Lead Scoring
A systematic method for ranking leads by fit and engagement so sales works the accounts most likely to convert first.
What is lead scoring?
Lead scoring is the process of assigning a numeric value to every lead in your database based on two independent dimensions: fit (how closely the person and company match your ideal customer profile) and engagement (how actively they are showing buying intent through behavior). The combined score determines which leads are passed to sales, in what order, and with what urgency.
Fit is scored on firmographic and demographic attributes — industry, company size, region, job title, seniority — and is largely static. Engagement is scored on behavior — pricing page visits, demo requests, email replies, product signups, event attendance — and decays over time. Keeping the two dimensions separate matters: a perfectly-fitting CFO who has never engaged needs outbound nurture, while a student who downloaded five ebooks needs neither. Blending them into one number hides exactly the distinction sales needs.
Lead scoring is not the same as lead qualification (a human or automated conversation that verifies need, budget, and timeline) or lead routing (the assignment of a lead to an owner). Scoring feeds both: it decides who is worth qualifying, and routing decides who does the qualifying.
When to implement
Implement lead scoring once you have enough inbound volume that sales cannot follow up with everyone — typically 500+ new leads per month — and at least 6–12 months of closed-won history to calibrate against. Below that volume, a simple demo-request-first rule outperforms a scoring model and costs nothing to maintain.
Step-by-step workflow
- 1
Define the ideal customer profile with sales and CS
Owner: Marketing Operations + Sales leadership
Pull the last 12 months of closed-won and closed-lost deals and identify the firmographic attributes that separate winners: industry, employee count, revenue band, tech stack, geography. Agree the ICP definition in writing with sales and customer success — the score is only trusted if the people receiving leads co-authored the definition.
- Export closed-won/lost deals with firmographic fields
- Rank attributes by win-rate lift and deal size
- Document ICP tiers (A/B/C) and get sign-off from sales leadership
- 2
Assign fit points to profile attributes
Owner: Marketing Operations
Translate the ICP into a fit score, typically 0–100. Weight the attributes that actually predict revenue, not the ones that are easy to capture. Include negative scoring for disqualifiers: students, competitors, personal email domains, unsupported regions.
- Score title/seniority, company size, industry, region
- Add negative scores for hard disqualifiers
- Cap each attribute so no single field dominates
- 3
Assign engagement points to behaviors
Owner: Marketing Operations
Score behaviors by proximity to purchase intent. High-intent actions (demo request, pricing page, trial signup) should be worth an order of magnitude more than passive ones (blog visit, email open). Email opens are unreliable since Apple Mail privacy protection; weight clicks and replies instead.
- Group behaviors into high / medium / low intent tiers
- Set point values with a clear gap between tiers
- Exclude bot and internal traffic from scoring
- 4
Add score decay
Owner: Marketing Operations
Engagement from 90 days ago is not buying intent today. Apply time decay — for example, halve engagement points after 30 days of inactivity and zero them after 90 — so the score reflects current interest, not accumulated history.
- 5
Set the MQL threshold jointly with sales
Owner: Marketing Operations + Sales leadership
Pick the score at which a lead becomes an MQL and is passed to sales. Set it by back-testing: at a candidate threshold, how many of last quarter's leads would have qualified, and what share of those actually converted? Tune the threshold to the volume sales can actually work at agreed SLA.
- 6
Wire the score into routing and SLAs
Owner: Revenue Operations
Automate the handoff: when a lead crosses the threshold, it is routed to an owner (see lead routing) with a follow-up SLA, and the score, score reasons, and key behaviors are visible on the record so the rep knows why the lead qualified.
- 7
Review score performance monthly
Owner: Marketing Operations
Each month, compare MQL-to-opportunity and MQL-to-won conversion by score band. If high scorers don't convert better than low scorers, the model is decorative. Collect rep feedback on false positives and adjust weights.
- Report conversion by score decile
- Review rejected MQLs with SDR team
- Log every scoring change with date and rationale
- 8
Recalibrate quarterly against closed revenue
Owner: Marketing Operations + RevOps
Quarterly, re-run the win/loss analysis and adjust attribute weights, decay rules, and the threshold. If volume and data maturity allow, evaluate whether a predictive (machine-learned) model outperforms the manual rules — but only after the manual model's data hygiene is proven.
Roles & responsibilities
| Role | Responsibility |
|---|---|
| Marketing Operations | Owns the scoring model: design, implementation, monitoring, and recalibration. |
| Sales leadership | Co-signs ICP and MQL threshold; commits to follow-up SLAs on qualified leads. |
| SDR / BDR team | Works scored leads within SLA and feeds back false positives and false negatives. |
| Revenue Operations | Implements routing, SLA timers, and reporting infrastructure downstream of the score. |
| Demand Generation | Uses score bands to decide nurture vs. accelerate treatment for each segment. |
Tool stack
Marketing automation platform
HubSpot · Marketo · Pardot (Account Engagement) — where the scoring model usually lives
CRM
Salesforce · HubSpot CRM — where the score must be visible to reps
Enrichment
Clearbit · ZoomInfo · Clay — fills the firmographic fields fit scoring depends on
Predictive scoring (optional)
MadKudu · 6sense · Breadcrumbs — worth evaluating only after manual scoring works
Product analytics (for PLG)
Amplitude · Mixpanel · Segment — source of product-usage signals for product-qualified leads
Key metrics
| Metric | Definition | Formula | Typical target |
|---|---|---|---|
| MQL-to-opportunity conversion rate | Share of MQLs that become sales-accepted opportunities. | Opportunities created from MQLs ÷ total MQLs | 10–25% depending on motion |
| Score-band lift | How much better top-decile leads convert than bottom-decile. The core validity check of the model. | Top-decile win rate ÷ overall win rate | ≥ 3× for a useful model |
| MQL rejection rate | Share of MQLs sales explicitly rejects as unqualified. | Rejected MQLs ÷ total MQLs | < 20% |
| Speed to lead | Time from crossing the MQL threshold to first sales touch. | Timestamp of first touch − timestamp of MQL | < 1 hour for high-intent leads |
| Scoreable-field completeness | Share of leads with enough firmographic data to compute a fit score. | Leads with complete fit fields ÷ total leads | > 90% after enrichment |
Common failure points
| Failure | Symptom | Fix |
|---|---|---|
| One blended score | High-fit/no-intent and low-fit/high-activity leads get identical treatment; reps distrust the number. | Split into separate fit and engagement scores; route on the combination, not the sum. |
| Scoring what's easy, not what predicts | Email opens and page views dominate; MQLs don't convert better than raw leads. | Back-test every attribute against closed-won data; delete weights that show no lift. |
| No score decay | Leads who binged content a year ago still sit at the top of the queue. | Add time decay to all engagement points; recompute nightly. |
| Threshold set by marketing alone | Sales ignores MQLs; a shadow qualification process emerges. | Set and revisit the threshold jointly; tie it to sales capacity and an explicit follow-up SLA. |
| Set-and-forget model | Model reflects the ICP and product of two years ago. | Monthly performance review, quarterly recalibration, changelog for every adjustment. |
| Bot and internal traffic pollution | Random leads spike to MQL after security scanners or employees hit the site. | Filter known bots, internal IPs, and email-scanner clicks before points are awarded. |
Frequently asked questions
- What is the difference between lead scoring and lead grading?
- In tools that use both terms (notably Pardot), the score measures engagement behavior while the grade (A–F) measures profile fit. The industry mostly says 'fit score' and 'engagement score' for the same distinction. The important part is keeping the two dimensions separate, whatever the labels.
- Should we use predictive (AI) lead scoring instead of rules?
- Predictive models can outperform manual rules, but only with volume (thousands of leads per month), clean CRM outcome data, and 12+ months of history. Most teams get more value from a well-maintained rules-based model first; the data hygiene work required for it is also the prerequisite for any predictive model.
- What score should trigger an MQL?
- There is no universal number — thresholds are only meaningful within your own model. Set it by back-testing candidate thresholds against last quarter's conversion data, and by sales capacity: the threshold should produce roughly the MQL volume your SDR team can work within SLA.
- How often should the scoring model be updated?
- Review performance monthly (conversion by score band, rejection rate) and recalibrate weights quarterly. Also re-score after any material change to ICP, pricing, or product — the model encodes assumptions that those changes invalidate.
- Does lead scoring apply to product-led (PLG) companies?
- Yes, but the strongest signals move from marketing behavior to product usage: activation events, seats invited, feature depth, usage frequency. This variant is usually called PQL (product-qualified lead) scoring, and it follows the same fit + engagement structure with product events as the engagement dimension.
Download the SOP
The standard operating procedure for this process — purpose, roles, step-by-step procedure with checklists, metrics, and failure modes — is available as a Markdown file you can drop into Notion, Confluence, or any wiki and adapt.
↓ Lead Scoring SOP (.md)Related processes
- Lead RoutingThe rules and automation that assign every inbound lead to the right owner, fast, with an SLA — so no qualified lead waits or falls through the cracks.
- MQL-to-SQL HandoffThe agreement and mechanics by which marketing-qualified leads become sales-accepted pipeline — definitions, SLAs, feedback loops, and the accountability between the two teams.
- Outbound SequencingDesigning and running multi-touch, multi-channel outbound cadences — email, phone, LinkedIn — that turn a target account list into booked meetings.
Cite this page
“Lead Scoring: definition, workflow, roles, metrics & SOP.” b2bprocess.com, updated 2026-07-08. https://b2bprocess.com/lead-scoring