Visitor Intelligence · Capability
Session intelligence: what a buyer actually tried to do
Session intelligence turns raw website and app behavior into structured buyer context: which content sections held attention, what was clicked and whether it worked, how far forms progressed and where they stalled, what was searched and selected — plus a narrative interpretation of the whole visit.
Four questions a page-view report cannot answer
What mattered?
Semantic page sections are tracked by heading — how long each was actually visible, re-entries, and what remained on screen when the visitor left. Attention is measured against the real page, not assumed from a URL.
What did they try?
Every meaningful click is recorded with its outcome — navigated, opened a modal, downloaded, errored, or did nothing. Downloads resolve immediately so leaving the page never reads as “nothing happened.”
Where did they stall?
Form progression keeps the step, questions reached and completed, validation blocks, submit attempts, and the exact field where someone stopped — and validation-blocked is distinct from ordinary abandonment.
What can you trust?
Human behavior is separated from bots and scanners, events are buffered and retried with stable identities, and reduced-coverage conditions are marked instead of hidden.
From raw events to a session you can act on
- Capture. Meaningful screen, content, action, form, search, and selection signals are collected — including SPA route and same-URL screen changes, true scroll depth on the real scrollable surface, and deliberate intent like search terms, settled filters, pricing choices, and downloads.
- Normalize. Noisy interactions consolidate into final states: repeated clicks become attempt counts, form activity becomes a journey, and events resolve into session boundaries.
- Interpret. Each session gets engagement, intent, conversion, entry/exit, and prior-visit context — plus a coherent narrative explanation of the visit a rep can read before a call.
- Surface. High-intent, returning, identified, form-incomplete, converted, campaign-sourced, and bot cohorts become directly queryable — and routable into CRM, alerts, and GTM Agents.
What the tracker keeps — and deliberately avoids
| Keeps | Avoids | Marks |
|---|---|---|
| Search terms, semantic section names, form labels and progression, settled selections, action outcomes, copy source, download type | Arbitrary form answers in the journey model, clipboard contents, and high-risk fields such as passwords, card or bank details, and government identifiers | Bot and scanner traffic, and reduced-coverage conditions — so missing or automated behavior is never silently treated as buyer intent |
What this changes for the team
Sales
A usable explanation of what a prospect cared about, tried, and struggled with before reaching out — especially when the visitor is a known CRM contact.
Marketing
Content interest, campaign quality, form friction, search demand, and conversion behavior at the visitor level.
Growth / RevOps
Routing and workflow rules built from actual intent, identity, return behavior, outcomes, and capture quality — not raw page views.
Frequently asked questions
How is session intelligence different from web analytics?
Analytics counts URLs; session intelligence interprets behavior. Instead of reporting that a visitor opened five pages, it reports which content sections they actually read, what they clicked and whether the click worked, how far they progressed through a form and where they stopped, what they searched for, and what the visit adds up to as buyer intent.
Does it work in single-page apps?
Yes. Lumitra understands route, hash, and meaningful same-URL screen changes, so tabs, wizard steps, and dialogs are represented as real screens instead of producing fake page-view noise. Scroll depth is measured on the page's real scrollable surface, including inner overflow containers, rather than assuming the document is always the scroll root.
What does Lumitra record about forms?
The journey, not the answers. Lumitra keeps form titles, the step reached, questions reached and completed, validation errors, submit attempts, and the field where someone stopped — while form-journey analytics avoid storing arbitrary typed values, and high-risk fields such as passwords, card or bank details, and government identifiers are excluded from capture.
What is an action outcome?
The result of an interaction, recorded alongside the interaction itself. A click is classified by what happened next: navigation, a modal opening, loading, content change, a visible success or error, a download, a new tab, or no observable response. Repeated clicks on the same target become an attempt count instead of duplicate intent events.
How does this feed buyer intent?
Each session gets engagement, intent, conversion, entry/exit, and prior-visit context plus a narrative of the visit — and sessions from bots and scanners are separated first. High-intent, returning, identified, form-incomplete, converted, campaign, and bot cohorts are directly queryable, so intent is something you can route on, not a vibe.
Read a real session narrative
Book a demo and see a live visit interpreted — sections read, clicks resolved, form friction, and the story a rep gets before the call.