A lead scoring model that weights buying signals

A lead scoring model that weights buying signals ranks accounts by two things at once: how well they fit your ideal customer, and whether something just happened that makes them likely to buy now. Most teams score only the first. That is why their best-scoring accounts often sit cold for months.

This guide shows you how to build a lead scoring model that blends fit with signals, so reps work accounts that match and are in motion, first.

TLDR

• Fit-only scoring tells you who to target, never when. It ranks a dormant perfect-fit account the same as one that just raised a round.
• A better model gives each account a fit score and a signal score, then combines them.
• The signal score has three parts: recency, type, and size. Recent, high-value, big-money signals score higher.
• Weight the signal score a little above fit, because timing is what most outbound gets wrong.
• Refresh scores daily so new signals lift accounts and old signals decay. Static scores go stale fast.

Why does fit-only lead scoring miss good leads?

Fit-only scoring misses good leads because it answers the wrong half of the question. It tells you which accounts match your ideal customer. It says nothing about when any of them is ready to buy.

Picture two accounts that both fit your profile perfectly. Same size, same sector, same region. One raised a Series B last week. The other has had no news in two years. A fit-only model scores them identically, so a rep works them in whatever order the list loads.

That is a wasted opportunity. The funded account has new budget and shifting priorities right now. The quiet one might buy someday, but not because anything changed this week.

Most outbound fails on timing, not targeting. You are rarely short on accounts that fit. You are short on knowing which of them is in motion today. A scoring model that ignores signals cannot tell you that. For the wider case, see what buying signals are in B2B sales.

How do you build a lead scoring model that weights signals?

Build it from two scores you add together: a fit score and a signal score. Fit answers “should we ever sell to them.” Signals answer “should we call them this week.” You need both, and you weight signals a little higher.

Here is the method, in five steps.

1. Score fit out of 50. Use the firmographics you already trust: stage, sector, size, region, tech in use. A clean match scores high, a loose one scores low, a non-fit scores zero. Cap it at 50.
2. Score the signal out of 50. Build this from three parts that multiply together: recency, type, and size. A fresh, high-value, large event scores near 50. No signal scores zero.
3. Add them for a total out of 100. Total score = fit score + signal score. Sort your list by this number, highest first.
4. Set a work threshold. Decide the score below which a rep should not bother. A common line is 60, which forces both a decent fit and a real signal.
5. Refresh daily. Recompute signal scores every day so new events lift accounts and old ones fade. Fit changes slowly, signals change fast.

The signal score is where the model earns its keep, so score its three parts deliberately.

Signal factor	What it measures	How to score it
Recency	How long ago the event fired	Full weight in the first 2 weeks, half by 6 weeks, near zero after 3 months
Type	Which signal it is	Funding highest, then exec move, then acquisition, then hiring growth
Size	How big the event is	A larger round or a more senior hire scores higher than a small one

A simple way to combine them: pick a base value for the signal type, scale it by a recency multiplier from 0 to 1, then add a small bonus for size. Funding round base 40, fired 5 days ago so recency near 1, large round so a size bonus of a few points, lands near 45 out of 50.

What does a scored account list look like?

Here is the model run on four accounts. Each one fits the profile reasonably well, so fit scores are close. The signal score is what spreads them out and decides the work order.

Account	Fit (50)	Signal	Recency	Size	Signal (50)	Total (100)
Northwind (fintech)	44	Series B funding	4 days	$40M	47	91
Cedar Labs (devtools)	40	New VP of Sales	3 weeks	senior	30	70
Atlas Freight (logistics)	46	Acquired a competitor	5 weeks	mid	22	68
Bryce Retail (ecommerce)	42	None in 8 months	n/a	n/a	4	46

Read the order. Atlas has the best fit at 46, but its signal is five weeks old, so it lands third. Northwind has slightly worse fit but a fresh, large funding round, so it tops the list at 91. Bryce fits fine but has no signal, so it falls below the 60 threshold and waits.

This is the whole point. Fit alone would have put Atlas first and treated Bryce as a live lead. The signal score corrects both: it pulls the freshly funded account up and pushes the quiet one down. Reps now open their day with Northwind, not Bryce. For how to act once an account rises, see the signal-based selling playbook.

How do you keep lead scores fresh as new signals land?

You keep scores fresh by recomputing the signal half on a schedule, not by setting it once. Fit barely moves month to month. Signals move every day, both up when a new event fires and down as an old one ages.

Two things have to happen automatically.

• New signals lift accounts. When a funding round, exec move, or acquisition lands on an account, its signal score jumps and its total climbs the list the same day. A round you learn about a week late has already lost recency points it can never get back.
• Old signals decay. A score earned by a three-month-old round should not still read as a live trigger. The recency multiplier drags it down over time, so the account drifts back below the threshold unless a new signal arrives.

This only works if the signals feeding the model arrive in hours, not in a monthly batch. A batch update cannot lift an account inside the one-to-two-week window when a fresh round still matters. The data has to be one resolved company per event, deduplicated so one round does not double-count, and fresh enough to act on. That is what Datahyena delivers as a feed you can pull from your scoring job: funding rounds, acquisitions, and executive moves over an API.

What are the common mistakes in signal-based lead scoring?

The two that quietly break a model are static scores and over-weighting vanity signals. Both make the list look busy while pointing reps at the wrong accounts.

• Static scores. A score computed once and never refreshed treats a stale event like a live one. The account that raised six months ago still sits near the top, long after the budget was spent. If your model does not decay, it lies a little more every week.
• Over-weighting vanity signals. Not every event means budget. A press mention, a webinar signup, or a generic “we are growing” post is weak compared to a funding round or a new VP. Score weak signals low. If a soft signal can push an account above the work threshold on its own, your weights are off.
• No fit floor. A huge signal on an account that does not fit is still a bad lead. Keep fit in the total so a giant round at a company you can never sell to does not jump the queue.
• Counting one event many times. The same round reported by five outlets must score as one signal, not five. Duplicate events inflate scores and waste rep time on phantom urgency.

Get those right and the model stays honest: high scores mean a real fit plus a real, recent reason to call.

Score your first account on a live signal

The fastest way to test this model is to score one real account. Pull a live funding event with 50 free credits, no card required, and run it through your fit-plus-signal formula. When you are ready to wire signals into your scoring job, the signals overview shows everything we track.