Lead scoring your ops team can explain in one sentence

A lead score should pass one test: can the person who owns the pipeline explain, in one sentence, why a given lead is a 72?

If the answer is “the tool decided,” the score will be ignored within a quarter. Sales will go back to sorting by gut feeling, and the number will survive only as a column nobody reads.

We build scoring in CRMs designed around how you sell. The rule is the same whether it is HubSpot, Zoho, or a custom build: every point has a reason a person can read.

What a score is for

A lead score is a sorting aid, not a verdict. It answers a narrow question: “Who should we call first this morning?”

It is not:

• A prediction of revenue
• A replacement for talking to the person
• A way to decide who deserves a reply (everyone who writes in gets one)
• Proof that marketing is working

Keep that scope and the score stays small and useful. Expand it and it becomes a political object, argued over in meetings.

Two dimensions, not one number

Most useful scoring separates two things that often get blended:

• Fit: does this look like a customer we serve well? (industry, size, location, role)
• Intent: did they do something that suggests they want to talk? (asked for a quote, booked a call, visited pricing twice)

A large company that downloaded one PDF is high on fit and low on intent. A student who emails us three times is high on intent and low on fit. Merging them hides the difference. Showing both lets the right person decide.

In practice we store two numeric fields and one plain-language summary field.

The rules we start with

For a typical Montreal B2B services SME, a first version fits on one page. Example criteria (yours will differ):

Fit

• Located in Quebec or Ontario: +10
• Role is owner, director, or operations lead: +10
• Company size inside your sweet spot: +10
• Free-mail address and no company name: −10

Intent

• Submitted the contact form with a described project: +25
• Booked a call: +30
• Visited the services or pricing page more than once in 14 days: +10
• Unsubscribed or replied “not now”: −20 and stop sequences

You will notice what is missing: dozens of micro-behaviours, a black-box “engagement index,” points for opening an email. Opens are a messy signal, and we do not score on them.

The reason field is the whole trick

Every score gets a reason text field populated by the same rules that set the number. For example:

Fit 30: Quebec, owner, 10–50 staff. Intent 55: described project in form; booked call. Last updated by rule set v3.

When a rep opens the record, they see why. When they disagree, they can say which line is wrong. That is a fixable conversation. “The score feels off” is not.

If your platform cannot write a reason string, we use a note or custom property. The cost is small; the trust gain is large.

Honest inputs only

Scoring is only as honest as its inputs. We avoid:

• Scraped or purchased enrichment presented as fact. If a field is inferred, we mark it as inferred
• Rented lists. We do not build scoring on top of data that should not be in the CRM in the first place; see CASL and first touches
• Invisible tracking. If a behaviour is scored, the privacy page should be able to say that behaviours are recorded, in the language of the page
• Protected or sensitive attributes. No scoring on things you would not want to explain to the lead

Quebec’s Law 25 asks you to know what you collect and why. A short scoring document, listing each input, supports that. We are not your counsel, but we can make the document easy for your counsel to review.

What we put in the CRM, concretely

A standard slice looks like this:

1. Fit score (number) and intent score (number)
2. Score reason (text, auto-written)
3. Rule version (so you can see which logic produced the number)
4. Lifecycle stage with clear entry criteria
5. Source captured at form submit, not guessed later
6. Language of the record, so the follow-up matches

None of this is exotic. It lives in the same system as your site, CRM, and search work, and it gets better when the form on the site already writes structured fields instead of an email.

The monthly review that keeps it honest

Scoring drifts. Your offer changes, your market changes, a campaign floods the CRM with a new kind of lead. So we add a review:

• Pull the last 90 days of closed deals, won and lost
• Look at the score each had at first touch
• Ask: did high scores close more often? Did good deals start with low scores?
• Change one rule, write down why, increment the rule version
• Re-check next month

If the score does not separate good from bad after a few cycles, that is also a result. Sometimes the right move is to simplify to a single “call today / call this week / nurture” flag. We would rather give you three honest buckets than a number that implies precision.

When not to score at all

Skip scoring if you get fewer than thirty qualified enquiries a month. At that volume, your team can read each one. A score adds a second thing to maintain and a false sense of order.

Also skip it if your CRM data is not clean yet. Scoring a messy database just ranks the mess. Fix duplicates, required fields, and source capture first; scoring becomes cheap afterwards.

What we will not promise

We will not promise a score that predicts revenue, or that your conversion rate will rise by a percentage. We will promise that every number in your CRM has a written reason, that you can change the rules without us, and that you own the configuration and the data.

If we ever hand you a model you cannot inspect, we have failed the brief.

Next step

Send us three won deals and three lost ones, with the first message each lead sent. We will show you what a scoring first draft would have said, in plain language. Write to hello@aervio.io or call +1 438-498-3913, weekdays 9–17 ET. Our CRM service page has the longer picture.

Questions

Do we need machine learning for lead scoring?

Rarely. Most small and mid-sized teams get better results from five to eight explicit rules than from a predictive model they cannot inspect. A model needs volume and clean history; many Montreal SMEs have neither yet.

How many fields should a lead score use?

Start with fit (who they are) and intent (what they did), each with a few visible criteria. If you cannot list every criterion on one page, the score is too complicated for your team to maintain.

How often should we review the score?

Monthly at first, then quarterly. Compare the score at first touch with the outcome: closed-won, closed-lost, or still open. Change one rule at a time and write down why.