Lead scoring: how marketing and sales score leads together

Lead scoring is the model that gives each contact a score based on how well they fit the customer profile your company serves well and on the interest they show, so the sales team knows who to talk to first and when a lead becomes sales’ responsibility. The model that works comes from an agreement between marketing and sales on what a lead ready for a conversation looks like, with the weights taken from the history of won deals. Points invented in a planning meeting produce a score that reps learn to ignore within a few weeks. If you lead marketing, sales or sales development, your job is to approve the criteria with the other team, define what happens in each band and review the conversion of each band every quarter.

In the diagnostics I run, the scene repeats. The company switched on scoring two years ago, someone gave ten points per email open and twenty per content download, and the MQL threshold was set at one hundred because it was a round number. In the SDR queue, the highest scores belong to students and vendors, and reps go back to picking leads by company name while the CRM recalculates the score every day.

What is lead scoring and where does it fit in the funnel?

Lead scoring is the automated part of qualification. It works with what the CRM knows before any conversation: company and job title data, coming from the form or from data enrichment, and the contact’s behavior on the website, in emails and at events. The SDR conversation, which confirms pain, context and decision maker, remains the step described in lead qualification. The score decides the order of the queue and the moment of the handoff, and the conversation decides whether the lead enters the pipeline.

That pre-conversation work matters because the B2B buyer spends most of the journey away from the seller. Gartner (2020) reported that B2B buyers considering a purchase spend only 17% of that time meeting with potential suppliers. Part of the rest leaves a trail in your channels, and lead scoring organizes that trail so the SDR reaches the contact while they are still assessing the problem.

Why doesn’t scoring invented in a meeting work?

Scoring defined in a meeting adds up what each person in the room believes about the customer. Marketing gives weight to the content that took effort to produce, sales gives weight to the job title that usually signs, and nobody checks afterwards whether leads above the threshold convert better than the ones below it.

The model sales respects starts with history. You pull the deals won in the last 12 to 18 months, look at what those customers had in common when they were still leads and compare them with all the leads from the same period. If most wins visited the pricing page before the first meeting and only a small share of leads did the same, that visit deserves a high weight. If the ebook download shows up in the same proportion among won and lost deals, it is worth little, however much marketing likes the piece. With that data on the table, each team defends the weight it proposes with a number. If the CRM does not store lead source, the first version uses the judgment of the most experienced SDRs as a hypothesis, with a date set to check it against data.

Which lead scoring criteria to use: fit and interest

I work with two separate axes. Fit measures how closely the contact and their company resemble the customer your company serves well, and interest measures how strongly recent behavior points to a project in motion. With the axes kept apart, the team can read the score: high fit with low interest calls for nurturing and, in priority accounts, active prospecting, and high interest with low fit calls for screening before it takes up a rep’s time. In a single combined score, the student who read twenty articles ties with the director of an ideal account who asked for a demo.

Fit: the match with the ICP

The fit axis comes from the ICP, the ideal customer profile that marketing, sales and CS approved: industry, company size, region served and job title. The ICP exclusion criteria become strong negative points or a block, because no behavior makes up for an account the company decided not to serve. Job title weighs according to the role in the buying committee, and the manager who sponsors this kind of project is worth more than the assistant researching for the boss.

Interest: the behavior that comes before a purchase

A request for a demo, a quote or a sales contact is a declaration of interest and sends the lead straight to the SDR, whatever the score. Repeated visits to pricing, customer stories and integration pages signal evaluation. Reading articles and downloading top-of-funnel content signal study of the topic and are worth little on their own, while webinars and email replies sit in the middle, with whatever weight the history confirms.

Negative points

Negative points take out of the queue what the sum of behaviors would put there by mistake: competitors, vendors, students, candidates who visited the careers page and current customers, who go to post-sale. A personal email address, when the sale requires a business contact, and newsletter unsubscribes also subtract points. The rejection reasons sales records point to the negatives still missing.

Score decay over time

Without decay, the score only grows, and a lead crosses the MQL threshold after years of collecting points without having done anything recent. I recommend that interest points lose half their value every 30 days without activity, or expire between 60 and 90 days, depending on the buying cycle. Fit does not decay, and keeping size, industry and job title current is the work of CRM data quality.

Lead scoring criteria template

The table is an illustrative example, with hypothetical points for a company that sells a B2B service to mid-sized businesses, on a 0 to 100 scale on each axis. It shows the logic of each row, and your model takes its weights from your own history.

Criterion	Axis	Points	Rationale
Industry within the ICP	Fit	+25	Industries that close faster and renew more
Company size within the ICP range	Fit	+25	Range where deal size and retention hold up
Management role in the area that uses the solution	Fit	+20	Role that sponsors the project in most wins
Region served	Fit	+10	Outside it, implementation and service get expensive
Competitor, vendor or student	Fit	-100	Exclusion: no behavior makes up for it
Demo or quote request	Interest	+50	Declaration of interest; goes to the SDR even below the threshold
Two or more pricing page visits in 30 days	Interest	+20	Evaluation signal present in most wins
Visit to customer stories or integrations	Interest	+10	Buyer checking whether the solution fits their context
Top-of-funnel content download	Interest	+5	Shows up equally among won and lost deals
Careers page visit	Interest	-30	Candidate, no buying project
30 days without activity	Interest	Half the points	Old interest is worth less

Email opens were left out on purpose, for the reason the mistakes section explains. Apart from the direct request, which follows its own rule, no single criterion takes a lead to MQL: fit needs to be in range and interest needs to be recent.

Scoring bands: cold lead, MQL and SQL

The score changes the operation when each band has a destination, an owner and a deadline. The thresholds below follow the same hypothetical example. The SQL band does not depend on the score, because a lead only reaches it after the SDR confirms pain, context and decision maker in a conversation.

Band	Entry (example)	What happens	Who acts	Response time
Cold lead	Fit below 40 or interest below 20	Content nurturing	Marketing	No sales deadline
Warming up	Fit from 40 to 59 with some interest, or high fit with no recent interest	Segmented nurturing; active prospecting in priority accounts	Marketing and SDR	Light cadence, within the week
MQL	Fit from 60 and interest from 40	Automatic handoff to the SDR, with the behavior history visible in the record	SDR	First attempt on the same business day
Direct request	Demo, quote or contact request, with no exclusion criterion	Jumps the queue, with an alert to the SDR and the manager	SDR	Within one hour, during business hours
SQL	Conversation with pain, context and decision maker confirmed	Meeting with the account executive and an opportunity in the pipeline	Account executive	Meeting within the SLA deadline

The short deadline for direct requests is grounded in research. In a study published in Harvard Business Review in 2011, covering 1.25 million leads from 42 US companies, those that tried to make contact within an hour were nearly seven times as likely to reach a meaningful conversation with a decision maker as those that tried an hour later, and more than 60 times as likely as those that waited 24 hours or longer. The study is fifteen years old and comes from another market, and I use the result as a direction. The deadlines for each band go into the SLA between marketing, sales and post-sale, and a lead that misses its deadline shows up on the manager’s dashboard.

How do you calibrate lead scoring every quarter?

Once a quarter, the operations person pulls from the CRM the leads that entered each band and tracks how many became SQLs, opportunities and wins, over a window that matches the sales cycle. In a model that works, conversion grows clearly from one band to the next. In a hypothetical example, if MQLs convert to SQL at 25% and warming leads convert at 20%, the threshold separates very little, and the SDR spends time on a queue that belongs in nurturing.

The second reading starts from the wins: which band each won deal was in when it became an opportunity. Many wins below the threshold point to a signal the model does not score yet, such as a partner referral. With fewer than twenty or thirty wins in the period, widen the window before changing weights.

What leadership sees	Likely cause	Adjustment
MQL converts only slightly above warming leads	High weight on top-of-funnel actions	Raise the threshold or lower the content weight
Many wins below the threshold	Buying signal outside the model	Add the signal
Rising MQL rejections for fit	New source outside the ICP	Review negatives and sources
Queue larger than SDR capacity	Threshold too low	Raise the threshold
Empty queue	Threshold too high or decay too fast	Lower the threshold and check the window

Conversion by band goes into the RevOps metrics matrix, next to the MQL acceptance rate and the time to first attempt. Change only a few weights per review and log each change, or nobody will be able to say which adjustment improved the queue.

Who owns the lead scoring model?

The model owner is someone with no volume target, usually the person in charge of revenue operations or of the CRM, who configures the rules, documents each version and presents conversion by band at the review. Marketing and sales leaders approve each change of criterion or threshold together, and post-sale flags the signals of current customers. If marketing can lower the threshold alone to hit its MQL target, the queue fills with leads sales rejects. Marketing answers for the conversion of the MQL band and for nurturing the bands below it. Sales leadership sets the deadline and attempts per band, requires a rejection reason and brings to the review the cases where the score got it wrong.

What are the most common lead scoring mistakes?

Scoring email opens

Some email apps load message images automatically to protect user privacy, and the system logs as an open a message nobody read. Scoring opens inflates the score of people who only receive the newsletter. If you want to use email, score clicks on evaluation links, such as pricing and customer stories, and check in your history whether they separate won deals from lost ones.

Never reviewing the model

The company launches a product or changes its website, and the criteria keep rewarding the behavior of two years ago. The quarterly review only happens with a date on the calendar and an owner.

Sales ignoring the score

When sales ignores the score, the cause usually lies in the model or in the lack of consequences. If the queue has already disappointed the rep, leadership fixes it through calibration before demanding discipline. If the score is good and nobody uses it, what is missing is automatic routing by band and a task with a deadline, plus a record that shows why the lead scored, so the SDR opens the conversation with context.

What about artificial intelligence in lead scoring?

Predictive scoring uses machine learning to estimate, from history, the likelihood that each lead becomes a customer, and several CRM platforms already offer the feature. It finds combinations the manual table misses, such as the sequence of pages visited, as long as the CRM holds enough won and lost deals, a filled-in lead source and reliable stage dates.

I recommend AI once the simple model has run for two or three quarters with reviews, when the data is organized and the team trusts the score. Predictive scoring comes in as a second opinion, compared with the manual model in the same review. If it discards leads without anyone looking, the company documents the criterion and provides for human review, as I explain in AI and LGPD, Brazil’s data protection law.

Roadmap for leadership to implement lead scoring

1. Pull the baseline. MQL acceptance rate, conversion from MQL to SQL and to won, time to first attempt and rejection reasons.
2. Compare wins and leads from the last 12 to 18 months. List the attributes and behaviors that show up more often among wins than among all leads.
3. Write the criteria on both axes. Fit from the ICP, interest from history, negatives from the exclusions and rejections, and the decay rule.
4. Define the bands. Threshold, owner and deadline for each one, with the direct request kept apart from the points total.
5. Take it all to the SLA and the CRM. Marketing and sales approve the text, and the process owner configures scoring, routing, tasks and a dashboard by band.
6. Train the team before turning the model on. SDRs, reps and marketing analysts learn to read the score and the lead’s history and to log rejections with a reason.
7. Review every quarter. Conversion by band, wins below the threshold and rejections by reason, with a few adjustments at a time.

Lead scoring and L&D: what HR needs to assess

How to recognize the need

SDRs who pick leads by company name and a marketing team celebrating MQLs that sales rejects point to a model with no agreement between the teams, or to a team that never learned to use it. Before treating the case as individual resistance, ask the leaders who approved the criteria and when the last review took place.

Who takes part

The core is marketing and sales leadership, with the sales development manager, the SDRs and the analysts who run the queue and nurturing, plus the operations or CRM person who configures the model. The model only takes hold when marketing and sales learn together to use and review the score, which is why, in in-company training, I put both teams in the same room to calibrate the criteria on the company’s own leads and deals.

How to measure

Record before the program the MQL acceptance rate, the time to first attempt and the conversion from MQL to SQL, and compare them at the first quarterly review. Conversion by band all the way to won deals answers after at least one average sales cycle.

If your company has scoring switched on in the CRM and the team still picks leads by gut feeling, what is missing is agreement between the teams and training. In the in-company Revenue Operations training, I work with the marketing, sales and post-sale leaders on your operation’s history, and the group leaves with the SLA between marketing, sales and CS in writing, including the lead scoring criteria and the deadline for each band, and with the RevOps metrics matrix that tracks conversion by band in the CRM you already use.

Frequently asked questions

What is lead scoring?

Lead scoring is the model that gives each contact a score based on fit with the ideal customer profile and on the interest shown in their behavior. The score sets the order of the SDR queue and the moment the lead becomes an MQL and moves to sales.

What is the difference between MQL and SQL in lead scoring?

An MQL is a lead that reached the score band agreed between marketing and sales, with enough fit and recent interest. An SQL is a lead the SDR confirmed in a conversation, with pain, context and decision maker. The score takes a lead to MQL, and only the conversation takes it to SQL.

Which lead scoring criteria should B2B sales use?

Use two axes. On fit, industry, company size, job title and region from the ICP; on interest, contact requests, visits to pricing and customer story pages and event attendance. Add negative points for competitors, students and current customers, and a decay rule for old interest.

How often should you review the lead scoring model?

Every quarter, comparing each band’s conversion to SQL and won and checking which band the won deals were in. A product launch, entry into a new industry or a website change calls for a review outside the calendar.

Is it worth using artificial intelligence in lead scoring?

Yes, once the simple model has worked for a few quarters and the CRM holds reliable wins, losses, sources and dates. Before that, the predictive model learns from incomplete data. When the score decides discards without human review, document the criterion to comply with data protection law.

How do you train marketing and sales to use lead scoring?

Train both teams together, with the company’s own leads and won deals. SDRs learn to read the score and the history before the call, marketing analysts learn to read conversion by band, and leaders practice the quarterly review. Measure the MQL acceptance rate and the time to first attempt before and after the program.