Lead Scoring Mistakes to Avoid:Nine Ways It Goes Wrong
Lead scoring mistakes to avoid: nine choices that turn a score into a number nobody trusts, each with a plain fix you can make this afternoon, today, for free.
The biggest lead scoring mistakes are failures of honesty, not maths: scoring without a volume problem, rewarding engagement over buying, adding fit to intent, uncapped signals, passive qualification, no reasons, drifting inputs, no saved score, and refitting on too few deals.
Lead scoring mistakes to avoid, and why they all look reasonable
Nobody sets out to build a bad score. Every one of these mistakes made sense at the time.
That's what makes them expensive. They don't announce themselves. The score keeps producing numbers, the numbers keep looking official, and meanwhile the person who built it quietly stops sorting by that column and goes back to answering whoever wrote last.
There are nine lead scoring mistakes. They fall into three families, and the pattern underneath all of them is the same: they are failures of honesty about what the score knows, not failures of mathematics. Which is good news. Mathematics is hard. Honesty is available to you this afternoon.
| Family | The mistakes | What it costs you |
|---|---|---|
| Building it for the wrong reason | One, two, three | A score that measures the wrong thing |
| Building it loosely | Four, five, six | A queue full of the wrong people, the classic lead scoring mistakes |
| Never checking it | Seven, eight, nine | A model that stays as good as your first guess |
Family one: building it for the wrong reason
Mistake one: scoring before there is a volume problem. If you can read every lead yourself each week, a score is overhead. Your own judgement, applied to every lead, beats any model you could write, because you see context no rule can encode. A score replaces good judgement with worse arithmetic. Score when leads outnumber your capacity to read them. Not before. Whether you're there yet is the subject of whether lead scoring is worth it for a solopreneur.
Mistake two: scoring engagement instead of buying. Without a written target, the model rewards your most enthusiastic readers. Those are frequently students, competitors, job seekers and people who just like your writing. They are not your buyers. Fix: write one sentence before you assign a single point. This person is worth a sales conversation with me now, and a meaningful share of people like them become real opportunities. Check it against the people who actually bought.
Mistake three: adding fit and intent into one number. The sum hides the one distinction you most need before a call. Is this person able to buy, or merely active? A flood of activity from someone who can never buy ends up outranking a quiet signal from someone who could. Fix: make fit a gate (fits, does not fit, unknown) and let intent rank the people who pass. This one gets its own treatment in fit vs intent in lead scoring.
Free guide · 21 pagesRation the one thing you cannot scale.The Lead Scoring Playbook: deciding which leads deserve your time.Get the free guide →Family two: building it loosely
Mistake four: leaving repeatable signals uncapped. Page views, opens and visits can repeat forever. Without a ceiling, the score drifts into a traffic counter, and someone who reads your posts daily for a year outranks someone who asked for a call yesterday. Fix: give every repeatable positive signal a cap on its total contribution, and give the whole score a ceiling so it reads against a known maximum.
Mistake five: making marketing qualified reachable on passive signals. One missing condition and every regular newsletter reader lands on your call list. Fix: whatever the threshold, require at least one deliberate act, such as a form, an assessment or a page only buyers visit. Spelled out step by step in how to define a marketing qualified lead.
Mistake six: accepting a number without reasons. You can't open a conversation with a score. So you ignore it and go back to working the inbox. Fix: make every score explainable, line by line, largest contribution first. "Applied for the programme, read the pricing page twice this week" is a first line for your email. A bare number is an excuse.
Family three: never checking it
Mistake seven: never noticing the inputs drifted. You rename a tag, relaunch a form, restructure your site. A valuable signal quietly starts contributing nothing, and no error appears anywhere. You're the person who launched the form, so you're the person least likely to catch it. Fix: keep a running list of inputs that map to no signal, and look at it after every launch. If it grows, something important has stopped counting.
Mistake eight: keeping only today's score. Scores change every night. If you don't save what each lead scored on the day it reached you, the audit is impossible, and the model stays exactly as good as your first guess for as long as you use it. Fix: snapshot the score and its parts at every stage change, from day one. You can't recover it later.
Mistake nine: refitting on a handful of deals. A few closed customers have something in common by chance. Rebuild around them and you've fitted the model to noise. Fix: set a closed-deal trigger in advance, and until you hit it, only repair broken inputs and reorder tiers when the pattern is clear. How to run the check properly is in how to audit a lead scoring model.
Which of these are you making right now?
Be honest. Most people recognise three.
If you can read every lead and you built a score anyway, that's mistake one, and you can stop reading. Switch it off and use your head.
If your best-scoring contacts are friends, fans and competitors, that's mistakes two and three. Your score is a popularity contest.
If your call list is full of people who never answer, that's four and five. Cap the repeatable signals and require a real action.
If you sort by the score column but can't say why anyone is high, that's six. Add reasons or stop pretending you trust it.
If you can't remember the last time you checked that the score still works, that's seven, eight and nine, all at once. The most common family is the last one. It's also the most forgivable, because the cost is invisible. Nothing breaks. The model just slowly stops being right.
Free assessmentHow trapped are you, really?Take the Corporate Suffocation Index, the free solopreneur readiness assessment.Take the assessment →Three small lead scoring mistakes that cause a lot of damage
These didn't make the nine, but they travel with them.
Penalising personal email addresses. This is a standard tip and it's right for companies that sell to other companies, where a personal address means an individual outside your market. If your buyers are individuals, other solopreneurs, freelancers, professionals buying for themselves, then a personal address is the norm. Penalise it and you remove your best customers.
Huge negative scores for disqualified contacts. A big negative number can still be outweighed by enough activity, and it buries the reason inside arithmetic. Make exclusions explicit, with a stated reason: your own addresses, test addresses, competitors, people who never agreed to hear from you.
Showing never-scored as zero. A contact the scoring hasn't yet seen and a contact it has seen and found nothing for are different things. Showing both as zero tells you a lead is cold when nothing has looked yet. Show the first as unscored.
Free guide · 21 pagesRation the one thing you cannot scale.The Lead Scoring Playbook: deciding which leads deserve your time.Get the free guide →The ten-minute check
You don't need a project. You need ten minutes and a willingness to be annoyed.
- Open your contacts sorted by score. Read the top ten. Would you call them?
- Find one customer who bought. What did their score say at handoff? If you can't answer, that's mistake eight.
- Look at the share of contacts above your line. If it's most of them, tighten.
- Pick any high scorer and ask for the reasons. If there are none, that's mistake six.
- Check the list of inputs that score nothing. If you've never made one, make it.
Fix the first thing you find. Not all of them. One.
The queue you work from is something you can trust only if you keep the reasons visible, keep the score each lead had when it reached you, and check it against what closed. That's the whole discipline for avoiding lead scoring mistakes. The rest is detail.
Free assessmentHow trapped are you, really?Take the Corporate Suffocation Index, the free solopreneur readiness assessment.Take the assessment →Frequently asked questions
What are the most common lead scoring mistakes?
The commonest is never checking the model against outcomes. A score that is not compared with what closed stays only as accurate as your first guess, and drifts as your product and audience change.
Is it a mistake to score leads at all?
It is if you can read every lead yourself. Scoring solves a volume problem. Without one, your own judgement beats the arithmetic, so wait until leads outnumber the time you have.
Why does a score without reasons fail?
Because a number gives you nothing to say. You ignore it and go back to the inbox. A score with visible reasons hands you the opening line of the conversation.
How many signals should my model have?
A dozen is generous for most small businesses. Keep only the behaviours you genuinely record today, for most contacts. A signal you plan to capture next quarter scores nothing and makes the model look more complete than it is.
The full method, with the checklists, is in The Lead Scoring Playbook, a free 21-page playbook.
How trapped are you, really?
Take the Corporate Suffocation Index, the free solopreneur readiness assessment.