B2B Buying Signals: Who Sets Your Scoring Weights?

The weights behind B2B buying signals: who decides the score 

A sales leader looks at a lead in the queue and asks the one question nobody in the room can answer: “Why is this a 78?” It’s a 78 because of weights someone chose, for a reason nobody wrote down, and that person has probably left the company. You’re routing leads to the right rep on the strength of that number, which means you’re routing on an argument you can no longer read. 

Which buying signals b2b teams collect matters less than who decides how much each one counts, and how to set those weights so the number survives a challenge from sales. A demo request, a pricing-page visit, and a third-party topic surge can all feed the same score. The number that comes out depends entirely on the weights, and on who chose them. 

A score is an argument, not a measurement 

A composite lead score is an argument dressed as a measurement. A measurement has a unit and a referent: 72 degrees Fahrenheit, 15 milliseconds, $40,000. A lead score compresses several kinds of evidence into one number, and the compression is a choice. Someone decided a demo request is worth 30 points and a webinar registration is worth 5. That ratio is a claim about your buyers, not a fact. 

When the argument behind the number is invisible, the field stops trusting it. In Salesforce’s 2024 State of Sales report, only 35% of sales professionals completely trust the accuracy of their organization’s data. A rep gets a “hot” lead that goes nowhere twice, decides the score is noise, and starts working the queue by gut. Nobody logs that decision. The model still produces 78s, but the people at the end of the pipe quietly opted out. 

Three ways weights get set, and what each optimizes for 

There are three honest ways to set the weights behind a lead scoring model. The problem is using one without knowing what it optimizes for. 

Vendor default. Fast, reasonable, and available on day one. The catch: default weights encode the vendor’s data strengths, not your buying reality. A provider with deep third-party topic coverage will weight topic surges heavily, because that’s the signal it can see best. That’s not dishonest; it’s structural. You’re inheriting a view of importance shaped by what the platform measures well. 

Regression against closed-won. Defensible and grounded in your own history. You look at which behaviors preceded closed deals and let the math assign weight. The catch: regression optimizes for the past. It describes the buyer who bought last year, at last year’s price, for last year’s product. The moment the product changes or you enter a new segment, the weights drift, and drift is silent. 

Explicit business judgment. The slowest to set up and the only one that survives an argument with sales. You name why each behavior should count and write the reasoning where someone can read it later. More work up front, but it’s the only method where a rep’s “why is this a 78” gets an answer. 

Method	Optimizes for	Right when
Vendor default	Speed to launch	You’re new to scoring and need a starting baseline
Regression on closed-won	Fit to your history	Your product, price, and market are stable
Explicit judgment	Defensibility	Sales needs to trust and act on the score

Most teams should start with a default, pressure-test it against regression, and end at explicit judgment. The three aren’t rivals. They’re a sequence. 

Why better data beats a better model 

Better data beats a better model, and marketers already know it. In Digitalzone’s 2026 survey of 1,500 B2B marketers, respondents ranked data quality and accuracy as the single biggest driver of campaign performance, above creative, messaging, targeting, and technology. 

That’s not just a Digitalzone finding. In Lotame and Cint’s 2024 global survey of 1,200 marketing professionals, 97% of marketers and agencies reported encountering barriers in data orchestration and use. A sophisticated model over weak inputs just produces confident wrong answers faster. Complexity doesn’t fix bad data; it launders it. 

For routing, this becomes a concrete problem: the unit problem. When the underlying signal can’t name a person, no weighting scheme can recover one. You can tune the ratios all day. If the input is an anonymous account-level surge, the output is still an anonymous account-level surge with a number stapled to it. 

What the unit of scoring should be 

The contact should be the unit of scoring, and the account should be the roll-up, not the reverse. 

Picture two accounts, both scoring 60. The first 60 comes from one analyst who’s read every page, downloaded every asset, and attended two webinars. The second 60 is spread across six people: a VP on the pricing page, two directors who opened an email, a manager at a webinar, and two others who touched a single asset. The first is one motivated individual who may have no budget. The second is a buying committee waking up. According to Forrester’s 2026 State of Business Buying report, the typical B2B buying decision now includes 13 internal stakeholders, and that number rises for complex purchases. An account-level model that only sees “60” cannot tell them apart, and it will route them identically. 

Summaries hide distribution, and distribution is what matters when you’re deciding who to call. In Digitalzone’s own campaign data, accounts engaged by two or more tactics surface 72× more buying-committee members than single-tactic accounts. That gap comes from having contact-level signals to weight in the first place. Score the people first. Roll them up second. Here’s a 3-step method for layering contact signals on top of account data. 

A method for setting weights deliberately 

Set weights in five steps, in order. The discipline isn’t in the math. It’s in writing down the argument so the next person can challenge it. 

1. Name the single decision this score serves. Routing, sequencing, spend allocation, or sales alerting. Pick one. A number trying to serve two decisions does neither well.

2. List the behaviors that plausibly precede that decision. For routing to an AE: a demo request, a pricing-page visit, repeat visits from multiple people at one account. Leave off the behaviors that feel important but don’t precede the decision you named.

3. Assign a weight to each, with a one-sentence written rationale. The rationale is the deliverable, not the number. “Pricing-page visit = 25, because in our last two quarters it preceded a sales conversation more often than any other single behavior” is a weight someone can argue with. “Pricing-page visit = 25” alone is not.

4. Test the ranking against closed-won and closed-lost from the last two quarters. If your top-scoring leads include a pile of closed-lost and your closed-won sat mid-pack, the weights are wrong. Adjust before anything reaches a rep.

5. Fix a review cadence and a trigger for off-cycle review. Quarterly is reasonable. Name the events that force an early look: a pricing change, a new product line, a new segment. Weights drift as buyers and markets shift, and a cadence catches drift before sales does. (For the cost of getting this wrong, see what happens when stale intent signals reach your pipeline.) 

A worked example 

Say the decision is routing to an account executive. A defensible starting table might look like this, with each weight carrying its reason: 

Behavior	Weight	Rationale
Demo request	40	Highest-intent single action; directly requests a conversation
Pricing-page visit	25	In the last two quarters, preceded a sales conversation more than any other page
3+ contacts active at one account	20	Signals a committee, not a lone researcher (the routing case that matters)
Content download (mid-funnel)	10	Real interest, but weak on its own without a second signal
Webinar registration	5	Low commitment; registration is not attendance

When a rep asks why a lead scored high, you open the table: a demo request plus a pricing-page visit plus one more active contact. Which accounts you work first is a separate decision. That’s where a measurement framework that connects scores to pipeline outcomes comes in. 

The score has to be for something 

If you can’t say in one sentence what your score is for, the score isn’t ready to route anything. The weights aren’t a technical detail buried in the platform. They’re an argument about your buyers, and an argument nobody can read is one nobody will follow. Name the decision. Write the rationale. Review it on a cadence. That’s how you turn B2B buying signals into a score your team will actually follow. 

Two things have to be true for any of this to work. You need contact-level signals to weight in the first place, and enough reach across the committee that the roll-up describes a real group. If your data can’t name the people, talk to Digitalzone about contact-level signals for your target accounts. If your visibility across the buying committee is the gap, let’s talk about seeing the whole committee. Better demand starts with better understanding. And understanding starts with knowing what your score is actually for. 

FAQs 

What is a lead scoring model, and how do weights fit in? 

A lead scoring model assigns a single number to a lead based on their behaviors and attributes: a page visit, a download, a job title, a topic surge. Weights are how much each of those inputs counts toward the final number. The signals are the raw evidence; the weights are the argument about what that evidence means. Two teams with identical signals and different weights will produce completely different scores from the same buyer. 

How should I weight buyer intent signals for routing leads? 

Start by naming the single decision the score serves. For routing, that’s getting a lead to the right rep. List only the behaviors that plausibly precede a real sales conversation, assign each a weight with a one-sentence written rationale, then test the ranking against closed-won and closed-lost deals from your last two quarters. Weighting buyer intent well is less about the math and more about writing down why each behavior counts, so sales can challenge and trust the result. 

Why do regression-based lead scores stop working over time? 

Regression sets weights by looking at which behaviors preceded past wins, so it optimizes for the buyer you had last year. When your product, price, or market changes, the behaviors that predicted a deal shift too, but the model keeps scoring against the old pattern. A fixed review cadence, with triggers for pricing or product changes, is how you catch drift before sales does. 

Should B2B buying signals be scored at the account level or the contact level? 

Score at the contact level and roll up to the account, not the reverse. An account score of 60 driven by one obsessive researcher describes a completely different situation from a 60 spread across six roles in a buying committee, and an account-level model can’t tell them apart. Accounts don’t buy. People do. A score that can’t resolve to a named person can’t tell a rep who to call.