Lead Scoring & Routing: How to Drive Real Sales & Marketing Alignment

Sales and marketing alignment: the lead scoring gap most teams miss

Sales and marketing alignment is one of the most cited challenges in go-to-market strategy, and one of the least operationalized. LinkedIn research found that for the vast majority of sales and marketing professionals, better alignment is their single largest area of opportunity. Just as many sales and marketing leaders believe greater alignment is critical for sustained business growth.

The problem is not a lack of ambition. It is a lack of operational infrastructure. Teams pursue alignment as a concept rather than building the scoring and routing systems that make it real.

ZoomInfo, an all-in-one AI GTM Platform, offers a real-world example of what can happen when sales and marketing teams build scoring and routing systems that actually work together. This article covers both the methodology (what lead scoring is and how to build a model) and the real-world implementation (how ZoomInfo does it), so you can take the framework back to your own stack.

What is lead scoring?

Lead scoring is a method sales and marketing teams use to rank prospects by their likelihood to convert, assigning point values to behaviors, firmographic attributes, and engagement signals. In B2B lead scoring, the goal is to surface the highest-value accounts and contacts so sales reps spend time on opportunities most likely to close, not on leads that are not ready.

The practical outcome is a prioritized queue: reps work the leads most likely to convert first, and marketing routes the rest into nurture until intent signals strengthen.

Why does this matter? Sales reps spend a significant portion of their time on prospecting and prioritization activities rather than active selling. According to Salesforce State of Sales research, that proportion is substantial enough that improving prioritization has an outsized effect on revenue productivity. A well-calibrated lead scoring model functions as a seller productivity lever and a marketing ROI tool simultaneously: marketing gets credit for MQLs that actually convert, and sales gets a queue they can trust. Without a scoring model, both teams are flying on instinct, and the handoff between them becomes a recurring source of friction.

A lead scoring model formalizes that handoff. It defines, in measurable terms, what a qualified lead looks like and when it is ready for a rep. The next section covers what goes into that model.

Explicit vs. implicit lead scoring attributes

Every lead scoring model draws on two categories of input: what a prospect tells you and what a prospect shows you.

Explicit attributes are firmographic and demographic data a prospect provides directly, such as job title, company size, industry, geography, or the technologies they use. Implicit attributes are behavioral signals inferred from prospect actions: pricing page visits, demo requests, content downloads, email opens, repeat site visits, and intent data spikes from third-party sources.

The most effective B2B lead scoring models combine both dimensions. Fit without intent misses timing: a perfect-ICP account that shows no engagement signals is not ready for a sales conversation. Intent without fit wastes rep capacity: a highly engaged visitor from a company that will never buy still costs the same follow-up time as a real opportunity.

How to build a lead scoring model in 6 steps

Building a reliable lead scoring model is not a one-afternoon exercise, but it does not require a data science team. The process below is designed for RevOps practitioners who own the CRM and the routing logic.

Step 1: Define your ICP criteria

Start with the firmographic and technographic filters that qualify an account as in-profile. This is the foundation of your explicit scoring layer. Company size, industry, geography, and technology stack are the most common dimensions, but the right criteria are specific to your win data.

"Every business tends to know the types of companies they are able to sell to easiest, and the people who buy from them most often," says Will Frattini, head of enterprise growth strategy at ZoomInfo. "We're no different, and by understanding data attributes about the companies and people who tend to sign off with us, we were able to build out a clear model on what would be our 'hottest' leads."

Pull your last 12 months of closed-won deals and look for patterns: which industries, headcount bands, and titles appear most often? That cluster is your ICP. Define it in measurable terms so it can be encoded into scoring rules.

Step 2: Identify behavioral signals

Map the actions that indicate purchase intent at each funnel stage. Not all behavioral signals carry the same weight. A pricing page visit signals active evaluation; a blog post read signals early research. A demo request is a direct buying signal; an email open is a soft engagement indicator.

Work with your marketing team to audit the touchpoints in your MAP and CRM. List every trackable action and assign it a rough intent tier (high, medium, low) before you assign point values. This prevents the common mistake of treating all engagement as equivalent.

Step 3: Assign point weights

Calibrate weights against historical win-rate data, not intuition. Pull your closed-won and closed-lost data and look at which attributes and behaviors were present in deals that closed versus those that did not. The attributes that appear disproportionately in closed-won deals earn higher weights.

ZoomInfo's scoring model evaluates win-rate projection, channel, and segment on a trailing 90-day basis for every individual salesperson. "What's our projected win rate? Where did this lead come from? Based on historical data, what are the better performing channels?" Frattini says. "Segment typically determines whether a lead goes to a commercial sales rep or a corporate rep. This is what's scored on a trailing 90-day basis, for every individual salesperson on our team."

Use the sample scoring matrix below as a starting point and calibrate the weights against your own win data:

Step 4: Add negative scoring

Negative scoring is one of the most underused levers in lead scoring model design. Subtracting points for disqualifying signals keeps your MQL queue clean and prevents reps from wasting time on leads that look engaged but will never convert.

Common negative signals include: unsubscribes, competitor domain email addresses, student or personal email addresses, and extended inactivity windows. Set a floor score (e.g., any lead below 0 points enters a suppression list rather than a nurture program) to prevent re-routing of clearly disqualified leads.

Step 5: Set your MQL threshold

The MQL threshold is the score that triggers a sales handoff. This number must be agreed upon jointly by sales and marketing, not set unilaterally by marketing operations. If sales does not trust the threshold, they will ignore the queue.

A typical starting range is 50–100 points, calibrated against your historical data. Set the threshold too low and you flood reps with unqualified leads; set it too high and you starve the pipeline. The threshold should also inform your lead routing and scoring logic: leads above the threshold route to frontline reps, leads below route to nurture or SDR sequences.

Step 6: Test, calibrate, and review

Run your model against a trailing 90-day dataset before going live. Compare the leads your model would have flagged as MQLs against the leads that actually converted during that period. Adjust weights where the model diverges from reality.

Commit to a quarterly review cadence. Win-rate patterns shift as your market evolves, your ICP expands, and your product changes. A scoring model calibrated on last year's data will degrade without regular recalibration.

That methodology only holds up in practice when the routing infrastructure behind it can move as fast as the model demands. Here is how ZoomInfo applies it.

Not all leads are created equal

One of the greatest challenges facing any sales leader is balancing the quality of incoming leads with their sales team's ability to execute on and close them.

During a company's crucial early growth periods, many sales leaders route their strongest leads to their best salespeople. At face value, the logic makes sense: strong leads mean greater revenue potential, and skilled, experienced reps are more likely to close those deals.

The situation becomes more complicated, however, when you factor in new hires and ramp times. Not every salesperson is of equal ability, and less-experienced reps still need sufficient opportunities to develop their skills, not to mention the quotas they're expected to attain.

The solution, for us, was to rethink how the warmest leads were routed. How do you determine which reps should be assigned which leads?

Form fills: Behind the scenes

Like many companies with a well-developed sales motion, ZoomInfo relies on form submissions from our website as a vital source of leads.

In line with industry best practices, we ask for as little information as possible from our site visitors. Fewer form fields means a higher likely conversion rate, and much of the data we need to route those leads effectively is derived after the fact by cross-referencing submitted information with data in ZoomInfo's platform.

Although ZoomInfo's proprietary data enriches form submissions with more than 200 additional data points, we typically focus on just half a dozen or so when determining how that lead should be routed.

"Every business tends to know the types of companies they are able to sell to easiest, and the people who buy from them most often," says Will Frattini, head of enterprise growth strategy at ZoomInfo. "We're no different, and by understanding data attributes about the companies and people who tend to sign off with us, we were able to build out a clear model on what would be our 'hottest' leads.

"We started to see a trend that our sellers would win or lose more often with businesses of a certain size, certain firmographic, industry, and so on, as well as certain titles and buyer personas. Once we got that profile dialed in, we took it a step deeper and saw that certain salespeople on our team are better equipped and able and willing to close business faster with certain personas."

With this data in hand, we set out to evaluate which leads were closing at higher rates, and referenced that with how we assess our reps' performance.

Playing to reps' strengths

Before we examine the performance of specific salespeople, we evaluate the strength of an inbound lead via our website, specifically, requests for demos, by three criteria:

1. Win-rate projection

2. Channel

3. Segment

Some leads inevitably score higher than others. Businesses within certain industries, of a certain headcount, with certain technologies in their stacks are significantly more valuable and are therefore scored higher.

ZoomInfo's scoring model evaluates each of these three criteria on a trailing 90-day basis for every individual salesperson, using win-rate projection to estimate close likelihood, channel to identify which lead sources have historically performed best, and segment to determine whether a lead routes to a commercial rep or a corporate rep. This rigorous approach ensures that only quality leads reach frontline teams and that reps are assigned leads fairly based on their performance.

According to ZoomInfo's internal lead routing data, only 32–34% of form-fill submissions are routed to frontline sales teams. The remaining 66–68% of those submissions are routed elsewhere, such as nurture programs, as they don't yet meet our qualification criteria.

This offers a high degree of confidence in the quality of incoming leads for our sales teams, but it also poses unique challenges for sales leaders and their reps.

"If I'm a sales rep, I'm scored on my three-month dollars booked, relative to how many meetings I said were a good fit," Frattini says. "So if I complete a bunch of meetings from marketing and I'm getting one of the 32% of 'hot' leads that were routed, and I say it was a good meeting, if I don't close that business, that's a significant missed opportunity.

"In the same vein, if I don't say it was a good meeting, my leadership wants to ensure that lead gets routed again to another sales person to ensure we don't miss out on a win for the business."

From click to call in 90 seconds

To ensure the right leads are routed to the best reps at the correct moment, incoming leads are enriched and scored across several different criteria at the moment data is submitted via a form.

However, that is only the beginning. Once leads have been scored and routed appropriately, and requests for a demo identified, those leads are then scored once again to ensure they are routed to the reps with a keen understanding of the prospect's industry.

All this happens within just 90 seconds.

"Somebody could be on our website, filling out a form, and in a half-hour, they've already moved to a different meeting," says Deeksha Taneja, ZoomInfo's vice president of growth and optimization. "After 30 minutes, the likelihood of booking a demo with them drops by roughly half. If you don't get to them that same day, your chances drop to around 20%."

ZoomInfo's internal conversion data shows that after 30 minutes, the likelihood of booking a demo drops by roughly half; by end of day, chances fall to around 20%. Momentive cut speed-to-lead from 20 minutes to 60 seconds using ZoomInfo Operations, proving the 90-second benchmark is achievable in practice. That result is driven by the same enrichment-and-routing architecture described throughout this article: the GTM Context Graph processes 1.5B+ data points daily, fusing firmographic, behavioral, and intent signals to surface not just what a lead looks like, but why certain profiles convert.

While timing is the most important factor in successfully booking demos, it is far from the only one. Other signals, such as the channel of incoming leads, tell us a great deal about our prospects. Inbound leads from pricing pages demonstrate significantly higher intent than a request for more information, and are scored and routed accordingly.

Leads that do not clear the routing threshold are not discarded. This is the group typically routed into ongoing outreach and nurture programs, monitored over time to identify spikes in intent that could indicate a stronger likelihood of conversion.

"You can think of intent as a cumulative score," Taneja says. "A lead may have lower scores in terms of firmographics or demographics, but if intent is getting higher, even though the overall score is lower, that's when we might consider putting that lead in front of an SDR."

As complex as lead routing and scoring can be, it is just one side of the equation. Even the most carefully vetted leads are of little use if frontline sales teams lack the resources to work them. Our sales and marketing teams work closely and meet frequently to ensure that individual reps have sufficient capacity to work and close leads effectively.

For most RevOps teams, the bottleneck is not the scoring logic, it is the engineering cycle required to rebuild routing flows every time the model changes. GTM Studio's codeless interface lets operations teams configure enrichment, scoring thresholds, and routing rules in a single workflow, without writing SOQL queries or opening a change management ticket. When the scoring model needs recalibration, the change ships the same day.

AI and predictive lead scoring

Predictive lead scoring uses machine learning models trained on historical win/loss data to assign conversion probability scores automatically, without manually defined rules.

Traditional rule-based scoring is transparent and controllable: you define the criteria, assign the weights, and can audit exactly why any lead received its score. The tradeoff is that rule-based models require manual recalibration as market conditions change. Predictive models improve over time as they ingest more outcome data, but they require sufficient data volume to train reliably. A common threshold is 1,000 or more closed opportunities in your CRM before a predictive model produces reliable signals. Below that threshold, a well-calibrated rule-based model will outperform a predictive one because the training data is too thin to generalize.

ZoomInfo enables a form of predictive scoring by reasoning across 1.5B+ daily data points, fusing CRM signals, behavioral data, intent data, and conversation intelligence to surface accounts most likely to convert, not just accounts that fit a static profile. This is the capability Frattini describes when he talks about understanding which companies and personas close fastest: the human-validated ICP criteria become the foundation that the intelligence layer builds on, continuously refreshed against live signals rather than a trailing snapshot.

The practical guidance for teams evaluating which approach to use:

• Rule-based scoring is the right starting point for teams with fewer than 500 closed deals in their CRM. It is auditable, adjustable, and does not require clean historical data at scale.

• Predictive scoring becomes viable when you have 1,000 or more closed opportunities and a CRM foundation that is reasonably complete. Without clean underlying data, predictive models inherit the same gaps and produce unreliable scores.

For most RevOps teams, the answer is a hybrid: rule-based criteria for explicit fit signals, and AI lead scoring layered on top for intent and behavioral pattern recognition. ZoomInfo operates in that hybrid mode, grounding its reasoning in verified firmographic and technographic data while continuously updating behavioral and intent signals.

Sales and marketing alignment: from concept to process

One of the problems facing sales and marketing leaders is that there is no one-size-fits-all way to achieve better alignment between sales and marketing teams. Even two companies of similar headcount in the same industry may have significantly different needs and obstacles.

As a result, many GTM leaders end up pursuing "alignment" as an aspirational concept rather than viewing it through the lens of actionable processes. This can result in misguided internal initiatives, wasted spend, and frustrated sales and marketing teams.

"A lot of businesses overestimate how sophisticated their alignment strategy needs to be," Frattini says. "What made our early model so successful is that it was scalable."

Platforms like ZoomInfo apply reasoning directly to CRM and enrichment data to build a scoring propensity model grounded in actual win patterns, compressing the time between enrichment and routing decision without requiring a custom engineering build.

Our sales leaders have been refining this lead scoring model over the past eight years or so with great success. The process may be faster today, but the core data points we look for and the criteria we use to evaluate our salespeople have remained virtually unchanged as this system has rolled out across our sales organization.

Sales and marketing alignment is often spoken of as a goal unto itself. In fact, true alignment is nothing more than identifying and implementing predictable, replicable systems that ensure business goals are being met.

"Sales and marketing alignment strives for better outcomes, but it's a little ethereal to say, 'Let's align these two disciplines,'" Frattini says. "It was much more tangible for us to identify the fact that almost 70% of form submissions didn't meet our criteria to be a good-fit account or person yet, and would have been more expensive for us to put those leads in front of an SDR or salesperson until they were more qualified or ready."

Alignment SLA framework

Alignment requires operational agreements, not just shared goals. Four commitments that turn sales marketing alignment from a concept into a process:

• Joint MQL definition: Sales and marketing agree on the score threshold that triggers a handoff. This number is not set by marketing operations alone. If sales does not own the definition, they will not trust the queue.

• Follow-up SLA: Sales commits to contacting routed MQLs within a defined window. The 90-second benchmark described earlier in this article is achievable with automated routing, and the data on demo booking drop-off after 30 minutes makes the business case for urgency clear.

• Feedback loop: Sales flags mis-scored leads so marketing can recalibrate the model. Without this loop, scoring models drift from reality as market conditions change. A simple Salesforce field ("lead quality: good fit / not a fit / too early") gives marketing the signal it needs to adjust weights.

• Quarterly review cadence: Both teams review model performance against trailing 90-day win data. Scoring criteria that predicted conversion last year may not predict it today. The review cadence is what keeps the model honest.

How ZoomInfo powers lead scoring and routing at scale

ZoomInfo's all-in-one AI GTM Platform brings together the data, intelligence, and workflow automation that make lead scoring and routing a repeatable system rather than a manual process.

ZoomInfo's B2B data foundation covers 500M contacts, 135M+ verified phone numbers, 200M+ verified business emails, and 30,000+ technologies tracked across 200+ categories. This means the firmographic and technographic inputs that drive explicit scoring criteria are accurate and continuously refreshed, not stale snapshots inherited from a batch append six months ago. The result is a scoring foundation that reflects current account reality rather than the state of your CRM at last year's territory planning cycle. See how Snowflake saw 2x conversion on ZoomInfo-scored accounts, along with 90% higher opportunity open rates, by building their scoring model on verified firmographic data.

The GTM Context Graph processes 1.5B+ data points daily, fusing CRM data, behavioral signals, intent data, and conversation intelligence into a unified reasoning layer. This is what enables predictive scoring that goes beyond static fit criteria: the Graph surfaces why certain accounts convert, not just which ones match the ICP profile. The intelligence layer continuously updates as new signals arrive, so a lead that was not ready last quarter can resurface when intent spikes. Smartsheet's 84% MQL increase after implementing ZoomInfo's scoring and enrichment, along with a 26% improvement in opportunity rate, reflects what happens when the intelligence layer connects firmographic fit with behavioral and intent signals rather than treating them as separate inputs.

GTM Studio gives RevOps teams a codeless interface to configure enrichment, scoring, and routing workflows without engineering tickets. The same routing logic that ZoomInfo uses internally, the 90-second enrichment-to-routing cycle described throughout this article, is configurable for your team without SOQL queries or change management cycles. Momentive compressed speed-to-lead from 20 minutes to 60 seconds using ZoomInfo Operations, demonstrating that the 90-second benchmark is not a ZoomInfo-specific outcome but a repeatable result of the right enrichment and routing architecture.

See how ZoomInfo's AI GTM Platform can compress your lead routing cycle and improve scoring accuracy, request a demo.

Frequently asked questions about lead scoring

What is the lead scoring process?

Lead scoring follows six steps: define ICP criteria, identify behavioral signals, assign point weights, add negative scoring for disqualifying signals, set an MQL threshold with sales, and calibrate the model quarterly against win-rate data. The goal is a repeatable system that routes the right leads to the right reps at the right time. For the handoff step, see how lead routing and scoring works as the operational layer that executes on the model's output.

What is an example of lead scoring?

ZoomInfo's own model evaluates three primary criteria: win-rate projection (historical close rates for this account profile), channel (where the lead originated, since pricing page leads score higher than general information requests), and segment (company size and firmographic fit). According to ZoomInfo's internal lead routing data, only 32–34% of form-fill submissions score high enough to route to frontline sales reps; the rest enter nurture programs until intent signals strengthen.

How do you calculate a lead score?

A lead score is calculated as a weighted sum of individual criteria: add positive point values to qualifying signals (pricing page visit = +15, VP+ title = +20, demo request = +25) and subtract points for disqualifying signals (unsubscribe = -10, competitor domain = -20). Set an MQL threshold, typically 50–100 points, and calibrate the weights against your historical win-rate data rather than intuition. The scoring matrix in the model-build section above provides a starting template with rationale for each weight.

What is predictive lead scoring and how does it differ from rule-based scoring?

Predictive lead scoring uses machine learning trained on historical win/loss data to assign conversion probability scores automatically, without manually defined rules. Rule-based scoring is transparent and controllable but requires manual recalibration; predictive scoring improves over time but needs sufficient data volume, typically 1,000 or more closed opportunities, to train reliably. Platforms like ZoomInfo combine both approaches through the GTM Context Graph, which reasons across 1.5B+ daily data points to surface accounts most likely to convert.

How do you align sales and marketing on lead scoring?

Alignment requires four operational agreements: a joint MQL definition (the score threshold that triggers a sales handoff), a follow-up SLA (how quickly sales contacts a routed lead, with the 90-second benchmark achievable through automated routing), a feedback loop (sales flags mis-scored leads so marketing can recalibrate the model), and a quarterly review cadence (both teams review model performance against trailing 90-day win data). Without these agreements, lead scoring becomes a marketing metric that sales ignores. See how Momentive cut speed-to-lead from 20 minutes to 60 seconds to understand what automated routing makes possible on the follow-up SLA front.