[GTM glossary](/glossary.txt) Lead scoring models Sales Development & Prospecting # Lead scoring models In short Lead scoring models are the specific designs used to turn lead data into a score. The main types are manual rule-based, predictive, and two-dimensional fit and intent models, and each suits a different team size and data volume. This page compares the models. For the overall practice of ranking leads, see lead scoring. ## Main types of model ### Manual, rule-based Someone assigns points by hand to traits and actions. It is transparent and quick to start, and works well with small data. It relies on the author's assumptions, so it needs regular review. ### Predictive A statistical or machine learning model learns from past won and lost leads and outputs a probability. It can find patterns people miss, but needs enough historical outcomes and is harder to explain to reps. ### Fit and intent grid Instead of one number, leads are placed on two axes: how well they match the target customer, and how much interest they show. Each cell maps to an action, such as "call now" or "nurture". ### Account-level Scores roll up all contacts at a company, which suits team-based buying where no single lead tells the full story. --- ## How to choose a model * **Little data, new motion:** Start manual and keep it simple. * **Lots of closed deals:** Test a predictive model against your manual one. * **Long, multi-stakeholder deals:** Consider account-level scoring. * **Need clear actions:** Use a grid so each score implies a next step. --- ## Example Say a team has a manual model where a demo request earns 30 points. It then trains a predictive model on a year of outcomes and finds that visits to an integrations page predict wins better than ebook downloads. The team adjusts the manual weights to reflect that and keeps both models running for comparison. --- ## Common mistakes * **Choosing complexity first:** A black-box model with thin data is worse than a clear manual one. * **No holdout test:** Without comparing against a baseline you cannot tell if the model helps. * **Training on biased history:** If reps only called certain leads, the model learns their habits, not buyer behavior. * **No explanation for reps:** Show the top reasons behind a score to build trust. --- ## Lead scoring models FAQs **What are the types of lead scoring models?** The main types are manual rule-based models, predictive models built on past outcomes, fit and intent grids that rate two dimensions separately, and account-level models that combine all contacts at a company into a single score. **What is the best lead scoring model?** It depends on your data and sales motion. Teams with little history usually start with a simple manual model. Teams with many closed deals can test a predictive model. Compare any model against a baseline using real outcomes. **What is the difference between predictive and manual lead scoring models?** A manual model uses points that people assign based on judgment, so it is transparent but assumption-driven. A predictive model learns patterns from historical won and lost leads, so it can be more accurate but needs enough data and is harder to explain. **How do you build a lead scoring model?** Analyze past customers to find shared traits and behaviors, choose a model type, assign weights or train it, set a handoff threshold, then validate the scores against outcomes. Revisit the model on a regular schedule. Share term --- ## More related phrases * [Predictive scoring](/glossary/predictive-scoring.txt) * [Propensity model](/glossary/propensity-model.txt) * [Fit score](/glossary/fit-score.txt) ## Similar phrases * [Predictive lead scoring](/glossary/predictive-lead-scoring.txt) Predictive lead scoring uses a model trained on past deals to rank the leads in your database by how likely each one is to become a customer. * [Lead scoring](/glossary/lead-scoring.txt) Lead scoring is a method of ranking leads with points for fit and behavior, so sales can focus first on the people most likely to become customers. * [AI lead scoring](/glossary/ai-lead-scoring.txt) AI lead scoring uses machine learning to rank leads by how likely they are to buy, learning from past wins and losses instead of relying on manually assigned points. * [Fit and intent matrix](/glossary/fit-and-intent-matrix.txt) A fit and intent matrix plots accounts by how well they match your ICP and how actively they are showing buying signals, to decide the right next action. * [Account scoring](/glossary/account-scoring.txt) Account scoring assigns a numeric value to each company based on fit and engagement criteria, so teams can compare accounts consistently and decide where to focus sales and marketing effort.