Two categories of deduction
Fit deductions cover what the company or contact is. These are stable attributes that do not change based on what the person did today.- Company size below the serviceable floor - Industry outside the ICP or on an excluded list - Competitor domain - Region you do not sell into - Role with no purchasing influence and no path to it
Behavior deductions cover what the person does and how recently. These are the signals that separate an active buyer from a subscriber.- Visiting only careers or support pages - Unsubscribing from email - No activity for a defined period - Repeated content downloads with no product page visit
Fit deductions should be large enough to disqualify on their own. Behavior deductions should be smaller so they accumulate into a picture rather than veto a record on one action.
Score decay is negative scoring
The most valuable deduction rule in most models is time. A lead that visited the pricing page in March and did nothing since is a different prospect from one that visited yesterday, and a static model treats them identically.
Decay handles this without a separate rule set. Subtract points on a schedule for inactivity so scores fall as engagement ages. Without decay, scores only rise, records accumulate points forever, and the MQL queue eventually fills with leads whose intent expired a year ago.
Validate against rejection data
Negative rules are guesses until you check them against what sales actually rejects.
Pull the leads sales rejected over a full quarter and look at where they scored. Rejected leads clustering at high scores means the deduction rules are missing the attributes reps care about, and the rejection reason codes name those attributes directly. The most common gap is a fit attribute the model never captured, usually company size or an industry the sales team knows does not buy.
Run the check in the other direction too. If reps ask for leads that turn out to be sitting just under the threshold because of a deduction, that rule is too aggressive and needs to be resized.
What it protects
Every unqualified lead that reaches a rep costs response time that a real lead needed. It also erodes trust in the queue, and once reps stop opening MQLs on arrival, the entire routing and response system stops working regardless of how fast it assigns records.
The downstream effect is the one that shows up in the numbers. Fewer real opportunities get created than the demand plan assumed, and the shortfall lands in pipeline coverage a quarter later, where it looks like a demand generation problem rather than a scoring problem. Clean deductions are the cheapest control available, since they cost model maintenance rather than budget. They also improve the inputs that any sales forecast depends on.
Frequently Asked Questions
Why use negative scoring instead of just filtering leads out?
Filters are binary and brittle. A hard exclusion on free email domains blocks the founder of a real prospect who signed up with a personal address. A deduction lets that record still qualify if the rest of its signals are strong enough. Negative scoring handles the gray area, and most of the volume lives in the gray area.
What should lose a lead points?
Two kinds of signal. Fit deductions cover attributes that place the record outside the ICP, such as company size below the serviceable floor, an excluded industry, or a competitor domain. Behavior deductions cover actions that predict low intent, such as visiting only the careers section, unsubscribing, or going inactive for a defined period.
How much should a negative rule deduct?
Enough to matter. A deduction that cannot pull a lead back under the threshold on its own does nothing except make the model look thorough. Size the strongest fit deductions so a single one disqualifies the record, and keep behavior deductions smaller so they accumulate rather than veto.
How do you know if negative scoring is calibrated correctly?
Look at the scores of leads sales rejected last quarter. If rejected leads cluster at high scores, the negative rules are missing the attributes that reps actually reject on. If leads that reps wanted are sitting just under the threshold with deductions applied, a rule is too aggressive. The rejection reason codes tell you which rule to change.
Put these metrics to work
ORM builds custom revenue forecast models that turn concepts like negative lead scoring into prescriptive action for your team.
Schedule a Demo