as of the day before they converted. If their median pre-conversion score was, say, 63, your SQL threshold of 75 is too strict — you'd have called them late. Move thresholds to where real buyers actually scored.
- False-positive check. Score the last 50 leads sales marked "dead on arrival." If many scored 70+, find the inflating signal — most commonly branded-navigational entries from existing users — and add the corresponding negative rule.
- Watch the misses. Once a month, pull five high-scoring leads that never converted and watch their session recordings. If the 80-point leads all stalled at the same pricing-table confusion, your model was right about intent and your page lost the deal — a CRO fix, not a scoring fix. (This recordings-first diagnosis loop is the same one from our conversion playbook.)
Setting this up in Clicktics
Everything the model consumes is data Clicktics already collects from one ~24 KB script:
- Entry intent: every visitor's landing page and referrer are on their journey timeline, and the Search Console integration shows which Google queries actually drive clicks to each landing page — your intent-class mapping, measured instead of guessed.
- Behavior points: pricing views, return visits, session duration, funnel starts, and chat opens are tracked per visitor automatically; conversion events cover the funnel-start signals.
- The pipeline: Clicktics' lead management scores leads on behavior and moves them through New → MQL → SQL → Customer stages, so your thresholds map one-to-one onto pipeline stages your team already works. Chat is built in for the Dana moment, with notifications so the "answer it now" play actually happens now.
- Calibration: session recordings for the watch-the-misses review, and Discovery AI for questions like "which landing pages produced last quarter's customers?" — the backtest, without the spreadsheet.
Five mistakes that break intent-based scoring
- Scoring the form, not the journey. A demo request from an informational path and one from a transactional path are different leads. The form is one signal; the path is the context.
- Treating all branded search as hot. "Clicktics pricing" is a buying signal; "clicktics login" is a customer. Split branded queries by the page they land on.
- Ignoring decay. Without time decay, your MQL list becomes an archive. Intent expires — 30-day half-lives keep it honest.
- Over-weighting fit. A perfect-ICP lead from a "what is" query is a future customer, not a current one. Fit routes leads; intent times them.
- Never re-scoring. Intent escalation is the strongest signal in the model, and it only exists if every new visit triggers a re-score.
Frequently asked questions
What are SEO intent signals in lead scoring?
They're the evidence of what a lead was searching for when they found you: the intent class of their query (informational, commercial, transactional, navigational), inferred from their organic landing page and confirmed with Search Console query data. They reveal funnel stage before any on-site behavior happens.
How do I know which keyword brought a specific lead?
You usually can't see it per-visitor — search engines withhold it. Instead, use the landing page as the intent proxy and Search Console to see each page's real query mix. A lead landing organically on a comparison page arrived via commercial queries with high confidence.
How many points should each signal be worth?
Start with weights proportional to buying proximity — entry intent up to ~30, behavior ~35, fit ~20, momentum ~15 — then calibrate against your last 50 customers' pre-conversion scores. The exact numbers matter less than backtesting them; every business's grid drifts to its own shape within a quarter.
Does intent-based scoring work for ecommerce?
Yes, with product intent replacing query intent classes: category-page landings (browsing) vs "buy [product]" landings (transactional), plus cart starts as the funnel signal. The structure — intent + behavior + momentum with decay — transfers unchanged.
What's the difference between lead scoring and lead grading?
Scoring measures engagement and intent (will they buy now?); grading measures fit (should we want them?). The model here blends both, with intent and behavior carrying 80 of 100 points — because timing kills more deals than fit does.
Score the journey, not just the form
Your leads are telling you how ready they are before they ever identify themselves — through the searches that brought them and the pages those searches chose. Clicktics captures the whole story in one place: Search Console queries beside visitor journeys, behavior-scored lead management with pipeline stages, live chat with notifications, recordings for calibration, and Discovery AI — one ~24 KB script, free for up to 180 days, no credit card.
Start your free trial → and let your next hot lead introduce themselves by their search.
Tomás García
Tomás García writes for the Clicktics blog about session replay, analytics engineering, and building privacy-first products that agencies love. Reach the team at [email protected].