SOURCE-BASED ANALYSIS · NO HANDS-ON RATING
Write the decision the score supports
HubSpot documents engagement scores, fit scores and combined scores for supported objects and subscriptions. The score property can then be used by segments, workflows and reports. Before assigning points, state the decision: for example, which records receive human review this week. A score without a named decision becomes difficult to validate.
Create fixtures before tuning thresholds
Assume a fictional business values an approved company profile and a completed meeting more than an email open. Prepare synthetic fixtures representing high fit with low engagement, low fit with high engagement, missing data and a clearly ineligible record. Define expected relative order without inventing a conversion rate.
| Fixture | Expected interpretation | Review question |
|---|---|---|
| High fit, no recent action | Fit signal only | Should it enter outreach automatically? |
| Low fit, repeated email activity | Engagement without fit | Can activity overpower exclusion rules? |
| Missing company data | Uncertain fit | Is missing treated differently from negative? |
This is an illustrative validation set, not measured buyer behavior.
Review group limits and thresholds
Document each rule, point value, group limit and threshold label. Test the sample records with the product's preview controls when available. A High, Medium or Low label reflects configured ranges; it is not a verified probability of purchase.
- Name the decision owner and eligible population.
- Define positive, negative and missing-data fixtures.
- Record rules, deductions, limits and thresholds.
- Preview distribution and inspect outliers.
- Review the score property's use in workflows, segments and reports.
- Approve activation and a dated review interval.
Treat edits as model changes
HubSpot states that turning on or updating a score evaluates records using the applicable current and historical data, and downstream tools using the score may be affected. Compare before-and-after fixtures and document the expected movement between thresholds before saving a material change.
Accept the model when reviewers can explain why each fixture received its relative position and downstream actions match the declared policy. This guide does not claim predictive accuracy, measured uplift or a live account test.
Sources and verification
Official documentation checked on 2026-10-29. Worked examples are hypothetical planning exercises, not measured product benchmarks.