Know what every lead is worth
Craft scores every lead from the customer's own words: urgency, financing, decision-makers, hesitation. That joins consumer and property data, and the lead gets a dollar value while the call is still live. Managers see the reasons behind every score, so they trust the queue.
Book a Demo
Danielle Foster
44% close on similar opportunities · 6 mi away
+31%
Sales lift
50%+
Model inputs from the conversation itself
2.3×
Top-decile lead value vs. average
~1%
Predicted vs. realized revenue gap
Scored, matched, booked
Craft scores the lead during the first conversation, picks the rep most likely to win it, and books while the homeowner is on the line.
Who needs the new bath?
The homeowner with a 780 credit score? Or the one who tells your agent her father slipped getting into the tub last month, and asks for the soonest possible appointment?
Demographic models optimize for the first caller.Craft uses that data too, then listens to the call: more than half of the model's inputs are things the customer said.
What the model heard
Signals Craft hears on every call:
A custom model for your company
The model predicts three numbers for every lead: chance they demo, chance they buy, and project size. What the customer said on the call joins consumer and property data, and the product is a dollar figure, in real time. Every closed opportunity it learns from is yours, so two companies in the same city end up with different models.
Demos
78%
Chance they demo
Closes
54%
Chance they buy
Project size
$17,400
If they buy
Score
$7,330
Expected value
What the model saw
Assignments managers sign off on
Craft shows the evidence behind every score, right in the product: financing asked, soonest appointment wanted, spouse not on the call. Every assignment comes with its reasons, so managers can check the model's work in seconds. They keep final say with a drag, and the plan re-optimizes around them.
Marcus Rivera → The NaylorsWalk-in bath · Thu 9:00 AM$14,100Why this assignment
Assignment that knows your team
Craft builds each rep's day around predicted value: geography, capacity, and their win rate on this exact opportunity type. New reps ramp on middle-of-the-road leads until their numbers earn the big ones. When a higher-value lead appears, it reassigns and reschedules the rest.
Danielle Foster3 of 4 booked · 52 min · 24 miTyler Brooks to Danielle Foster
$8,120
Mesa · 9:00 AM
Previous customer
$6,480
Gilbert · 12:30 PM
Marcus Rivera to Danielle Foster
$4,910
Chandler · 3:30 PM
Scheduling that predicts the win
While the homeowner is still on the line, the model ranks every open slot by predicted close rate for this exact lead. Your agent offers the best one first.
Times to offer · ranked by predicted close
Danielle Foster46% closeDanielle Foster41% closeMarcus Rivera33% closeA model that retrains on every outcome
Every won opportunity, lost bid, and no-show flows back in, and the model shows its work: predicted value tracks realized revenue within about 1%, decile by decile, on appointments it has never seen. This month's scores know more than last month's.
Grouped by predicted value at scoring time · n = 21,412
FAQs
AI lead scoring ranks every incoming lead by how likely it is to close, so your team works the best opportunities first. Craft builds the score from three sources: what the homeowner actually said on the first call, property and consumer data, and the closed outcomes of similar leads in your CRM.
Demographic models rank a lead by how much the household resembles past buyers. Craft adds the strongest signal available: first-party call behavior. Financing questions, urgency, project scope, and who was on the phone are heard on the first call, joined with consumer and property data, and scored in real time. Closed outcomes from your CRM confirm which signals actually predict a won opportunity for your company.
Yes. Craft trains the model on your own calls, your market, and your closed opportunities, and it retrains as new outcomes arrive. Two companies selling the same product in the same city end up with different models, because their reps, lead sources, and customers are different.
Any home improvement vertical: baths, flooring, roofing, windows and doors, siding, kitchens. Craft trains a separate model for each company, on your own calls and closed opportunities, so the signals reflect what actually predicts a win for what you sell.
Because they can check them. Every score shows the signals behind it: financing questions, urgency, who was on the call. Every rep assignment shows its reasoning, and a manager can override any of them with a drag while the plan re-optimizes around the change.
Yes. Craft weighs geography, capacity, and each rep’s historical win rate on similar opportunities, then books the appointment with the rep most likely to win it. When a higher-value lead appears, Craft can reassign and reschedule the displaced appointment automatically.
Yes. Craft checks real availability while the homeowner is still on the line, prioritizes same-day and high-value slots, and books directly into your calendar. Confirmations, reminders, and rescheduling are handled automatically.
Craft syncs scores, assignments, and appointments to ServiceTitan, Salesforce, LeadPerfection, JobNimbus, MarketSharp, and 20+ other systems. Closed outcomes flow back from the CRM so the scoring model keeps learning from your actual results.
1,100,000+
conversations behind every score
See what your leads are worth
Bring last month's lead list. We'll score it in front of you and walk through the reasons behind every number. It takes about 30 minutes.