Automatic lead scoring a number instead of a gut call
Every new contact gets a number the moment it appears in the CRM: for industry, company size, job title, and for what that person did on your site and in your emails. A hot lead goes to a rep with an alert and a 60 minute deadline, a warm one into a sequence, a cold one into automated follow-up. The first call goes where the chance is biggest.
- 7 timesclose to how much more often a company that calls within the first hour qualifies a lead, against one that waits an hour longer. Scoring decides who gets that hour. Audit of 2,241 companies, Harvard Business Review
- 3 bandshot, warm, cold. Each has its own path: a call within the hour, an email sequence, or follow-up with nobody involved
- 2 to 6 weeksfrom workshop to a model in production, tested first against your own leads from the past few months
The problem
Why does a director with a budget wait behind a student?
An enquiry from the operations director of a manufacturing company with 80 staff lands in the same queue as a form filled in by a student gathering data for a thesis, and a competitor checking your price list. The rep works from the top, because the time of submission is the only hint on offer. The first minutes of the day go to a contact who will never buy. The director waits.
The more sources, the worse it gets. Website form, chat, webinar, trade fair list, referrals, an ad landing page. Nobody has time to judge each enquiry against all the others, so the call is made by eye, differently by every person, and there is no way to check it afterwards. The data that would settle it already sits in the CRM: industry, company size, pages viewed, emails opened. Nobody adds it up into one number.
The cost is countable. In an audit of 2,241 companies published in Harvard Business Review the average response time to an inbound web enquiry was 42 hours, and 23 percent of companies never replied at all. Companies that called within the hour qualified a lead close to 7 times more often than those that waited an hour longer. One hour is not enough for everybody. The question is who gets it, and scoring answers that before the rep opens the CRM.
- A hot lead waits, because three forms with no potential sit ahead of it in the queue
- Every rep judges leads differently and nobody knows who is right
- Behavioral data from your site and your emails sits unused in the CRM
- A competitor checking your price list gets a call before the director with a budget
Why now
What changed, so that scoring goes live in weeks rather than a quarter?
The data is already there. HubSpot, Pipedrive and Salesforce record which pages a contact viewed, which emails they opened and what they downloaded. Company size and industry come from enrichment against the email domain or from a public company register, job title from LinkedIn. Nothing has to be collected from scratch. It has to be weighted and added up.
Native lead scoring in HubSpot or Salesforce is enough when there are a handful of criteria and all the data sits in one CRM, and the rules for reading and writing those fields are published openly, for example in the HubSpot developer documentation. When part of the signal sits outside the CRM, because the webinar runs in another tool, chat in a third and the trade fair list in a spreadsheet, we move the model into n8n or Make. The workflow gathers the signals, adds up the total and writes it back into the CRM as a field. The rep still works in their own tool, the list is just sorted differently.
The third change is the AI agent. A model assigns a number, but it will not answer a sales manager asking which leads from this week score above 80 and have had no contact yet. BOND, our company AI agent, answers that from the CRM in 5 to 10 seconds, with no filter, no view and no report.
What changes
What does a number on every lead actually change?
Order by potential, not by the clock
Every new contact gets a score the moment it lands in the CRM, before a rep ever sees it. The list is sorted by the points total, not by the time the form came in.
Score calculated within seconds of the CRM record, for every source: form, chat, webinar, trade fair, referral
Two dimensions in one number
Firmographic fit: industry against your ICP, company size from enrichment or LinkedIn, job title, the tools the company runs. And behavior: pages viewed, the price list, case studies, clicks in emails, a return visit after days of silence.
Two groups of criteria, one total, three thresholds agreed with your reps in the workshop
A hot lead gets a call within the hour
A lead above the threshold fires an alert to Slack, Teams or SMS together with what pushed the score up. The rep calls knowing this person hit the price list twice and opened the proposal.
Alert carrying the reason for the score and a 60 minute contact deadline
Warm and cold leads stop eating a rep's day
A warm lead enters an email sequence, a cold one gets automated follow-up. A person steps in once the score crosses the threshold, not at every single form submission.
Three bands, three paths; only the top one is handled by hand
A score that ages
Negative points for a role with no say in the purchase, for a competitor domain and for silence, say 30 days with no action at all. A lead that stopped responding drops down the queue instead of hanging at the top of the list.
Negative criteria and decay after an agreed number of silent days, written into the model, not into the manager's head
Questions about the funnel in a plain sentence
How many hot leads sat for more than an hour with no contact. Which ICP segment has the highest average score. Who has the most leads above 80 today. The AI agent answers from the CRM.
An answer in 5 to 10 seconds, with no filters, no views and no CRM login
And in your company
What if it does not work here?
Four sentences we hear most often, and how we answer them.
Our reps will trust their own eye anyway, not a number.
That is the most common reason scoring dies after a month. So the weights come from how your reps recognize a good lead today, and we test the model against historical leads and compare it with their judgement before it goes live. The rep also sees what the lead scored points for, not just the total.
We have too few leads for this to make sense.
If a few enquiries a week arrive from one source, you do not need scoring and we will say so in the consultation. It earns its place when leads come in from several sources at once and a rep cannot keep judging each one against the rest. Until then it is better to sort out the CRM.
HubSpot has its own lead scoring, why add anything on top?
It does, so does Salesforce, and with a handful of criteria drawn from one CRM that is enough. That is what we will tell you. When the signals sit outside the CRM, because the webinar, the chat, the trade fair list and the company register live elsewhere, the model moves into n8n or Make and the result comes back to the CRM as a field you sort by.
We have no company data, only a first name and an email from the form.
The email domain gives us the company, enrichment against that domain or a public register gives size and industry, the LinkedIn job title gives the place in the decision. Behavioral signals are already collected by your CRM or your email tool, they only need weighting. Gaps in the data get scored too: a lead with no company gets fewer points, not zero.
- CRM
- HubSpot, Pipedrive, Salesforce, Zoho
- Lead sources
- Website form, chat, webinar, trade fair list, referrals, ad landing pages
- Company data
- Enrichment by email domain, public company registers, LinkedIn
- Behavioral signals
- GA4, email sequences in the CRM, content downloads, Google Sheets
- Alerts
- Slack, Microsoft Teams, SMS, email
- Automation layer
- n8n, Make, native CRM scoring, custom API integrations
Scoring criteria and where the data comes from
The criteria that most often make up the model. Weights and thresholds are set in the workshop, because what signals readiness to buy in a services business means something different in manufacturing or in e-commerce.
| Criterion | Where the data comes from | How it moves the score |
|---|---|---|
| Industry against the ICP | company register lookup, CRM field, email domain | plus on a match, zero when it is missing |
| Company size | headcount from enrichment, a public register or LinkedIn | plus inside the ICP range, for example 10 to 200 people |
| Job title of the contact | form, LinkedIn, email signature | the closer to the decision, the heavier the weight; a role with no say scores negative |
| Visit to the price list or case studies | GA4, CRM tracking | strong plus, because it signals a comparison of offers |
| Opens and clicks in a sequence | the email tool inside the CRM | plus rising with every click, not for an open on its own |
| Return visit after days of silence | GA4, CRM | plus, because the lead came back without an email from you |
| Competitor domain or no activity | domain list, date of the last action | minus; the score decays after an agreed number of days |
Weights and thresholds are not a template. A first rollout covers the criteria you already have data for, and the rest goes in once the measurement at 30 days shows a signal is missing.
How we work
Five steps from a gut call to a number on every lead
Every step ends with something you can judge before you decide on the next one.
Workshop
We talk to your reps about how they recognize a lead worth their time today, and we write those criteria down together with the exceptions. We check which data already sits in the CRM and which has to be pulled in from enrichment, LinkedIn or GA4.
Model design
Weights for the firmographic and the behavioral criteria, the hot, warm and cold thresholds, negative points and the decay rule for inactivity. Choice of tool: native CRM scoring, n8n or Make.
Test on historical leads
We run the model against leads from the past few months and compare the score with what actually happened: who bought, who vanished, who the reps rated differently. Where the model is wrong, we correct the weights before production.
Production and training
The score is written into the CRM, alerts on hot leads go to Slack, Teams or SMS. The team learns to read what a lead scored points for, and the sales manager changes weights and thresholds without raising a ticket with us.
Measurement at 30 and 90 days
We check whether hot leads really do buy more often than cold ones. If they do not, the model is weighted wrong and we fix it straight away, instead of waiting until the reps stop believing in it and go back to judging by eye.
Before and after
How a rep's day changes
Before
- Opens the CRM and calls from the top of the list, because the time of submission is the only hint
- The first half hour of the day goes to a student and to a competitor checking the price list
- Judges leads by eye, differently from the colleague at the next desk
- A lead who hit the price list three times looks the same as one who typed an email and vanished
After
- The list is sorted by score, and the first call goes to the lead with the best chance
- The hot lead lands on Slack with its reasons before the rep finishes their coffee
- Warm and cold leads are handled by a sequence, a person steps in above the threshold
- The score ages on its own, so the list does not silt up with dead contacts
The first scores show up on the day of the switch. Whether the model is right we check at 30 and 90 days, comparing the band a lead was in with what happened to it.
For the person signing this off
What a head of sales gets out of it
You stop adjudicating whether a rep was right to drop a lead, because the decision now carries a number you signed off yourself. The team's time goes to the contacts with the best chance, not to the ones that happened to arrive first.
The second thing counts in the conversation with the board and with marketing. You know which channel brings leads with the highest average score, and how many hot leads waited longer than an hour. That is an argument in the budget conversation that you do not have today.
- Qualification criteria written down, instead of a separate gut call in every rep
- The model tested against your historical leads before it goes into production
- Weights and thresholds changed on your side, without raising a ticket with us
- Measurement at 30 and 90 days: do hot leads buy more often than cold ones
What it costs
The workshop has a published price, the rollout is quoted after it
The process workshop costs EUR 3,500 net per week, and runs from one to seven weeks. For scoring it is usually one or two weeks, depending on how many sources the leads come from and how many reps we have to listen to. We quote the rollout after the workshop, once it is clear how many integrations are in play and what state the CRM data is in.
You pay the cost of running the model directly to the tool provider, that is n8n, Make or the enrichment service, or you pay nothing at all when it sits on your own infrastructure. No margin from us. Where an AI agent is part of the setup, a single query costs the provider roughly EUR 0.005 to 0.07.
Process workshop
EUR 3,500 net
per week, scoring usually takes 1 to 2 weeks
A self-contained piece of work. It ends with qualification criteria written down with their weights, and a plan you could execute without us.
- A map of your lead sources and of how the team judges them today
- A list of criteria with weights, thresholds and negative points
- A tool recommendation: native CRM scoring, n8n, Make or the API
- The metrics to measure at 30 days
- Most popular
Workshop and scoring rollout
Quoted after the workshop
scope follows from the number of lead sources and integrations
The option most companies choose. A working model 2 to 6 weeks after the workshop, tested against your own historical leads.
- Everything in the workshop
- The model in n8n, Make or natively in the CRM, wired into your lead sources
- Enrichment with company size, industry and job title data
- Alerts on hot leads to Slack, Teams or SMS
- A test on historical leads and a switch-over with monitoring
- Training for the team on reading the score and changing the weights
Scoring and an AI agent for sales
Quoted after the workshop
for teams that ask about the funnel a dozen times a day
Everything in the rollout plus BOND, which answers questions about leads in a plain sentence.
- Everything in the scoring rollout
- An AI agent wired into the CRM: which leads above 80 have had no contact, which segment scores highest
- Alerts on hot leads still uncontacted after an hour
- Access for the reps and for the manager, in English and in Polish
If the workshop shows that native scoring in your CRM is enough, or that there are too few leads for a model to mean anything, we say so and show you what to set up yourself. The matter ends at the workshop, and you keep a written set of qualification criteria that stands up without us.
Frequently asked questions
What is automatic lead scoring?
It is a workflow that rates every new contact the moment it appears in the CRM and gives it a points total for how well the company fits your ICP and for what the person did on your site and in your emails. The total sets the band: a hot lead goes to a rep with an alert, a warm one into a sequence, a cold one into automated follow-up. The rep sees a list sorted by potential, not by the order the enquiries arrived in.
How long does a lead scoring rollout take?
One or two weeks of workshop and 2 to 6 weeks of rollout, depending on the number of lead sources and the integrations that have to be wired up. A model with the basic firmographic and behavioral criteria, one CRM and two or three sources sits closer to the lower end of that range.
Which criteria go into the scoring model?
Two groups. Firmographic: industry against the ICP, company size from enrichment, a public register or LinkedIn, the job title of the contact, sometimes the tools the company runs. Behavioral: how many pages were viewed, a visit to the price list or the case studies, clicks in an email sequence, a content download, a return visit after days of silence. On top of that, negative points for a role with no say in the purchase, a competitor domain and a lack of activity.
Do we need n8n or Make if HubSpot already has lead scoring?
Not always. When there are a handful of criteria and all the data sits in one CRM, native scoring in HubSpot or Salesforce is enough and that is what we will tell you. An outside workflow earns its place when the signals are scattered: the webinar in one tool, chat in another, the trade fair list in a spreadsheet, company size in a register. Then n8n or Make collects them, adds up the total and writes it into the CRM.
Where do you get company size and industry data from?
The email domain identifies the company, enrichment against that domain or a public company register gives size and industry, LinkedIn gives the job title. When the form carries too little to match a company, we try the name and the domain, and if nothing matches the lead scores on behavior alone. Gaps in the data are information too and the model treats them as such, instead of pretending they are not there.
What happens to a lead with a low score?
It goes into automated follow-up or an email sequence and takes up none of a rep's time. A score is not a verdict: when the same person comes back to the site, opens the price list or clicks in an email, the points go up and the lead can move into a higher band. At that point the rep gets an alert exactly as they would for a new hot lead.
How do you check the model works rather than guesses?
Twice. Before production we run it against historical leads and compare the score with what actually happened to them and with how the reps rated them. After production we measure at 30 and 90 days whether hot leads buy more often than cold ones. If there is no difference, the model is weighted wrong and that is what we fix.
Will the reps know why a lead has the score it has?
Yes, and that is the condition for them using the model at all. The alert on a hot lead lists the criteria that pushed the rating up: two visits to the price list, a director-level job title, a company of 60 people in an industry inside the ICP. In the training we show how to read that, and the sales manager gets access to the weights so they can change them without us.
Let us find out who your team calls first
The free consultation takes 30 minutes. You tell us where your leads come from and how the reps decide who to call first today, even if that rule exists only in their heads. We show you which criteria can be turned into a number straight away, and which ones need data you are not collecting yet. You do not have to share your CRM at this stage.
We run at most two rollouts at a time, so that someone is genuinely sitting on each of them.
Book a free consultationPS. If a rep on your team lost an hour this week on a lead that will never buy, this is exactly the problem. A gut call is different in every person and cannot be checked afterwards. A number on a lead can be. That is where the workshop starts.