Lead scoring means giving contacts points so you know whom to contact first. Points come from two places: fit, meaning how closely a contact matches your ideal customer (industry, size, region), and behavior, meaning what they do (downloaded your free resource, clicked a link, replied to an email, asked for a meeting). Anything that speaks against a deal subtracts points or rules the contact out. Above a threshold you set in advance, you get in touch personally. For a small business, a spreadsheet and five to eight rules are enough.
In my business, Olaf, my AI employee for finding clients, pre-sorts candidates against fixed criteria: every entry gets a reason, evidence and an A, B or C rating. How he finds them is in Finding New Clients as a Freelancer; every post on the topic is collected on the lead magnets overview.
What lead scoring measures: fit and behavior
A good model keeps three kinds of points apart.
High fit with little activity is a good customer who isn't shopping yet. Lots of activity with poor fit likes clicking but won't buy. HubSpot's tool splits them the same way: "engagement scores" for actions, "fit scores" for properties such as company size (as of November 2026).
A simple lead scoring model in six steps
- Derive your ideal customer from real customers. What do your five best jobs have in common? Without that picture, a score only measures activity.
- Pick three to five fit criteria. Only what you know for sure or can check publicly.
- Pick three to five behavior signals. A reply counts more than a click, a meeting request most of all. Count clicks per person only if you are allowed to (see privacy below).
- Define exclusions. A clear no gets so many negative points that the contact can never reach the threshold.
- Set a threshold and what happens next. Above it you reach out personally; below it, the newsletter simply continues.
- Review after three months. Did the customers come from above the threshold? If not, change the points.
Keep it small: with thirty rules, nobody knows later why a contact has 47 points. What to send contacts below the threshold is covered in Lead Nurturing Emails: Examples by Stage.
Lead scoring example: a table to copy
An example, not a case from my own work, and every number is made up: say you set up digital bookkeeping for small trade businesses and collect contacts with a checklist as your free resource.
Threshold in this example: 50 points, then you get in touch within two days. Two contacts:
- Contact 1: a carpentry shop with eight people in your region, downloaded the checklist and replied to your second email. 20 + 15 + 10 + 5 + 25 = 75 points. You call.
- Contact 2: an advertising agency, downloaded the checklist and clicked your packages three times. −100 + 5 + 30 = −65 points. No call; the newsletter continues.
Contact 2 clicked more. Without fit, you would have called the agency first. How to build the free resource that triggers the first points is in How to Create a Lead Magnet (8 Real Examples). Ideas for one are collected in Lead Magnet Ideas and How to Test Them.
When lead scoring is worth it, and when it isn't
Lead scoring solves an ordering problem. If you don't have one, you don't need it.
I run my business as a solo freelancer, and on my website the bottleneck check uses a single rule instead of a score: if you say more people than just you work in your business, the results page and the email include a link to book a call. That is how my privacy policy describes it in section 7.9. How the check is built as a quiz is in Quiz Funnels: Building One with AI.
Lead scoring tools for small business
* Affiliate link: if you sign up with Brevo or buy a plan through this link, I earn a commission. Your price stays the same.
Check that the feature is in your plan. Tools for finding contacts also advertise fit ratings; what you take on legally with them is in Lead Scraping Tools Compared.
How an AI employee pre-sorts leads
Olaf rates businesses before anyone has spoken to them. I give him a location, an industry and the traits of my ideal customer; he searches Google Maps and public company pages, checks every candidate against the same criteria and returns only the matches as a rated, documented list, each lead with source, reason and rating.
His page shows the numbers from my own runs: 198 candidates screened in eight assignments, and as a rule of thumb 3 to 5 real leads per 20 candidates. He discards false alarms, such as an old-tech site with a modern design. Olaf contacts no one, and he doesn't decide which target audience makes commercial sense. Why research and outreach must stay separate is in Automating Cold Outreach: Limits for AI; where B2B candidates come from otherwise is in Finding B2B Leads: LinkedIn and Alternatives.
The email that follows is prepared by Sonja: she checks name, address, demo link and price against the source and drops the draft into my inbox; nothing goes out until I say so explicitly. In one follow-up round she deliberately left a recipient out because the first email had promised personal consideration. No score would have caught that. The split: the score sorts, a person decides who gets which message.
Lead scoring and GDPR: profiling
This is not legal advice. For contacts in the EU, four parts of the GDPR matter.
Lead scoring is usually profiling. Article 4(4) defines profiling as automated processing of personal data to evaluate personal aspects of a person, explicitly including "personal preferences, interests, reliability, behaviour". If a natural person stands behind the contact and software calculates their points, you are in this territory.
Article 22 covers decisions based solely on automated processing. People have the right not to be subject to such a decision if it "produces legal effects concerning him or her or similarly significantly affects him or her". The profiling guidelines of the Article 29 Working Party, endorsed by the European Data Protection Board in 2018, say that in many typical cases targeted advertising based on profiling will not have a similarly significant effect, but name exceptions such as tracking people across websites, devices and services. Human oversight there has to be meaningful, "rather than just a token gesture". In practice: treat the score as a sorting aid and look at the contact before deciding anything that matters to them.
Transparency under Articles 13 and 14. For automated decisions within Article 22, Article 13(2)(f) and Article 14(2)(g) also require "meaningful information about the logic involved". If you count clicks per person, it belongs in your privacy policy and your consent text; mine says in section 7.5 that signing up includes tracking opens and clicks per subscriber, and in 7.9 what the bottleneck check stores.
Objection to direct marketing. Under Article 21(2), anyone can object at any time to processing for direct marketing, "which includes profiling to the extent that it is related to such direct marketing". After that, the data is no longer processed for that purpose. In your model, an objection is not a negative score; it ends the marketing.
Store only the criteria that actually earn points. What an AI should see of your customer data at all is covered in AI and Privacy: What the AI Gets to See.
Frequently asked questions
What is lead scoring in simple terms?
Points for fitting your ideal customer and for behavior such as replies, deductions for clear exclusions. Above a threshold, you reach out personally.
What are good lead scoring criteria for a small business?
Industry, size and region for fit; free resource, pricing click, reply and meeting request for behavior. The strongest signals are the ones where the person writes to you.
How many leads do I need before lead scoring makes sense?
No fixed number. Rule of thumb: once you can't read new contacts yourself on the day they arrive.
Is lead scoring allowed under GDPR?
It is usually profiling and needs a legal basis, transparency and a right to object to direct marketing. Decisions based solely on automated processing with significant effects fall under the stricter rules of Article 22.
Where to go from here
Write down your five best jobs and what they had in common. That gives you your fit criteria; add three signals and a threshold. If you want to talk through the workflow in your business that depends on you most, book a strategy session. Olaf is available as a ready-made template in Premium in my community AI Employee Framework, with a scoring matrix you adapt to your own criteria. My weekly newsletter comes with any template on /en/vorlagen. Which next step right after sign-up can trigger the first signal is covered in Thank You Page Examples and What to Include.
