Lead Scoring Models: Types, Criteria and a Worked Example

This guide covers the models in use, the criteria worth scoring, a calculation you can copy, and how to decide what happens when a score crosses the line. If you run a sales team of two to ten people, it is written for you.

TL;DR

• A lead scoring model assigns points to a lead’s characteristics and behaviour, then adds them into one score that ranks who deserves attention first.
• Most small teams do best with a rule-based model that combines fit (who they are) and intent (what they do), plus negative scoring.
• A predictive model trained on your own sales history needs sufficient relevant, clean outcome data. Requirements vary by platform; some tools use lead-conversion data or a pooled model. Start with weights you can explain.
• A score only helps if it triggers an action: a call, a booking, a nurture sequence or a human handoff.
• Review the model with sales every quarter, using booked meetings and won deals as the test.

Table of Contents

• What a lead scoring model actually does
• The main types of lead scoring models
• Lead scoring criteria: what to score
• How lead score is calculated: a worked example
• Setting thresholds that mean something
• Turning a score into action
• The human handoff
• Limitations and integrations to check
• Setup effort: a realistic plan
• Total cost: what to count
• A buying checklist for lead scoring software
• Where Naya fits
• Our view on lead scoring
• FAQ
• Sources

What a lead scoring model actually does

A lead scoring model is a set of rules that turns what you know about a lead into a number. Crunchbase describes it as assigning values to leads based on their characteristics, behaviours and interactions, so the leads most likely to convert can be prioritised. The point is not the number itself. It is the decision the number makes for you, which is who gets a call in the next five minutes and who goes into a longer sequence.

For a small team this matters more than for a large one. You cannot afford to spend a morning chasing a student who downloaded a guide while a ready-to-buy managing director waits for a reply. A model makes the call consistent, so the answer does not depend on who happened to check the inbox.

A good model answers three questions:

1. Is this the kind of business or person we sell to? (fit)
2. Are they showing signs they want to buy now? (intent and engagement)
3. Is anything telling us to stop or slow down? (negative signals)

The main types of lead scoring models

The names vary between guides, but the models fall into a handful of families. Turtl lists five: firmographic, demographic, behavioural, engagement and predictive. Here is how they work and when each earns its place.

Another common split is between explicit and implicit data. Zapier explains that explicit data is information supplied by the lead, such as job title, company size and location, while implicit data comes from actions such as website visits, form submissions and email engagement. Turtl also includes publicly known facts in explicit scoring. Fit scoring can use explicit data. Intent scoring can use implicit data. A solid model needs both.

Which model should a small team choose?

Start with a manual, rule-based model that adds fit and intent. It is transparent: any salesperson can see why a lead scored 74. A predictive model trained on your own sales history needs sufficient relevant, clean outcome data. Requirements vary by platform; some tools use lead-conversion data or a pooled model. A model built from your own best customers is a practical starting point.

Lead scoring criteria: what to score

Criteria, factors and attributes all mean the same thing here: the signals you assign points to. Choose signals you can capture automatically and that you have seen separate buyers from non-buyers in your own pipeline.

Fit criteria (who they are)

• Industry: the sectors where you have won most often.
• Company size: employee count or revenue band that matches your capacity to serve.
• Region: the areas you actually cover.
• Job title and seniority: the person who can sign, or who influences the decision.
• Need or use case: for example, a missed-call problem, a quote backlog or slow follow-up.

Intent and engagement criteria (what they do)

• Asked for a quote, demo or callback.
• Visited the pricing or booking page more than once.
• Answered a qualification call and gave a timeline.
• Replied to an email or text message.
• Completed a form with a budget or volume figure.

Negative criteria (reasons to subtract)

• Unsubscribed or asked not to be contacted.
• A role or sector you do not serve.
• A personal email address where you only sell to businesses.
• Long inactivity, so scores decay over time.
• Existing suppliers or rivals researching your offer.

Negative scoring is the part most small teams skip, and it is the cheapest way to keep your call list clean. If a lead asks to stop, the score should drop sharply and the contact should leave every sequence.

How lead score is calculated: a worked example

One simple approach is a weighted sum. Turtl describes it as scoring each attribute from 0 to 10, multiplying by a weight and adding the results. Its fit formula is: industry points × 3, plus revenue points × 3, plus employee points × 2, plus region points × 2. With weights of 3, 3, 2 and 2, the maximum is 100.

Here is an illustrative example (the numbers are invented to show the method, not benchmarks). A UK building services firm with 40 staff enquires:

Now add intent, scored separately so a strong fit with no interest does not look the same as a strong fit that has asked for a quote:

Keeping fit and intent as two scores is easier to manage than one blended figure. A lead with high fit and low intent goes to nurture. A lead with high intent and low fit gets a quick, polite check. High on both gets a call now.

Setting thresholds that mean something

A threshold is the score at which a lead changes status, for example from nurture to sales-ready. Set it too low and salespeople drown in poor leads. Set it too high and good buyers wait.

A practical way to set it:

1. Take your last 50 to 100 leads and note which became meetings and which became customers.
2. Score them with your draft model.
3. Find the score above which most of the meetings and wins sit.
4. Set the threshold just under that point, then check how many leads per week it would send to sales.
5. Adjust until the volume matches what your team can handle.

The agreement matters as much as the arithmetic. Twilio’s guidance stresses that sales and marketing should agree on what makes a lead qualified and set the criteria and threshold together, communicating any later change. In a small business that conversation can take 30 minutes, and it prevents weeks of arguing about lead quality.

Turning a score into action

A score that sits in a field and does nothing is decoration. Decide in advance what each band triggers, subject to the applicable purpose, audience and channel permissions and suppression checks. Opted-out contacts must be excluded from outreach:

Speed is where small teams lose most. If a hot lead arrives at 7pm on a Friday, the model has done its job only if something responds. That is why scoring works best when it is connected to the response, rather than being a report you read on Monday. For more on what happens after a lead is scored, see our lead qualification process framework and our guide to AI lead qualification.

The same logic applies to nurture. Cool leads are not dead leads, but continued marketing contact must respect the applicable purpose, audience and channel permissions and suppression status. A lead score or existing record is not permission to keep contacting someone; opted-out contacts must stay excluded from outreach. This guide to lead nurturing covers how scores can steer which prospects get which follow-up.

The human handoff

Scoring decides who gets attention. It should also decide who gets a person. A sensible handoff rule has three parts:

• Trigger: a lead crosses the threshold, asks for a person, or mentions a large order, a deadline or a complaint.
• Context: the salesperson receives the score, what drove it, the conversation so far and the lead’s stated need, so nobody asks the same questions twice.
• Time limit: a named person owns the lead and a target for first contact, such as 15 minutes in working hours.

Write these rules down before you automate anything. When the rules are explicit, you can let software handle the routine questions and keep people for the conversations that close deals.

Limitations and integrations to check

Every model has limits worth knowing before you build.

• Data quality sets the ceiling. Scores based on missing job titles or wrong company sizes mislead. Enrichment can help fill missing company and contact details; coverage depends on the available sources, and accuracy should be checked.
• Engagement can be noisy. One person opening an email ten times is not ten buying signals. Cap points per action.
• Models age. Your market, offer and channels change, so weights that worked last year may not now.
• Scoring lives in your CRM. If scores do not sync to the system your team works in, they will be ignored. Check which CRMs your tool connects to, and how lead records, conversation history and bookings flow back.

Setup effort: a realistic plan

You can have a working first version within a couple of weeks if you keep it small:

1. Week 1: list your best 20 customers and find what they share. Draft 5 to 8 fit criteria and 5 to 8 intent signals. Agree the threshold with sales.
2. Week 2: connect your data sources (website forms, calls, email, CRM), turn on enrichment and build the action rules for each band.
3. Weeks 3 to 6: run it, read the scores against actual outcomes, and adjust weights.

Avoid the temptation to add 40 criteria. A model with 10 signals you trust beats one with 40 you cannot explain.

Total cost: what to count

Do not price lead scoring as a single licence line. The real cost has four parts:

• Software subscription: the platform that scores, responds and follows up.
• Data: enrichment and prospecting data, if charged separately.
• Setup time: your team’s hours defining criteria and connecting systems.
• Ongoing review: a monthly or quarterly check of the model.

Compare this against the cost of a missed lead. If your average deal is worth several thousand pounds and a handful of enquiries a month go unanswered, the arithmetic usually favours automated response. Use your own deal values, not an industry average.

A buying checklist for lead scoring software

When you compare tools, ask each one the same questions:

• Can I define fit and intent scores separately, and edit the weights myself?
• Does it support negative scoring and decay for inactivity?
• Can it enrich leads automatically so the fit score is complete?
• Does it score across calls, email, SMS and chat, not only web forms?
• Does it connect to my CRM, and does the score and its reasons appear on the lead record?
• What happens when a lead crosses the threshold: does something act, or only a flag appear?
• How does handoff to a person work, and what context travels with it?
• How is it billed, and what grows the bill as volume rises?
• How long until a first version is live, and who does the work?

Where Naya fits

Naya is a sales operating system for established businesses with small teams. It brings lead prospecting and enrichment, inbound call handling, outbound AI calling, lead qualification, appointment booking and follow-up into one connected system, with CRM integrations for HubSpot and Salesforce so lead records, conversation history, bookings and next actions stay in the tools you already use. Teams can run email, SMS and voice campaigns, use chat and shareable agent links, manage their own AI agents with voice cloning, and tailor the conversation to their own qualification criteria.

In practice that means the questions your model scores, such as timeline, volume or budget, can be asked in the first conversation rather than weeks later. See how lead qualification checks work on every call, or read about AI sales prospecting and lead enrichment.

Naya’s published plans start from £578 per month for Standard. Growth starts from £798 per month billed annually, or £997 per month on a month-to-month basis. Custom is quoted individually. You can compare them on the Naya site and book a strategy call to map your scoring criteria to a working setup.

If your main gap is missed enquiries, the AI lead follow-up capability is the place to start. If calls are the bottleneck, look at AI calling and lead qualification.

Our view on lead scoring

Most small teams do not have a scoring problem. They have a response problem. A perfect model that scores a lead at 90 on Monday for an enquiry that arrived on Friday has already lost.

We build Naya around the idea that scoring, qualifying, booking and following up belong in one connected system built around the business’s own sales process. When the first conversation captures the answers your model needs, the score is based on what the buyer said, not a guess from a page view.

Our advice is to keep the model short, make every point explainable and let routine activity run around the clock, so your people spend their time on valuable conversations, customer relationships and closing. Measure it by qualified enquiries, booked meetings, recovered quotes and the capacity your team gains.

Ken

FAQ

What is the best lead scoring model for a small business?

A rule-based model that scores fit and intent separately, with a few negative signals, is the best starting point. It is simple to explain and you can build it from your own best customers. A predictive model trained on your own sales history needs sufficient relevant, clean outcome data. Requirements vary by platform; some tools use lead-conversion data or a pooled model.

How is a lead score calculated?

You give each attribute points, multiply them by a weight that reflects its importance, and add the results. In the worked example above, each fit attribute is scored from 0 to 10. Intent actions then add or subtract points on top.

What criteria should a lead scoring model include?

Include fit criteria such as industry, company size, region and job title, and intent criteria such as quote requests, pricing page visits and replies. Add negative criteria such as unsubscribes, wrong sector and long inactivity. Keep the list to the signals you can capture automatically.

How often should I update my lead scoring model?

Review it at least quarterly. Compare scores against booked meetings and won deals, adjust any weights that are not predicting outcomes, and tell everyone on the sales and marketing side when criteria change.

Sources

• Lead scoring models: types, revenue brackets and formulas (Turtl)
• What is lead scoring? Lead scoring models and best practices (Zapier)
• What is Lead Scoring? A Beginner’s Guide (Crunchbase)
• The Complete Guide to Lead Scoring Models (Twilio)

Recommended

• The Lead Qualification Process: A Step-by-Step Framework
• Lead Qualification Checklist: What to Check on Every Call
• AI Lead Qualification: How It Works and How to Set It Up
• Lead Enrichment: How It Works and How to Choose Tools