Home·Modules·Occupancy & ICP
Occupancy & ICP · in validation

A hundred pending leads.
Which one do you serve?

Operators with more demand than capacity are making the most consequential decision in the business on instinct: which customer to take next. Occupancy scores every pending lead against the book you already have, and predicts what each is actually worth.

“We have over a hundred pending leads and no way to know which are worth serving first.”

Founder · asset-heavy manufacturer
app.allometry · Occupancy Pending leads · ranked by predicted LTV 117 open · scored against your existing book SERVE NEXT: 3 LEAD LOOKS LIKE PREDICTED LTV Midland Cold Storage enterprise · 14 sites · inbound your top 3 accounts $840K Rivera Freightmid-market, expanding$310K Bellwood Supplysteady, low expansion$145K Coastal Tool Hirechurn-shaped · caution$38K WHY MIDLAND RANKS FIRST Site count and mix match your three most profitable accounts · low cost-to-serve scored against your book, not a generic firmographic model
Pending leads ranked by predicted lifetime value — scored against your existing customers.
01 · How it works

Four steps. The decision stays yours.

01
Learn your book
Your existing customers become the training set — what they order, at what margin, how long they stay, how they expand.
02
Score the pipeline
Every pending lead compared against that book: which existing accounts does this one resemble, and how did those perform?
03
Predict value
A lifetime-value estimate per lead, with the cost-to-serve and expansion likelihood that drive it shown separately.
04
Rank capacity
Serve-next ordering, so limited capacity goes to the demand that compounds rather than the demand that shouted loudest.
02 · What it does

Built from what operators actually described.

Scored against your book, not a generic model

Firmographic scoring tells you a company is big. This tells you it looks like the three accounts that have been most profitable for you — which is a different and more useful claim.

Predicted lifetime value, not a grade

A letter grade cannot be argued with. A dollar figure with cost-to-serve and expansion likelihood behind it can be, and should be.

Per-address economics

For operators deploying to many customer sites, value is predicted per address — which is also how the subscription scales with the estate rather than a seat count.

Honest about status

In validation with a design partner, scored against a real book of roughly three hundred customers. We would rather prove this on live data than ship a model we cannot defend.

03 · The swarm

One agent per question. Per address.

Scoring an address against your own book tells you what a customer like this has been worth to you. It does not tell you what is happening around that address — who else is already serving it, what it costs to get a truck there, whether the area is growing. Those are six separate lookups, and no human does them for 100 pending leads. A swarm does.

ONE PENDING LEAD — 1 OF 100+ 1290 RUE NOTRE-DAME who else serves it 3 competitors cost to serve +18% freight demand signal 2 permits filed your comparables 7 like it, 5 profitable density nearby 4 of yours < 12km local market wages +6% YoY RESOLVES TO PREDICTED LIFETIME VALUE $214,000 over 5 yrs RANK 3 OF 118 PENDING WHY — FROM THE SIX LOOKUPS Density: 4 of yours within 12km+ Comparables: 5 of 7 profitable+ Permits: 2 filed within 800m+ Freight: 18% above your median Competitors: 3 already serving Serve beforethe 41 ranked below it Cost to produce this~6 agent credits
Six lookups no salesperson runs for a hundred leads — and the reasoning is shown, so you can disagree with it.

Why it has to be a swarm

These are six unrelated retrievals against six unrelated sources. Done in sequence for 118 leads it is a week of somebody’s life; done in parallel it is minutes and a handful of credits. This is the clearest case in the product where agents are not a garnish — the work is simply not done otherwise.

Grounded in your book, not a generic model

“Three competitors nearby” is trivia. “Three competitors nearby, and the last four addresses like that returned 60% of your median” is a decision. The local signals are only worth retrieving because there is a book to interpret them against.

It answers an open question

Competitive density was the data source we lost when Market Intelligence was retired, and it is what has kept Occupancy in validation. Assembling it per-address on demand is the alternative to buying a dataset — cheaper, current, and scoped to the addresses you actually care about.

Untested, and we will say so until it is not. The swarm is a design, not a shipped feature. The open question is whether six retrieved signals plus your own comparables produce a ranking an operator will actually act on, or something that merely looks thorough. That is answerable in a week against a real book, and until it has been, this page will keep saying so.
04 · The outcome

What it changes

When capacity is the constraint, which customer you take next is the highest-leverage decision in the business — and it is usually made from a spreadsheet of inbound, ordered by arrival date.

Pending leadsranked by value
Trained onyour existing book
Outputpredicted LTV, not a grade
Statusin validation
Agentic execution, human authority. The agent does the work and attaches a number; you keep the decision. Every consequential write is gated on your approval, and no recommendation that breaches a floor you set is ever shown as an option — it escalates instead. How the substrate works →
05 · What it costs

Prices like every module. $2,000 a month.

Billed annually — $24,000 a year — and included in Closed Loop. Currently in validation with a design partner — if you want to be part of that cohort, tell us.

06 · Alongside

The other modules.

Start with a number, not a demo.

A free Margin Scan reads your data and tells you what your pricing is leaking.