Fit + intent + capacity scoring: a three-score model that turns into meetings

Most lead scoring fails for one plain reason: it produces a number, and then nobody acts on it fast enough. Fit and intent without a way to act are just a leaderboard.

The fix is a third score that almost nobody runs: capacity. Below is a model you can build this week, the formula that combines the three scores, and the rubrics behind each one.

What two-score models get right, and where they stall

A fit and intent model answers two good questions:

• Fit score: should we sell to them?
• Intent score: should we sell to them now?

That split is sound, and it is now standard. HubSpot's scoring tooling formalizes fit and engagement as separate criteria, then lets you combine them. (HubSpot lead scoring tool, Build lead scores)

The stall happens after the score. A two-score model ranks leads but says nothing about whether the top of that ranking will get worked while it is still hot. So the best accounts sit in a queue marked "later," and later rarely comes. That gap is what the third score closes:

• Capacity score: can this lead actually be worked right now, at the speed the signal demands?

Capacity matters because speed is a conversion multiplier, not a nicety. InsideSales has published for years that contact and qualification rates climb sharply when you respond in the first minutes rather than hours. (InsideSales lead follow-up)

Define the three scores

Fit score (0-100)

How closely the account and the person match your ICP. Fit inputs usually come from:

• Firmographics: size, industry, geography
• Technographics: the tools and data stack they run
• Role and seniority: the person who can buy, or the person who creates urgency

Intent score (0-100)

How strongly the account is signaling active demand. Intent inputs fall into three groups:

• First-party: website activity, email engagement, demo requests, and product usage if you run PLG
• Second-party: partner referrals, co-marketing lists, marketplace co-sell signals
• Third-party: review sites and intent networks

Intent has a real market definition: buyer behavior used to predict purchase likelihood. Gartner Digital Markets describes intent as behavior across its properties that identifies accounts "with the highest intent to purchase." (Gartner Digital Markets B2B Intent Data)

Capacity score (0-100)

Whether the lead will actually get a first touch inside your SLA. Capacity inputs include:

• Who or what is on the hook to work it
• Whether the segment and territory are covered
• Whether the response SLA can be met today, or the lead will rot
• Queue health: backlog size and how old it is

For a human team this is rep workload and PTO. For Chronic, capacity is simpler and more honest: an autonomous operator that finds, writes, sends, and follows up has no "I'll get to it tomorrow." The constraint is sending capacity and deliverability headroom, not someone's calendar. Either way, the score answers the same question: will this lead get worked in time?

The formula

You want one number for routing and priority. Use this:

Priority score (0-100) = 0.45 x fit + 0.45 x intent + 0.10 x capacity

Fit and intent carry the outcome. Capacity is the tie-breaker that keeps the hottest lead from dying in the queue.

Shift the weights to match the motion:

• Inbound-heavy: 0.35 fit / 0.55 intent / 0.10 capacity
• Outbound-heavy: 0.55 fit / 0.35 intent / 0.10 capacity

You can build this in a spreadsheet, a formula field, or directly inside an operator that already holds the scores. Chronic scores fit and intent on every lead it discovers, then prioritizes with capacity-aware sequencing so high intent does not sit idle. That is the difference between a smart-looking score and a booked meeting. (Chronic lead scoring)

Build the model, step by step

Step 1: lock the ICP inputs

Scoring is only as good as the ICP behind it, and the ICP has to show up in data, not in a vibe. Start with three inputs at most:

1. Company size band (employees or revenue)
2. Industry (the two or three verticals you actually win)
3. A triggering stack signal (one technographic that matters)

If your ICP fields are not clean, fix that first. Chronic's ICP builder exists for this exact step. (ICP builder)

Step 2: build the fit rubric (0-100)

Score fit in three buckets.

Firmographics (0-50)

• Employee band matches ICP: +20
• Industry matches ICP: +15
• Geography matches selling coverage: +5
• Funding or growth profile matches motion: +5
• Negative fit (student, agency, consultant, competitor): -25 to -50

Technographics (0-25)

• Runs a required system (example: Salesforce if you sell a Salesforce app): +15
• Runs a complementary tool: +5
• Runs an anti-fit stack (locked into a direct competitor on a long contract): -10 to -25

Role and seniority (0-25)

• Buyer role (VP Sales, RevOps, demand-gen lead): +15
• Seniority (director and up): +5
• Functional adjacency (an SDR manager when you need sales leadership): +3
• Generic title match only, no function clarity: +0

Guardrails: cap fit at 100, and when you apply negative fit, apply it hard. Weak disqualification just wastes cycles.

These fields only stay populated at scale if enrichment is consistent. Chronic runs enrichment as a default step in the flow, not a separate tool you bolt on. (Lead enrichment)

Step 3: build the intent rubric (0-100)

Most teams already have enough intent signals. They just do not score them. Score on two dimensions: how strong the action is, and how recent.

First-party intent (0-70)

Buying actions

• Demo request or contact sales: +40
• Pricing page visit, twice in 7 days: +20
• Integration docs visit: +15
• Case study view in their vertical: +10

Evaluation actions

• Live webinar attendance: +10
• Product tour completed: +10
• Reply to outbound with a real question: +15

Awareness actions

• Blog view: +2
• Careers page: +1

Apply recency multipliers: x1.3 in the last 3 days, x1.1 in the last 7, x0.7 once it is older than 14.

Product intent for PLG (0-70, in place of some web intent)

If you run PLG, product usage is intent. That is what a product-qualified lead is: someone who experienced value through usage and behavior. (TechTarget PQL definition)

• Hit the activation ("aha") event: +30
• Invited two or more teammates: +15
• Connected an integration: +15
• Active on 3 distinct days in 7: +10

Third-party intent (0-30)

Optional and useful, not magic.

• G2 buyer-intent surges and category activity: +10 to +25 by intensity and stage. (G2 Buyer Intent documentation)
• Gartner Digital Markets intent activity in your category: +10 to +25. (Gartner Digital Markets B2B Intent Data)

Rule: third-party intent is additive, never authoritative. It does not outweigh a first-party buying action.

Step 4: add the capacity score (the piece most models skip)

A meeting needs the right lead, at the right time, worked fast enough. Capacity scoring turns "fast enough" into a rule instead of a hope.

If a human team works the leads

Score at the rep or queue level:

• Rep has under 15 open tasks and under 6 meetings today: +30
• Rep is on the inbound rotation: +25
• Lead is in the correct territory and segment: +15
• SLA can be met within 2 hours: +20
• Rep is on PTO or at meeting cap: -50
• Queue backlog over 50 unworked leads: -30

No dynamic rep context? Score at the queue level: backlog under 24 hours old +30, 24 to 72 hours +10, over 72 hours -20.

If an autonomous operator works the leads

Capacity becomes about the system's headroom, which is the cleaner version of the problem:

• Sending capacity available on a warmed mailbox in the right segment: +30
• Lead falls inside today's outreach window for the prospect's timezone: +20
• Deliverability is healthy on the matched domain: +15
• Daily send volume already at the safe cap: -40

This is not academic. Even Salesforce ops practitioners frame scoring as useless without routing and action behind it. (Salesforce lead management best practices) The point of capacity is that it is the only score that protects speed to lead, and speed to lead is what InsideSales data keeps tying to connect rates. (InsideSales lead follow-up)

Weighting by motion: SMB vs mid-market

SMB (volume, speed, lower ACV)

Reward intent over fit. SMB buyers move fast, so the model should too.

• Fit 35% / intent 55% / capacity 10%
• 75+: contact within 5 minutes
• 60-74: enroll in sequence, follow up within 24 hours
• Under 60: low-touch nurture

Mid-market (higher ACV, more stakeholders)

Fit matters more, because wrong fit burns cycles across more people.

• Fit 50% / intent 40% / capacity 10%
• 80+: first touch within a 2-hour SLA
• 65-79: standard sequence, 24-hour SLA
• Under 65: nurture or outbound later

Tie-break rules

When two leads score the same, decide by rule, not by who feels like picking the easy one. In order:

1. Intent recency: activity in the last 72 hours wins
2. Buying action type: a demo request beats a pricing visit
3. Persona authority: the economic buyer beats an influencer
4. Expansion potential: an existing-customer expansion beats a new logo, if your comp plan rewards it
5. Capacity reality: if nothing can work it today, route it to the autonomous first-touch queue rather than letting it wait

That last one matters because "we'll get to it tomorrow" is fiction more often than not.

The minimum viable stack

You do not need a mystery AI product to start. You need:

• A fit score field (0-100)
• An intent score field (0-100)
• A capacity score field (0-100)
• A priority-score formula field
• A priority-band field (A, B, C, D)
• A routing rule on band, segment, and capacity

The trap is needing five stitched-together tools to keep those fields current. Enrichment in one place, scoring in another, sequences in a third, routing in a fourth, and the score is stale before anyone acts on it. Chronic runs the whole loop in one flow: discovery, enrichment, fit and intent scoring, capacity-aware prioritization, and the outreach itself. (Sales pipeline, AI email writer)

Score bands and the action map

The number does nothing on its own. The action behind the band is what books meetings.

Band A: 80-100, contact now

1. Assign to whoever (or whatever) has the highest capacity score
2. Open a first-touch task due within 15 minutes
3. Trigger the call-plus-email sequence immediately
4. If no capacity is free, route to the autonomous first-touch sequence and escalate when capacity opens up

Band B: 65-79, contact today

• Assign within territory, start the sequence within 2 hours, follow up the same business day

Band C: 50-64, qualify low-touch

• Email-only sequence, light personalization, re-score on any new intent event

Band D: under 50, stop spending time

• Nurture, exclude from the active queue, keep for retargeting if relevant

To make this run on its own, map each band to a signal-led cadence so the outreach changes based on what the buyer actually did. Chronic publishes that exact approach. (Signal-led sales cadence, Signal library)

The weekly tuning loop

A scoring model is never done. It is either tuned or dead. Run the same 30 minutes every week.

1. Pull the scoreboard. For the last 7 days, by band: meetings booked, reply rate, connect rate, time to first touch, no-show rate.
2. Find two failure modes. Band A not converting means your intent weights or your speed are off. Band C converting well means you under-scored a signal.
3. Change only three things. For example: raise the pricing-page weight from 10 to 20, add negative fit for agencies, decay intent older than 14 days, or tighten the capacity penalty at the send cap.
4. Recalculate and compare. Aim for stability. If a week's edits flip 40% of leads between bands, the model is too sensitive.
5. Hold the SLA. Track speed to first touch like revenue. When you cannot meet it, capacity has to override the score. This is where most teams fail and then blame the scoring.

The deeper shift is from asking your data questions to having the work done for you. (From asking your data to doing the work)

Common pitfalls

1. One combined score only. You lose explainability, and reps stop trusting it.
2. Too many inputs. A model that needs 47 fields is fragile, not smart.
3. No decay. Old intent becomes a zombie score that clogs the queue.
4. No negative scoring. Bad fit needs real penalties.
5. No routing and no SLA. Then your hottest lead becomes last week's missed meeting.

Where Chronic fits

Apollo exports lists. HubSpot and Salesforce can score, but you still stitch the enrichment, scoring logic, sequences, and routing together yourself, which is how "lead scoring" turns into one more tab.

Chronic is an autonomous revenue operator: you give it a goal, and it runs discovery, enrichment, fit and intent scoring, capacity-aware prioritization, and the outreach through warmed mailboxes, surfacing approvals only for the decisions that matter. The three-score model in this guide is not a spreadsheet you maintain on the side; it is how the operator decides what to work next. (Lead scoring, Lead enrichment, AI email writer, Outbound workflow blueprint)

FAQ

What is a fit and intent scoring model?

It is a prioritization system that assigns a fit score (how well an account matches your ICP) and an intent score (how strongly it is signaling demand), then often combines them into one priority score for routing and outreach. Adding a capacity score makes the combined number actionable rather than just a ranking.

Why not just buy an AI lead-scoring product?

Because you still have to define your ICP, decide which signals matter, and enforce the SLA behind the score. A black-box vendor score is one reps do not trust and ops cannot tune. Start with rules you can explain, then let an operator run them.

Which intent signals work best without paying for third-party data?

First-party signals usually win: demo requests, repeat pricing-page visits, integration-docs views, and product-usage events. They are direct evidence of evaluation behavior, not inferred interest.

How do I score capacity if I cannot see workload?

Start at the queue level: backlog age, unworked lead count, and hours of coverage. Route Band A to the smallest backlog first. It is crude, but it stops the worst failure, which is hot leads sitting untouched. If an autonomous operator works the leads, score capacity on sending headroom and deliverability instead.

How often should I tune the weights?

Weekly, with at most three small edits. Models decay because your market and motion change, not because the math is hard.

What is the fastest way to implement this without engineering?

Use a spreadsheet rubric to generate fit and intent scores, import them into your fields, and build simple routing by band. To run it fully on autopilot, use a system that handles ICP, enrichment, scoring, and sequencing in one flow. Chronic does that out of the box. (Chronic lead scoring)

Build it this week

1. Define fit inputs (three at most) and score them to 100.
2. Define intent inputs from signals you already have and score them to 100.
3. Add a capacity score so Band A never waits for "when I have time."
4. Run the formula, route by band, and hold the SLA.
5. Tune weekly on meetings booked, not on clicks.