Build a Lead Score That Can Say No

A lead score that can only go up will eventually call almost everyone interested.

Every repeat visit adds points. Every page adds points. A job seeker, supplier or existing customer can move ahead of a suitable prospect simply by spending more time on the site. The score looks busy while the sales queue becomes less useful.

Build a score that can reduce confidence, route a record elsewhere or leave the next step undecided. The purpose is to help the team choose a sensible action, not to turn all activity into urgency.

Decide what the score is for

Choose one decision. For example: which eligible contacts should a representative review for a useful follow-up today?

That is narrower than predicting who will buy. It also makes the score easier to evaluate. You can inspect whether the review was useful, whether the record belonged in the queue and whether the representative had a reason to contact the person.

Write the intended action beside the score. Do not let another team reuse a review-priority number as a probability of purchase without establishing that it supports that interpretation.

When exploring how OSPRY connects visitor context with action, start with that decision. A better queue needs a clear job before it needs more inputs.

Put stop conditions outside the points

Some conditions should not be overcome by another page visit.

If a contact asked you to stop messaging, a high score does not erase the request. If the confirmed need is something your business cannot provide, more activity does not make the offer suitable. If the identity association is unresolved, investigate before treating the record as a known person's behavior.

Keep those conditions visible before a representative uses the score. They are routing and eligibility decisions, not ordinary deductions that a few positive signals can cancel.

Use facts the person supplied or your business can substantiate. Do not guess a service location from an IP address or infer that someone wants a job because they opened one careers page.

Separate fit from current interest

A suitable business can be researching for next year. A highly active visitor can be looking for something you do not sell.

Keep a short fit assessment beside the activity score. Record the relevant need, service area or business requirements, the evidence and any unresolved question. “Unknown” should remain available.

Then assess current interest using the activity and conversation you actually observed. A stated project question deserves different consideration from a generic page opening.

Avoid counting one underlying event several times. A page opening, a refreshed contact card and a notification about the same visit should not automatically become three independent reasons to call.

Reviewing visitor measurement features is easier when you ask how an input was established and whether it adds genuinely different evidence.

Write sample rules in plain language

Start with a small set that a representative can challenge. The following numbers are illustrative design choices, not industry benchmarks or a validated model.

For an eligible contact, you might propose:

• Add 20 points for an explicit, relevant project question.
• Add 10 points for a defined buying-page visit, counted once within a 14-day window.
• Add five points for a separately observed return that adds useful context.
• Subtract 15 points when the contact explicitly says the project falls outside the current planning period.
• Subtract 30 points when the inquiry is confirmed to be exclusively an employment request or supplier pitch, then route it to the appropriate team.
• Reduce old activity points by an agreed amount when no fresh relevant evidence arrives during a defined review period.

State what qualifies for each rule. “Relevant project question” needs examples. “Return visit” needs a reliable association. A timer should reduce old activity because it is older evidence, not pretend to prove that the person has lost interest.

Test the rules against awkward cases

Use examples that would embarrass the old score.

Imagine a hypothetical supplier who opens several buying pages and returns later. Under the sample rules, the capped buying-page activity adds 10 points, and the return adds five. The supplier then sends a message confirmed to be solely a pitch for their own services. The 30-point deduction produces a score of negative 15, and the inquiry is routed out of the sales-prospect queue.

That result does not mean the supplier is a bad person or can never become a customer. It describes the purpose of this inquiry.

Now imagine a suitable contact with a relevant project question, the capped buying-page visit and a meaningful return. The same rules produce 35 points. If they explicitly say the project is for a later planning period, subtracting 15 leaves 20.

The second contact may deserve preparation or a later review. The recorded timing should shape the response, even if the remaining score looks respectable.

Make the routing rule explainable

Choose review bands only after inspecting examples. If you use sample bands such as 30 and above for review, 10 through 29 for preparation and below 10 for holding, label them as a starting experiment.

No band should trigger a message without checking the relationship and the stop conditions. A low score should not prevent you from answering a direct customer question either.

Write the action, owner and reason for each route. A representative should be able to see “project deferred by the contact” rather than only “20.”

Check the integrations carrying those records. Preserve the explanation when the score moves into the CRM. A number without its important reasons is difficult to correct and easy to misuse.

Keep permissions beside the decision

The score describes evidence used for a review. It does not create permission to collect more information or use a contact channel.

Make recorded preferences and restrictions available before outreach. Check consent and data choices alongside the identity and activity record. Avoid putting sensitive details into event labels merely to make scoring more convenient.

Keep access to the score appropriate to the task. A team reviewing page performance may need aggregated patterns without a named list of people.

If someone corrects a mistaken association or asks for contact to stop, the relevant downstream queue needs the update too. Correcting one screen while leaving an old task active does not solve the operational problem.

Review the rejected and deferred records

A score should be open to correction.

During a small pilot, inspect records from every band, including ones the rules held back. Ask whether the fit evidence was accurate, whether an old event was counted twice and whether the negative rule described this inquiry rather than a permanent label about the person.

Record useful conversations, wrong associations, misrouted customer requests and representative overrides. Review why the overrides happened before changing the weights.

Keep the original rule version and the date of any change. Otherwise, a better-looking result may reflect a different definition rather than a more useful process.

The score earns its place when the team can explain why a record needs attention and when leaving it alone is the better decision. Start with a few clear rules, test the difficult cases and let the score say no when the evidence supports it.