INTENT AS CONTROL LAYER
Intent-based website personalization: route the page around the decision
Two anonymous visitors can look identical in analytics and still need completely different answers. One is asking whether the product fits. Another is deciding whether switching is worth the risk.
Intent-based personalization treats the decision being made as the primary control layer.
- 01Landing route
- 02Observable signal
- 03Intent state
- 04Proof module
- 05Next action
Start with six intent states
Do not begin with fifty segments. Begin with the six decisions that repeatedly appear in a B2B buying journey.
- Category fit
- “Is this product even for a company like mine?” Show ICP, use cases, outcomes, category definition and basic proof.
- Use case
- “Can it solve this specific job?” Show workflow, relevant features, use-case proof and an implementation example.
- Alternatives
- “Why should I choose this instead of X?” Show fair comparison, switching or migration, architecture, trade-offs and when the competitor is a better fit.
- Pricing
- “What does this cost and is it worth it?” Show pricing model, package fit, ROI, usage limits and commercial next step.
- Implementation and risk
- “How hard is this to adopt?” Show setup, integrations, security, migration, ownership and timeline.
- Action readiness
- “I already understand the product. What can I do now?” Offer trial, demo, quote, configuration or technical evaluation.
Signals should be first-party and explainable
Useful signals include landing route, pages viewed, page sequence, comparison interactions, pricing interactions, technical documentation, declared use case and CTA behavior.
A good intent system should answer: “Why did we classify this session as implementation intent?”
Entered through /eli-vs-scrunch → opened agent-ready architecture → read MCP content → implementation intent confidence increased.
Build decision rules before complex models
IF the visitor enters on competitor comparison, views migration or implementation, and has not started a trial; THEN prioritize switching proof, implementation detail and a technical evaluation CTA. The decision system should remain debuggable.
What happens when confidence is low?
Low confidence is a valid state. Do not force personalization when the signal is weak. Show the canonical experience, show a neutral mix of modules, or allow the visitor to self-select the path.
Intent should change during the session
discovery → alternatives → implementation → pricing → action
The site should respond to the newest useful evidence. Track signal → intent state → experience → action → qualified outcome, not only “pricing-intent page CTR increased.”
FAQ
Questions to resolve before you adapt the journey.
What is buyer intent on a website?
It is the decision or information need a visitor is trying to resolve, such as pricing, alternatives, implementation or use-case fit.
Can intent be detected without cookies?
Some intent can be inferred from the current session, page path and voluntary interactions. Available signals depend on implementation and consent requirements.
How many intent states should I use?
Start with a small number of commercially meaningful states. Five to seven is easier to validate than dozens of vague segments.
What if the intent model is wrong?
Use confidence thresholds, canonical fallbacks and self-selection. The system should be reversible.
Is this the same as lead scoring?
No. Lead scoring estimates commercial value or readiness. Intent personalization decides what information or action should be shown now.
CONNECT YOUR WEBSITE
See how Eli routes by intent
Route explainable session context to the evidence and action that resolve the current decision.
See how Eli routes by intent