How It Works: A Continuous AI Learning Loop

Most lead scoring is static. Someone sets a rule in a marketing platform once, gives a whitepaper download five points and a pricing page visit ten, and then nobody revisits it for a year. The model never finds out whether any of those points predicted revenue.

Bread & Butter works as a loop instead. Every visit feeds the scoring model, every conversion adds research to the profile, and every deal you mark as won teaches the system what a good visitor actually looks like on your site. The loop closes, and the scoring gets sharper on your traffic specifically.

In short: The journey is recorded from the very first visit, Focused Score ranks that visit against your own traffic rather than an outside standard, converting brings Profile Enrichment onto the profile, and marking the deal Won/Closed in Bread & Butter sets the Success Factor and recalibrates your Ideal Customer Profile, which in turn sharpens the scoring applied to the next visitor who lands.

Step one: the journey is recorded from the first visit

Tracking starts the moment someone lands. The journey captures what they looked at, in what order, how long they stayed, where they came from, which campaign brought them and how many times they have been back.

One tracking code covers as many web properties as you run, so a person moving between your main site, your blog, your documentation and a campaign landing page stays one profile with one journey rather than four disconnected sessions. At this point they are Stage 1, Visitor. Nobody is named yet and nothing is assumed about them. The record is simply built and kept, whether or not they ever convert.

Step two: Focused Score ranks the visit against your own traffic

Focused Score is live from the first visit. There is no warm-up period before scoring begins.

The important part is the frame of reference. Focused Score ranks a visit relative to your own traffic, not against some universal notion of a good visitor. Depth, return pattern, the specific pages that carry buying intent on your site, pacing and source all feed a score from 1 to 10. What counts as a strong session on a specialist advisory site and what counts as a strong session on a high-volume store are completely different, and the score reflects your mix.

Two supporting numbers sit alongside it. Your Focused Ratio is Focused visitors divided by all of your traffic, and it is compared against the Focused Index, the network benchmark. A healthy Focused Ratio runs from 10% to 20%. The first fifteen days calibrate your ratio against that Index, which is a calibration of the benchmark comparison, not a wait for scoring to switch on.

Worth knowing about the surfaces: the Focused Tab is ordered by most recent visit rather than by score, so it is a live view of who is around right now. The Nurture Action Queue is the surface that prioritizes, and it holds Engaged Leads with a Focused Score of 6 or higher.

Step three: conversion brings Profile Enrichment

The loop stays behavioral until the person acts. When they convert, through one of your forms, a subscription, a social login or a gated booking calendar, a verified email address arrives and Profile Enrichment runs.

Enrichment researches the person behind that address and attaches role, seniority, employer, company detail and career context to the profile. The research re-runs on a return visit, so a job change or a move to a new company does not leave you working from a stale record.

There is no two-tier enrichment any more. A profile is the user journey, recorded from the first visit, plus Profile Enrichment, researched when the email arrives. At this point the lead is Stage 3, Engaged Lead, and for the first time the scoring model has both halves of the picture: what this person did, and who they turned out to be.

Step four: Won/Closed sets the Success Factor

This is the step that makes the loop a loop, and it is the step teams most often skip.

When a deal closes, mark the lead Stage 8, Won/Closed. Won/Closed is the default Success Factor, which is the outcome the model treats as the definition of success. Everything upstream gets evaluated against it.

Mark it in Bread & Butter. Your CRM can stay the system of record for the commercial side of the deal, but the learning loop only sees the outcome when the stage is set in Bread & Butter, and a deal marked won only in the CRM teaches the model nothing. It is a few seconds of work per deal and it is the highest-leverage thing anyone on the team can do for scoring accuracy.

Step five: the Ideal Customer Profile recalibrates

Each Won/Closed lead hands the model a complete case study: the full journey from the first visit through every return, plus the enriched profile of the person who actually signed.

Across your wins, patterns emerge. Certain sources and campaigns show up repeatedly ahead of revenue. Certain page sequences turn out to be the ones buyers walk, while others are read by people who never buy. Certain roles, company sizes and industries close, and others do not. That cross-section of behavior and profile becomes your Ideal Customer Profile, recalibrated with every new win.

The recalibrated profile feeds straight back into step two. The next visitor who lands is scored against a model that has learned from your latest closed deal, so the ranking gets tighter on your traffic over time rather than drifting. Prospect AI uses the same recalibrated profile to find net-new people outside your traffic who match the shape of your best closed deals.

What the loop changes for your team

The practical effect is that prioritization stops being a matter of opinion. Reps open a queue built from what has actually closed for your business, rather than a list built from rules someone guessed at last year.

Enrichment arrives without anyone researching a name by hand, so the context a rep needs is already on the profile before the first call. Attribution becomes credible, because the sources that produce wins are visible as sources that produce wins, not just sources that produce sessions. And the model improves on its own as long as the team keeps marking deals Won/Closed, which means the scoring you get in month six is better than the scoring you got in month one without anyone tuning anything.