Most businesses do not have a lead quality problem. They have a lead ordering problem. Every enquiry lands in the same inbox, gets the same treatment, and the one that was ready to buy waits behind four that were never going to. At Growth Bully, a Malta performance marketing agency, this is the cheapest thing we fix in a sales process, and it rarely needs new software.
Almost everything published on lead scoring is written for a company with a marketing operations team and a platform licence to justify. This is the version for a business with two people in sales and a spreadsheet.
What is lead scoring?
Lead scoring is a rule that turns what you know about an enquiry into a number, so whoever follows up works the list in order rather than in arrival sequence. It combines who the person is with what they did, and its only job is to decide who gets called first and who can wait.
It is not a prediction and it is not a filter. A score of nine does not mean a sale and a score of two does not mean a bad lead. It means one of them should be called within minutes and the other can be emailed this afternoon. Scoring that is used to discard leads rather than to sequence them usually makes the numbers worse, which is covered in why leads do not convert.
Do you actually need lead scoring?
Only if you receive more enquiries than you can call inside the hour. Below that volume the correct answer is to call everybody quickly, and a scoring model is a solution hunting for a problem. Above it, the order you work the list in starts to decide your conversion rate.
Two other situations justify it earlier. The first is a wide spread in deal value, where a small enquiry and a large one look identical on the form. The second is a split in ownership, where marketing counts the lead and sales decides whether it was real. A shared score is the cheapest way to settle that argument before it becomes a monthly meeting.
What should a lead scoring model measure?
Two axes, kept separate. Fit is who they are: sector, size, the role of the person enquiring, location, whether they resemble customers you already serve well. Behaviour is what they did: the page the enquiry came from, how specific the request was, whether they left a phone number, whether they named a timeline.
Keeping the two apart matters because they fail differently. A perfect fit with no behaviour is a prospect for outbound, not a lead. Strong behaviour with poor fit is somebody who will take a call and never buy. Collapsing both into one number hides which of those you are looking at.
- Fit, five signals at most. Sector, company size, the enquirer role, whether they sit inside your service area, and whether they match a named existing customer.
- Behaviour, five signals at most. The page the enquiry came from, what the request actually said, whether a phone number was given, whether budget or timing was mentioned, and how many visits preceded the form.
- Nothing you cannot see. If a signal needs a tool you do not own or a form field nobody completes, it is not a signal. It is a wish.
Ten signals is already more than most teams need. Six, well chosen, beats twenty that nobody updates.
How do you set the points without guessing?
Work backwards from your last fifty closed deals and your last fifty that went nowhere. For each signal, count how often it appears in the won list against the lost list. A signal that shows up in most wins and few losses earns points. One that appears evenly in both earns nothing, however sensible it sounds.
Fifty of each is enough to expose the obvious patterns and small enough to finish in an afternoon. Keep the scale short, five points of range rather than a hundred, because a long scale invites false precision and arguments about whether a lead is a sixty three or a sixty seven.
Resist weighting by instinct. The signal every sales team overrates is enthusiasm on the first call, and the one almost everybody underrates is which page the enquiry came from. A request from a page about a specific service carries intent that a contact form on the homepage does not.
What score should trigger a sales call?
Set the threshold by capacity, not by theory. Count how many genuine first calls your team can make in a day, then place the cut off so roughly that many leads clear it. A threshold that lets through more leads than you can call the same day is not a threshold, it is a wish list.
Then write down what happens on each side of the line, and who owns it. Above the line is a call attempt within minutes, which is the single biggest lever in the whole process and the reason speed to lead beats almost every other optimisation. Below the line is an automated sequence that keeps the lead warm until behaviour pushes it up.
Should scores go down as well as up?
Yes, on two counts. Negative scoring removes points for signals that reliably predict wasted time: a role with no budget, a request outside your service area, an enquiry from a competitor address. Decay removes points as a lead ages, because interest is perishable and a score earned three weeks ago is describing a person who has moved on.
Decay is the more valuable of the two and the one most often skipped. Without it, a list sorted by score slowly fills with old leads that once looked excellent, and the newest enquiry sits underneath them. A simple rule works: a lead loses a point a week after the first fortnight, and drops out of the active list entirely after ninety days into a reactivation flow rather than the bin.
How often should you rebuild the model?
Review it quarterly against the deals that actually closed, and rebuild it whenever the offer or the market moves. The test is one number: of everything you won last quarter, what share scored above your threshold when it arrived? If it is under half, the model is describing a business you no longer run.
Keep the definitions, the owner and the review date on one page with the rest of your numbers, which is what a marketing measurement plan is for. Scoring that lives in one person's head is not a system, and it leaves with them.
Scoring sits on top of two things it cannot replace: an honest definition of what a qualified lead is, and a measured lead to meeting rate to tell you whether the ordering is working. The automation that runs the sequences and the decay is LeadLock, the follow up layer behind our lead generation work, and the outbound equivalent for named accounts is the Decision-Maker Pipeline. If you would rather see the gaps before building anything, the pipeline scorecard finds them in order.