Set it from history, not intuition
The defensible method uses your own outcomes.
Pull twelve months of closed-won deals and identify the score each one carried at the moment it was passed to sales. Then do the same for leads sales rejected. Those two distributions overlap, but the overlap has an edge, and the threshold belongs near the point where accepted leads start outnumbering rejected ones.
This gives you a number with an argument behind it. When marketing is short on volume at quarter end and asks to lower the bar, the conversation becomes a discussion about a known acceptance curve rather than a negotiation.
Capacity is the real constraint
Quality is the reason teams say they set a threshold. Capacity is what should set it.
If a team of six reps can make a genuine first touch on 60 leads each per week, the ceiling is 360 leads. A threshold that passes 500 does not deliver 140 bonus opportunities. It delivers 360 worked leads and 140 that sit unopened, and unworked leads convert at close to nothing. The excess volume also degrades the leads that do get worked, because reps triage by whatever is on top rather than by score.
Set the threshold so expected weekly volume lands beneath the capacity ceiling. Then raise volume by improving the model, not by lowering the bar.
Symptoms of a badly set threshold
- Too low. Acceptance rate falls, reps start ignoring the MQL queue, and marketing hits its number while pipeline does not move. - Too high. Reps ask for more leads, opportunity creation runs under plan, and prospects that were ready sit in nurture receiving another newsletter. - Drifting. The score distribution shifts upward over quarters while acceptance stays flat, which means the same number now describes a weaker lead.
Recalibrate on a schedule
Scoring models decay because the behaviors they measure become more common. A pricing page visit was a strong signal when the site had six pages. It is a weaker one after three years of content publishing.
Review the threshold quarterly against acceptance rate by score band, and rebuild it whenever the ICP changes. Volume that clears a stale threshold looks like demand in a planning meeting and does not convert, which surfaces later as pipeline coverage that was never real and a forecast built on opportunity creation that does not arrive.
Frequently Asked Questions
How do you set a lead scoring threshold for the first time?
Work backward from history. Score twelve months of closed-won leads with the current model and find the range they clustered in at the time they were handed to sales. Set the initial threshold near the bottom of that range, then adjust once you can see acceptance rates by score band. Setting it from intuition produces a number nobody defends when volume gets contested.
Should the threshold be based on quality or on sales capacity?
Both, and capacity is the constraint people forget. A threshold that passes more leads than the team can work inside the response window converts worse than a higher threshold passing fewer, because unworked leads convert at close to zero. Calculate what the team can genuinely touch each week and set the threshold so volume lands under that ceiling.
How often should a lead scoring threshold be recalibrated?
Quarterly, and immediately after any change to the ICP, the pricing model, or the scoring rules themselves. Score inflation is gradual. Behaviors that were rare signals when the model was built become routine as content volume grows, and the same numeric score means something weaker two years later.
Should there be more than one threshold?
Yes for most teams. Separate thresholds by segment, because an enterprise account showing moderate engagement is often worth more rep attention than an SMB account showing heavy engagement. A single global threshold forces one of those two segments into the wrong treatment.
Put these metrics to work
ORM builds custom revenue forecast models that turn concepts like lead scoring threshold into prescriptive action for your team.
Schedule a Demo