How to Build B2B Lead Scoring Without Historical Data

TL;DR. To build B2B lead scoring without historical data, start with explicit rules for customer fit and buying intent that sales can inspect. Use those rules to choose the next action, without claiming they predict a purchase. Add negative signals, expire old intent, and revise the rules using recorded sales outcomes.

What should your first lead score decide?

Choose one decision: which leads should sales review next? Keep that separate from forecasting revenue or estimating the chance of a sale.

Write an action for each priority:

• Sales review: check the evidence and decide whether to start a conversation.
• Nurture: offer a relevant next step while waiting for clearer buying evidence.
• Research: resolve missing information before assigning priority.
• Exclude: remove from sales outreach for a documented reason.

Define what “qualified” means before choosing signals. For example, require a supported use case, a relevant organization and an expressed need that sales can explore.

Agree who owns each queue and when it gets reviewed. MQL to SQL Handoff provides a framework for making that agreement explicit.

How do you separate fit from intent?

Keep two visible assessments. Fit asks whether you can serve this buyer. Intent asks what evidence supports a buying conversation now.

Start fit rules with your ideal customer profile. Choose criteria that change your ability to deliver value, such as the problem, operating requirements and serviceable market.

For intent, prioritize actions with clear meaning. A request to discuss a specific problem is easier to interpret than an unexplained page visit.

Dimension	Example evidence	Initial treatment
Fit	Confirmed problem your product addresses	Mark as supported fit
Fit	Required capability your product lacks	Flag a fit conflict
Fit	Organization details are missing	Mark as unknown
Intent	Requests a conversation about a stated need	Send for sales review
Intent	Describes an active evaluation	Review the context and next step
Intent	Reads educational content	Keep as an engagement signal

Treat these as proposed rules. Avoid presenting them as established predictors of conversion.

Keep unknown information separate from poor fit. An incomplete record needs research; a confirmed mismatch needs a different action.

How do you turn those rules into a working score?

Begin with a fit-and-intent matrix. It makes the reasoning visible without requiring invented numerical precision.

Illustrative starting model: these categories are operating choices, not measured predictions.

Fit	Current intent	Next action
Supported	Explicit buying request	Sales review
Supported	Educational engagement only	Nurture
Unknown	Explicit buying request	Prompt human review and research
Conflicting	Explicit buying request	Check the conflict before qualification
Conflicting	No buying evidence	Exclude from the sales queue

For each rule, record:

• The evidence required and where it comes from.
• Who or what the evidence applies to: person or account.
• The action it triggers.
• When it expires or needs checking again.
• The person responsible for resolving exceptions.

If your system requires points, use them to implement the agreed order. Label the weights as provisional. A larger score means higher priority under your rules; it does not mean a measured likelihood of buying.

Keep the underlying evidence visible beside the total.

Which negative signals should change the action?

Separate exclusion, reduced priority and missing information.

Exclusion rules might cover confirmed spam, duplicate records or enquiries outside your supported offer. Document the reason. Route existing customers and partner enquiries to their relevant teams rather than treating them as failed prospects.

Reduced-priority signals might include an explicit statement that a project is paused or that no evaluation is planned. Record what was said and the agreed next step.

Missing information should trigger research. Do not automatically penalize a lead for lacking a job title or company size.

Use specific outreach preferences to control outreach. An unsubscribe should stop the affected marketing messages, regardless of the score.

Apply negative evidence at the right level. One contact leaving an organization should trigger a contact update, rather than excluding the entire account.

How should score decay work?

Use decay to prevent old activity from retaining priority indefinitely.

For the first version, make intent signals expire after a defined review window. Choose that window as a provisional operating rule based on how your team handles enquiries. Review it once outcomes accumulate.

Keep the rules simple:

• Give each intent signal a timestamp and expiry condition.
• Remove expired intent from active priority.
• Preserve the activity history for review.
• Reassess fit when relevant information changes.
• Respect an agreed follow-up time when a buyer says “later”.

Do not treat another educational page visit as proof that an old buying request is active again. Require fresh evidence for renewed buying intent.

Write the decay rule before launch. Otherwise, individual salespeople may interpret the same ageing lead differently.

What does a worked example look like?

The following cases are illustrative.

Lead A matches your target customer and asks for a conversation about a supported problem. Route the lead for sales review. Sales still needs to confirm the situation; the score has not established budget, authority or purchase timing.

Lead B matches your target customer and reads several educational resources. Keep the lead in nurture. Under this model, reading activity alone does not trigger a sales handoff.

Lead C asks for a demo but provides little organization information. Send the request for human review and research. Missing fields should not bury an explicit enquiry.

Lead D previously described an evaluation, then says the project is paused. Replace the active-intent status with the stated pause and record any agreed follow-up.

Each decision should be explainable in one sentence: “This lead is here because of this evidence, and this is the next action.”

How do you improve the rules using actual outcomes?

Schedule a recurring review with marketing and sales. Examine records and evidence, including leads that received little attention.

Capture the rule version, evidence and priority at handoff. Then record what happened:

• Sales accepted or rejected the lead, with a reason.
• A conversation was held.
• A relevant problem was confirmed.
• An opportunity was created.
• The deal was won, lost or remained unresolved.

Keep “not contacted” separate from “contacted and unqualified”. Keep open opportunities separate from losses.

Review high-priority leads that proved unsuitable and lower-priority leads that progressed. If only prioritized leads receive attention, the review cannot tell you much about the leads you overlooked. Manually inspect some lower-priority records too.

Change one rule at a time where practical. Document the reason, retain the previous version and set a review point. For example, if content engagement repeatedly triggers conversations with no relevant problem, test removing it as a handoff trigger.

Use early outcomes to question assumptions. Avoid turning a handful of examples into a claim that the model predicts revenue.

What are the common mistakes, and when does this not apply?

• Combining everything into one opaque total. Show fit, intent and exceptions separately.
• Counting repeated activity without limits. Cap repeated engagement so it cannot substitute for fit or an explicit need.
• Spreading intent across an entire account. Identify which person expressed the need and why it matters to the account.
• Rejecting leads because fields are blank. Research unknowns before declaring a mismatch.
• Treating sales acceptance as final success. Follow accepted leads through later outcomes.
• Changing rules after every surprising result. Record exceptions and look for a reason to revise the policy.

If every inbound enquiry can receive a timely human review, start with routing and qualification rules. A scoring system may add little to that process.

For a small list of named accounts, account research and buying conversations may be the more useful starting point. The account-based marketing playbook covers that approach.

What should you check before launch?

• Each priority has an owner and a next action.
• Sales can inspect the evidence behind every decision.
• Unknown information is distinct from confirmed poor fit.
• Negative signals and outreach preferences override the relevant actions.
• Intent has an expiry or review condition.
• Outcomes and rule versions are recorded.
• A recurring review includes overlooked leads and unresolved cases.

Related reading

• How to Build an ICP — define the fit criteria behind your model.
• MQL to SQL Handoff — agree qualification, ownership and follow-up.
• Account-Based Marketing B2B Playbook — organize work around named accounts.