How Should You Set Lead Score Decay for Stale Intent?

Why are your hottest leads in the CRM often the coldest buyers?

Because most lead scores only add points and never take them away. A contact who downloaded three guides last spring still looks hot today, even if they went quiet months ago. Score decay fixes that by shrinking old engagement points over time, so the score reflects current intent instead of history.

It is the first pattern I look for in a CRM audit. Sales opens the "hot leads" view, calls the top ten names, and half of them say they solved the problem last quarter. Reps stop trusting the score after two weeks of that. Once trust is gone, the score becomes a number nobody looks at.

This article is about one fix: deciding how fast intent should fade, and setting that up so the score earns its place in your go-to-market motion again.

What is lead score decay?

Lead score decay is a rule that reduces the points a scored event contributes as that event gets older. A demo page visit from yesterday counts fully. The same visit from four months ago counts for less, or nothing. Decay keeps engagement scores honest about time, which raw point totals never are.

It is worth separating two kinds of scoring here. Fit scoring asks whether a company and contact match your ideal customer profile: industry, size, role, region. Fit rarely changes week to week. Engagement scoring asks whether this person is showing buying behavior right now. Engagement is the part that goes stale, so engagement is where decay belongs.

If your score mixes both into one number with no decay, a perfect-fit contact who did nothing for a year can outrank a decent-fit contact who visited pricing twice this week. That is backwards for any team that sells on timing.

How does score decay work in HubSpot?

HubSpot documents decay for engagement or combined score event groups. Its knowledge base says you can set decay intervals of every 1, 3, 6, or 12 months, with each month equal to 30 days. Each event decays on its own, follows linear logic, and the decayed values add up to the overall score.

The HubSpot example is useful. A rule gives 10 points for a form fill with decay set to 50 percent every month. One month later that event contributes 5 points. HubSpot says decay is based on the original event value, so 30 days after that, the form fill contributes 0 points. Decay also applies to historical data, so turning it on changes scores for old events immediately.

One limit matters for design. HubSpot says decay is not supported for segment membership criteria, because membership is a yes or no condition rather than a timestamp. Its suggested workaround is to put a date filter inside the segment, such as contacts who submitted a form within the past 30 days. Check HubSpot's current scoring docs before you build, since the scoring tools have changed more than once.

Why does intent fade at different speeds for different signals?

Because each signal says something different about timing. A pricing page visit says "evaluating now." A webinar signup says "curious this quarter." A newsletter click says "still interested in the topic." Treat them the same and your score either forgets serious buyers too fast or remembers casual readers too long.

I group signals into three speeds. Fast signals are the ones closest to a purchase decision: pricing page views, demo page visits, comparison page visits, and replies to sales emails. Medium signals show active research: webinar attendance, case study downloads, and repeat visits to product pages. Slow signals show general interest: blog reads, newsletter clicks, and social follows.

The speed you assign should match your sales cycle. For a product with a 30-day cycle, fast signals should lose most of their weight within a month. For an enterprise motion with a nine-month cycle, the same signals can hold weight longer, because buyers really do go quiet for weeks during procurement.

How do you pick decay settings without guessing?

Use your own closed-won deals. Pull the last 20 to 50 won deals, look at when each key signal happened relative to the opportunity being created, and set decay so the signal still counts during that window. The data you already have beats any benchmark you could borrow.

Here is the exercise I run. For each won deal, note the date of the first pricing visit, the first demo page visit, and any content downloads. Then note the opportunity create date. If most pricing visits happen within 30 days of the opportunity, a monthly decay on that signal makes sense. If downloads tend to happen three to six months earlier, give them a slower decay or a smaller weight.

If you have fewer than 20 won deals, do not overfit. Start with simple defaults: fast signals decay monthly, medium signals every three months, slow signals every six months. Revisit after a quarter of real data. A rough rule you review beats a precise rule you never check.

What mistakes break a decayed lead score?

The three I see most are decaying fit data, turning decay on without warning sales, and stacking decay on top of an already strict threshold. Each one makes the score swing in ways reps cannot explain, and an unexplained score gets ignored faster than a stale one.

Decaying fit data is the quietest mistake. Company size and job title do not become less true over time, so they should not lose points. Keep fit in its own score or its own group with decay off. In HubSpot terms, that usually means a separate fit group from the engagement group, which is where the documented decay setting lives.

The sales warning matters more than people expect. Because HubSpot applies decay to historical data, many contacts will drop in score the day you switch it on. If reps open their view and see half the list gone, they will assume something broke. Tell them first, and show three named examples of contacts that dropped and why. I covered the trust side of this in how to build a lead score sales will trust.

How should decay change your MQL threshold?

Lower it, or at least re-test it. A threshold set before decay assumed points piled up forever. With decay on, the same contact will carry a smaller live score, so the old cutoff will flag fewer leads. Re-check how many contacts cross the line each week before and after the change.

My method is simple. Before switching decay on, count how many contacts crossed your MQL threshold in each of the last eight weeks. After switching it on, look at the same count for the next four weeks. If volume drops sharply, look at who dropped out. If they are contacts sales would never have called anyway, the new volume is the honest one. If good accounts disappear, your fast signals are decaying too quickly.

This is also the moment to look at what feeds the score. If intent data from review sites or website visits carries a lot of weight, decide whether it needs its own speed. I wrote about weighing those sources in how to score intent signals from G2 and website visits.

Should a small team skip decay and use date filters instead?

Sometimes, yes. If your score has only three or four rules, a date-filtered rule such as "visited pricing in the last 30 days" is easier to explain than a decay curve. You lose the gradual fade, but you gain a rule every rep understands on first read.

For a seed-stage team with one founder selling, I usually recommend the date-filter version first. The founder can read the rule out loud and know exactly why a contact scored. When the team adds SDRs and the rule set grows past a handful of signals, gradual decay starts to earn its complexity.

If you want the bare-bones version, I walked through a small setup in building a three-property lead score in HubSpot. Add a date window to the engagement rules there and you have a lightweight decay model in an afternoon.

What should you do next?

Split fit from engagement, assign each engagement signal a fast, medium, or slow speed, and check those speeds against your won deals. Warn sales before you switch decay on, then compare weekly MQL volume for a month. Adjust the fast signals first, since they move the score most.

A lead score is a promise to sales that the top of the list deserves a call today. Decay is how you keep that promise as time passes. Without it, your score is a history report dressed up as a priority list.

If you want a second pair of eyes on your scoring model or help rebuilding it in HubSpot, reach out and let's chat.