Building a Marketo Scoring Model That Sales Actually Trusts

Why Sales Stops Using the Scoring Model

The conversation is always the same. Marketing presents a new lead scoring model. Sales agrees to work the MQLs. Three months later, sales has stopped following up on scored leads because "they're never ready" or "they don't know what we do." Marketing blames sales for not working the leads. Sales blames marketing for sending bad leads. The scoring model sits unused and the relationship deteriorates.

The root cause is almost never the technology. It is the design philosophy. Most scoring models are built from the marketing side — they reward behaviors that are easy to track in Marketo: email opens, page views, content downloads, webinar registrations. These are engagement signals. They are not buying intent signals. A prospect who has opened every email you have ever sent but has no budget authority and no active project is a high-scoring lead that sales correctly identifies as a waste of time.

A scoring model sales will trust starts with a different question: what does a contact do in the 90 days before they become a customer? The answer to that question — drawn from historical closed-won data — is the foundation of a model that predicts conversion rather than measuring engagement.

The Framework: Starting with Closed-Won Data

Before building any scoring rules, pull a sample of 50-100 closed-won opportunities from the prior 12 months. For each opportunity, look at the contact's Marketo activity log in the 90 days before the deal closed. Document every meaningful activity — content downloads, page visits, email clicks, form fills, event registrations — and look for patterns.

What you will typically find is that the activities that correlate with conversion are not the ones that received the most points in the existing model. Pricing page visits, demo page visits, case study downloads, and ROI calculator interactions consistently appear in the closed-won activity logs. Email opens and generic blog post views consistently do not.

• High-intent behavioral signals — pricing page visit (+25), demo request page visit (+20), ROI calculator interaction (+20), case study download (+15), competitive comparison content (+15), contact us page visit (+10)
• Medium-intent behavioral signals — product page visit (+8), feature-specific content download (+8), webinar registration with attendance (+10), webinar registration without attendance (+3)
• Low-intent behavioral signals — blog post view (+2), generic email click (+2), email open (+1). These are noise signals — they tell you the contact is alive but not that they are buying.
• Negative signals — unsubscribe (-50), bounce (-20), competitor company domain (-30), student or academic email (-20). Negative scoring is essential and almost always missing from first-generation models.

Demographic Scoring: Fit Before Behavior

Behavioral scoring measures what a contact does. Demographic scoring measures whether they are worth engaging at all. A contact at a 10-person startup who visits your pricing page should score lower than a VP at a 5,000-person enterprise who visits the same page — because the enterprise VP has the budget, authority, and organizational complexity that makes them a genuine prospect.

• Job title/seniority — VP and above (+20), Director (+15), Manager (+10), Individual contributor (+5), Student/intern (-20)
• Company size — 1,000+ employees (+20), 500-1,000 (+15), 100-500 (+10), under 100 (+5 or 0 depending on your ICP)
• Industry fit — industries in your ICP (+15), adjacent industries (+5), out-of-ICP industries (0 or -10)
• Geography — territories you actively sell into (+10), territories you cannot support (-20)

Getting Sales Buy-In Before Launch

The most important step in building a scoring model sales will trust is involving sales before the model is built, not after. Schedule a 60-minute working session with your top 3-5 sales reps — the ones with the best close rates — and ask them one question: what does a good lead look like to you, specifically?

Record their answers. You will hear things like: "They've looked at our pricing," "They mentioned a specific competitor," "They asked about implementation timeline," "They had a budget number in mind." These are the behavioral signals your model should prioritize. If those signals are not in your Marketo activity data, you need to create the touchpoints that generate them — a pricing page, an ROI calculator, a competitive comparison page — before the scoring model will work.

Maintenance: The Model That Decays

A scoring model is not a one-time build. Lead behavior patterns change as your product evolves, your buyer profile shifts, and your content strategy changes. A model built on 2022 closed-won data will be less accurate in 2024 — not because it was wrong, but because the signals have changed.

Establish a quarterly review process: pull closed-won data from the prior quarter, compare the pre-conversion activity patterns to your current scoring rules, and adjust weights where patterns have shifted. Document every change with the rationale. Share the review findings with sales so they see the model is actively maintained based on their conversion data — not a static artifact from a previous marketing ops manager.

ZSavvy's Scoring Engine for Executive Engagement

ZSavvy's scoring engine applies the same principles — closed-won data analysis, intent-weighted behavioral signals, negative scoring — but extends them to the executive engagement context. For nomination decisions, the relevant signals are different: account penetration depth, deal stage in Salesforce, relationship history with the executive, and the strategic value of strengthening that specific account relationship.

The approval probability prediction in ZSavvy's platform gives nomination reviewers a data-informed signal about which nominations are most likely to convert to attendance and which accounts are most likely to progress as a result of executive engagement — combining the scoring model principles used in Marketo with the specific data inputs that drive executive engagement program outcomes.