Lead Scoring for Small Business: A Simple Model You Can Build This Week

Every small business with a steady flow of enquiries eventually hits the same wall. There are more leads than there is time to chase them, so someone picks which to call first. Usually that choice is made on instinct, on whoever emailed most recently, or on whoever sounded friendliest. Sometimes that works. Often it means the buyer who was ready last Tuesday is still waiting for a reply while the owner is on the phone with someone who was never going to buy.

Lead scoring is the habit of writing down, in advance, how you decide who deserves attention first. It sounds like something for companies with a marketing department and a data team. It is not. A useful model fits on one page, uses information you already collect, and can be built in an afternoon. What takes longer, and matters more, is checking it against reality and agreeing with everyone involved on what the numbers mean.

This guide walks through a deliberately simple version: two axes, a short list of rules, a few thresholds that trigger a real action, and a way to check the whole thing against deals you have already won and lost. The point values below are illustrations to adapt to your own business. They are not benchmarks, and nobody can hand you the right numbers without seeing your customers.

Key takeaways

Why score leads at all

The honest case for scoring is not sophistication. It is consistency and speed. Three problems tend to show up in a small sales operation.

• Attention is the scarce resource. If you get thirty enquiries a week and can meaningfully follow up ten, the order matters. Reaching a ready buyer within the hour and a casual browser within the week is a better use of the same hours than doing it the other way round.
• Gut feel does not transfer. The owner may be good at sensing who is serious. A new hire, or the owner on a busy day, is not. Writing the criteria down turns private judgement into something a team can share and improve.
• Good leads go cold quietly. Nothing alerts you when a strong lead has gone four days without a reply. A model with a clear “act now” line makes the waiting visible.

What scoring cannot do is create demand, rescue a poor offer, or replace talking to people. It is a sorting tool. A sorted list of poor leads is still a list of poor leads, and a score is only ever an estimate of likelihood, never a prediction you should treat as certain.

Fit and engagement: two axes, not one number

Most beginner models make a single running total: add points for this, add points for that. The trouble is that two very different kinds of information get mashed together, and the total hides which one is driving it.

Consider two leads who both score 60. One is a company of exactly the right size in your target industry that has barely touched your website. The other is a student who has opened every email, visited the pricing page four times and requested a demo for a school project. The totals match; the right responses do not.

Separating the two axes avoids that.

You can then read the two together. High fit and high engagement is your priority list. High fit and low engagement is a nurture list: right people, wrong moment. Low fit and high engagement deserves a quick look, since enthusiasm sometimes reveals a use you had not considered, but usually it is a polite, time-limited reply. Low on both can safely wait.

You do not need software to hold two numbers instead of one. Two columns in a spreadsheet are enough to begin with, and a CRM that stores a score per contact can carry whichever you decide matters most.

Building a simple points model

Start from your customers, not from a template. Write down the five or six facts that, in your experience, separate the people who buy from the people who do not. If you sell to restaurants, that might be whether they have more than one site. If you sell consulting, it might be whether the person who enquired can sign off a budget. Then write down the four or five behaviours that, in your experience, show real intent.

Assign each a small number of points. Keep the scale coarse, since the difference between 7 and 8 points is imaginary. Multiples of five are plenty.

Notice the shape rather than the numbers. A direct request for a quote carries more weight than anything passive. Reading a page carries very little. Facts that would make a sale impossible are penalised heavily. Anything marked “only include this if” needs evidence before it earns a place.

Rules for writing the rules

• Every rule needs a reason you can say in a sentence. If you cannot explain why a signal matters, delete it.
• Prefer actions that take effort. Asking for a quote costs the person something. Opening an email costs nothing, and mail software sometimes opens messages automatically, which inflates the count.
• Cap repeating behaviours. Ten pricing-page visits should not outweigh one request for a meeting. Give each behaviour a maximum.
• Score only what you reliably capture. A rule that depends on a field that is empty for half your contacts quietly penalises everyone with a gap.

Negative scoring and decay

Models that only add points drift upward until everybody looks promising. Two mechanisms keep a score honest.

Negative points for disqualifying facts

Some information should push a lead down, and hard. Examples: the person is a competitor, a student, a job applicant, or a supplier pitching you. They are looking for work or information, not a purchase. Or they sit outside the geography or size you can serve. These deserve subtraction large enough to take the lead out of the priority list, not a small nudge that a few page views can undo.

Be careful with negatives based on guesses about people. Penalising a lead for a job title you merely assume means little, or for an email domain that is common among legitimate buyers, creates blind spots you will not notice, because you never follow up the leads you scored away.

Decay for stale interest

Engagement is perishable. A lead who asked for a quote six months ago and went silent is not a hot lead because of that request; the request just happened a long time ago. If your engagement points never expire, your top list will fill with people whose enthusiasm has long since moved on.

A simple rule is enough: engagement points older than a set period (thirty or sixty days is a common starting point; choose by your sales cycle) stop counting, or lose a fixed amount each week. A short-cycle business such as a home service might let interest fade in two weeks. A business with a long procurement process might hold it for a quarter. Fit does not decay, because who someone is does not expire, though it should be refreshed when facts change, for example when a contact changes company.

Thresholds that trigger a human action

A score matters only when it crosses a line that changes what someone does. Define two or three bands and attach a specific action, an owner and a time limit to each. The example below is again an illustration.

Two details make this work. First, the “act now” band should be small. If a third of your leads land there, the line is too low and the label stops meaning anything. Second, always record why a lead was disqualified. That record is the raw material for the calibration step later, and it is the only protection against quietly discarding people you should have kept.

Resist the urge to automate the human part away. A prompt that tells someone to call within an hour is useful. An automatic email that claims to be personal, sent because a number crossed a line, usually is not.

Agreeing what “ready” means

Marketing and sales often describe the same lead differently. The person generating enquiries wants credit for volume; the person following up wants fewer time-wasters. In a small business these may be the same person wearing two hats, or two people who rarely compare notes. Either way, the threshold for “ready for a conversation” (the thing larger companies call a marketing-qualified lead) needs an explicit agreement.

A short conversation covers most of it.

• Definition. Write one sentence describing a ready lead in plain words, and test the score against it. If the sentence says “asked for a quote or described a project with a deadline” and the score says a download counts, one of them is wrong.
• Response commitment. Whoever follows up agrees to act on ready leads within the stated time. In return, whoever generates leads agrees not to lower the bar to hit a volume target.
• Feedback loop. The follow-up person marks leads that turned out to be poor with a reason. The numbers get adjusted from those reasons, not from arguments.
• A named owner for the model. Someone is responsible for changing it, and changes are written down with a date.

If you are a one-person business, do the same exercise with your future self. The version of you that is tired on Friday afternoon will otherwise quietly redefine ready.

Calibrating against deals you have already won and lost

This is the step most guides skip, and it is the one that stops a scoring model from being a confident-looking guess. You already have the answer key: the deals that closed and the deals that did not.

The look-back exercise

1. Pull 20 to 30 recent deals. Include both won and lost ones, roughly balanced if you can. Include deals that stalled, not just ones with a clear outcome.
2. For each, go back to the information you had at the first meaningful contact, not what you know now. Hindsight will otherwise feed you signals nobody could have seen at the time.
3. Apply your draft model to each as if it were a new lead, and write down the score.
4. Sort by score. Check how many wins sit in your top band and how many losses sit there too.
5. Read the misses individually. Why did a low-scoring lead close? Why did a high scorer fall through?

You are not looking for a perfect sort. You are looking for the model to be clearly better than random order, and for the misses to teach you something. A rule that never appears among your wins should go. A signal present in nearly every win and absent from most losses should carry more weight. A recurring reason for losses, such as the buyer having no budget, is a candidate for a new rule.

Small data: do not over-fit

With 25 deals, patterns are fragile. If four of your six biggest wins happened to come through one referral partner, that does not mean referrals deserve a huge weight. It may be coincidence, or a single relationship that will not repeat.

• Prefer coarse rules. “Right industry” is robust. “Right industry and between 11 and 14 employees and contacted on a Tuesday” is a description of the past, not a pattern.
• Be sceptical of surprising results. If the data says something your experience contradicts, find a plausible mechanism before you act on it.
• Change one thing at a time. If you rewrite five rules at once you will not know which change helped.
• Wait for more evidence before fine-tuning. Until you have a few dozen outcomes under the new model, treat tuning as experiment, not as truth.
• Do not borrow someone else’s numbers. Another business’s weights describe their customers and their sales process, and the same label can hide very different behaviour.

The small-data position is not an excuse to skip scoring. A rough model built on your best judgement and sanity-checked against your own history is still better than no stated model. Just hold it loosely and expect to revise it.

Common ways scoring goes wrong

Too many rules

Every new rule feels like added accuracy. In practice, past a dozen or so, nobody can explain why a given lead scored 73, so nobody trusts it, so nobody uses it. Rules also interact in ways you did not plan, producing results you cannot trace. When in doubt, remove.

Scoring activity that means nothing

Page views, email opens and social follows are easy to measure, which is exactly why they get over-weighted. Many are noise: automated link scanners click links, mail apps preload images, and a visitor reading your blog may be a researcher, a competitor or a bored commuter. If a behaviour does not show up more often in your wins than your losses, weight it near zero.

Never revisiting the model

Your offer, prices, market and channels change. A model that was right a year ago can quietly sort leads wrongly today. Put a review on the calendar, quarterly for most small businesses, and repeat a cut-down version of the look-back exercise each time.

Hiding the reason for the score

A bare number invites distrust. If the person following up cannot see that a lead is 80 because they requested a demo and fit the target profile, they will override it with their own instinct, and the model never gets a fair test. Whatever tool you use, make it possible to see which rules fired and when the score last moved. Transparency is also your debugging tool: when a surprising lead appears at the top, you need to find out why in seconds.

Treating the score as a verdict

A score is a prompt to look, not a ruling. The salesperson who knows the lead has just changed jobs and now controls a budget should be free to move them up, and to say why. Record those overrides; they are some of the best evidence for what the model is missing.

How Scalry handles this

The model itself is something you define. Scalry does not decide what makes a good lead for your business, and the signals and points in this article are examples, not rules Scalry applies for you. What it provides is the place to keep the result and the history behind it.

Every contact in Scalry’s CRM carries a lead score number, along with a timestamp showing when that score was last updated. Each change to a score is written to a log that records the previous score and the new one, so when a lead jumps or drops you can look back and see how it moved rather than guessing. That serves the transparency point above: the reasoning stays inspectable. The CRM also offers a “score” custom property type, which is useful if you want to hold a second number of your own, for example separate fit and engagement scores alongside the main one.

Lead finder and lead scoring are plan features, so check which plan includes them before you plan around them. Beyond that, the discipline is yours: choose the rules, set the thresholds, agree on what ready means, and revisit the model when your deals tell you it has drifted.

Questions to ask before you commit

Whichever way you are leaning, these are the questions that change the answer. Work through them against your own numbers rather than anyone's feature matrix — including this one.

• What decision does the score change? If you cannot name an action tied to a threshold, you have a number without a purpose.
• Can the person following up see why a lead has its score? And when it last changed?
• Is fit kept separate from engagement? Or can a highly engaged poor-fit lead outrank a ready buyer?
• Does engagement expire? What happens to a lead who was active two months ago and has been silent since?
• Which rules have been checked against won and lost deals? And which are only assumptions?
• Who owns the model and when is it next reviewed?
• Can a human override a score and record the reason?

If you cannot answer the first three, you are not ready to choose between Scalry and your lead scoring setup yet — you are still deciding what you need, which is a cheaper problem to solve first.

A plan for this week

Day 1: define ready

Write one plain sentence describing a lead you would call within the hour. Agree it with anyone else who follows up leads.

Day 2: list your signals

Write down five or six fit facts and four or five engagement behaviours that you honestly believe separate buyers from non-buyers. Give each a reason in a sentence, and cut any you cannot justify.

Day 3: assign points and thresholds

Use coarse numbers. Add a few heavy negatives for disqualifying facts, a decay period for engagement, and two or three bands with an action and a time limit each.

Day 4: look back

Take 20 to 30 past deals, score them using only what you knew at the start, and see where wins and losses landed. Fix the obvious misses, changing no more than a couple of rules.

Day 5: switch it on and schedule the review

Start scoring new leads, make sure the score and its history are visible to whoever follows up, and put a quarterly review in the calendar. After a month, compare how the “act now” band actually performed and adjust from there.

Frequently asked questions

What is lead scoring in simple terms?

Lead scoring is a way of giving each lead a number based on how well they fit your ideal customer and how strongly they have shown interest, so you can decide who to contact first. You define the rules and points yourself, and the score is a prompt for action, not a guarantee.

Do I need software to score leads?

No. A spreadsheet with a few columns is enough to start. Software helps when you have enough volume that updating scores by hand becomes a chore, and when you want the score and its history attached to each contact record. The thinking matters more than the tool.

How many rules should a simple model have?

Roughly eight to twelve is plenty for a small business. Each rule should have a reason you can state in one sentence. Beyond that, the model becomes hard to explain, hard to trust and hard to debug.

What is the difference between fit and engagement?

Fit describes who the lead is and whether they could be a good customer, such as industry, size, role and location. Engagement describes what they have done recently, such as replying, asking for a quote or giving a timeline. Keeping them separate stops an enthusiastic poor-fit lead from outranking a ready buyer.

Why should lead scores decay over time?

Interest is perishable. A lead who asked for a quote months ago and then went quiet is not as promising as the same request made yesterday. Letting engagement points expire after a set period keeps the top of your list focused on people who are active now.

How do I check whether my scoring model works?

Look back at 20 to 30 recent won and lost deals, score each using only what you knew at first contact, and see whether wins cluster at the top. Read the misses individually and adjust a few rules at a time. With so few deals, treat the results as guidance rather than proof.

What is an MQL and who decides what counts?

MQL stands for marketing-qualified lead, meaning a lead judged ready for a sales conversation. There is no universal definition. The people who generate leads and the people who follow them up should agree on one in writing, and revisit it when the feedback from actual conversations says it is too loose or too strict.