---
name: "Build a lead qualification scorecard"
description: "Builds a lead scorecard whose fit and intent points are tested against won and lost deals, with thresholds sized to sales capacity and fair, privacy-safe criteria, so sales calls the leads most likely to buy first. Used when sales says the leads from marketing or ads are poor."
version: 2
verified_on: 2026-10-28
reverify_by: 2027-10-28
sources:
  - id: t1-hs-lead-scores
    title: "Build lead scores (HubSpot Knowledge Base)"
    url: https://knowledge.hubspot.com/scoring/build-lead-scores
    tier: T1
    accessed_on: 2026-10-27
  - id: t1-g-lead-form-create
    title: "Create lead form assets (Google Ads Help)"
    url: https://support.google.com/google-ads/answer/16726130
    tier: T1
    accessed_on: 2026-10-27
  - id: t1-m-crm-qualified
    title: "Set up your CRM for qualified leads (Meta Business Help)"
    url: https://www.facebook.com/business/help/279369167153556
    tier: T1
    accessed_on: 2026-10-27
  - id: t1-g-oci
    title: "Offline conversion imports FAQs (Google Ads Help)"
    url: https://support.google.com/google-ads/answer/10029210
    tier: T1
    accessed_on: 2026-10-27
  - id: t1-gdpr-art21
    title: "UK GDPR Article 21: Right to object"
    url: https://www.legislation.gov.uk/eur/2016/679/article/21
    tier: T1
    accessed_on: 2026-10-27
  - id: t1-eu-gdpr-art22
    title: "EU GDPR Article 22: automated individual decision-making, including profiling (as adopted)"
    url: https://www.legislation.gov.uk/eur/2016/679/article/22/adopted
    tier: T1
    accessed_on: 2026-10-28
  - id: t1-ico-b2b
    title: "Business-to-business marketing (ICO guidance)"
    url: https://ico.org.uk/for-organisations/direct-marketing-and-privacy-and-electronic-communications/business-to-business-marketing/
    tier: T1
    accessed_on: 2026-10-28
  - id: t1-ico-adm
    title: "Automated decision-making, including profiling (ICO detailed guidance)"
    url: https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/individual-rights/automated-decision-making/
    tier: T1
    accessed_on: 2026-10-27
---

> **How to use this file:** attach it to ChatGPT, Claude, Gemini or Copilot, replace every [PLACEHOLDER] with your details, and add any data or Deepstory evidence files the method needs.

**Best with:** Any chat AI that can analyse a spreadsheet; attach a CRM export of leads and deals with outcomes, with names and contact details removed.

## Quick start

1. Write down what a good customer looks like and what "ready for sales" means.
2. Split the score in two: fit (who they are) and intent (what they did, like asking for a demo).
3. Look at leads that became customers and leads that did not. Keep only the signals that are much more common in the ones that bought, and only signals you knew before sales called them.
4. Set the "call now" line so sales can call every lead above it within a day.
5. Do not score health, religion, ethnicity or other sensitive traits, and stop marketing to anyone who asks you to.
6. Let old activity expire so last year’s clicks do not count forever.
7. Check the scores against real sales every month and adjust.

## Before you start

- The problem is only that the lead form or ad brings poor leads; fix the source first. If sales also needs to choose which leads to call first, do both. See: Run lead form ads that bring real leads (https://deepstoryresearch.com/method/lead-form-ads-lead-quality)
- Customers buy at a checkout and nobody follows up leads (for example an online shop): a lead score has nobody to act on it; measure repeat purchase and customer value instead.
- Sales can already contact every lead within a working day: you do not need a call-now threshold. Set a one-working-day response rule and use only the disqualifiers and qualified-lead signals.
- You are a health, dental or medical provider: what a person asks about is usually health data, which this method forbids scoring. Score only non-health signals (channel, response, preferred time, location) or get privacy advice first.
- You have no CRM or pipeline yet; set one up first. See: Set up a CRM and sales pipeline (https://deepstoryresearch.com/method/crm-pipeline-setup)
- The score would decide whether someone gets credit, housing, a job, insurance, education or access to health services, or what price they pay. That is a significant automated decision with extra legal duties; get legal advice instead of using this method.

## Intake questions

Ask me these first:

1. **Who buys most, stays longest and is cheapest to serve?**
   Why: Fit criteria should describe your best customers, not every customer.
   If unknown, assume: Describe best customers from what sales remembers, and label every fit criterion UNTESTED until the back-test.
2. **What data arrives with each lead (form fields, company data, website and email activity), and are WhatsApp, phone or chat enquiries logged anywhere?**
   Why: A criterion you cannot capture reliably cannot be scored.
   If unknown, assume: Score only form fields and one or two tracked actions.
3. **Can you export the last 6 to 12 months of leads and deals with outcomes (won, lost, no response)?**
   Why: Back-testing shows which criteria actually separate buyers from non-buyers.
   If unknown, assume: Build a first version and label all points UNTESTED until at least 30 won deals with known lead data exist.
4. **How many leads a month, and how many can sales contact within a day?**
   Why: The sales-ready threshold must match capacity or good leads go cold.
   If unknown, assume: Ask who follows up leads and how many new leads each rep can call in a working day alongside open deals; if nobody knows, assume 5 a rep a day (ASSUMPTION). If nobody follows up leads, see Before you start.
5. **How long is the sales cycle, do you hold email permission for each lead, and how does sales decide a lead is qualified today?**
   Why: The SQL stage and the back-test’s won group need one written definition; the back-test leaves out leads younger than one sales cycle, Google accepts email-matched uploads only up to 63 days after the click, and the nurture path may email only leads with permission.
   If unknown, assume: SQL = a rep has confirmed need, budget and timing on a call; a 90-day sales cycle; and no lead is enrolled in marketing emails until their permission is checked. Label each an ASSUMPTION.
6. **Which CRM and plan do you use?**
   Why: Scoring tools and limits differ; HubSpot lead scoring, for example, needs a Professional or Enterprise plan.
   If unknown, assume: Ask where leads are logged today; if nowhere, set up a CRM first. If the CRM has no scoring tool, design the scorecard as a spreadsheet first.
7. **Which countries are your leads in, and do you score consumers or businesses?**
   Why: In the UK and EU, people can object to profiling for direct marketing, and consumer scoring carries more privacy risk; other countries have their own rules (for example India’s DPDP Rules, with notice and consent duties phasing in).
   If unknown, assume: Assume UK or EU rules apply and plan an opt-out path.

## Decision rules

| If | Then | Threshold | Type | Source |
|---|---|---|---|---|
| You design the scorecard. | Keep two parts: fit (who they are: industry, size, role, location) and intent (what they did: pages, emails, demo request). A high-fit, low-intent lead gets nurtured; a low-fit, high-intent lead gets a quick qualifying call. | - | Heuristic (rule of thumb) | - |
| You set up scoring in HubSpot. | Use a fit score for property values, an engagement score for events, or a combined score for both. Group related rules and give each group a limit so one behaviour cannot dominate; choose a score range such as -100 to 100. Engagement rules can use a timeframe and a frequency. | - | Platform rule | [t1-hs-lead-scores] |
| You have won and lost deal data. | For each criterion, compare its rate among leads that became won deals with its rate among all other leads from the same period (lost deals, disqualified and no response). Use only values the lead had when it arrived or before sales first contacted them; leave out anything sales filled in later. Leave out leads that arrived within the last sales cycle and leads whose deals are still open. Keep a criterion only if its rate among won deals is at least 10 percentage points above its rate among other leads; points = gap in percentage points / 2, rounded to the nearest 5; drop the rest. Treat any criterion seen on fewer than about 30 won deals as UNTESTED. Re-check with every 50 or so new closed deals. | - | Heuristic (rule of thumb) | - |
| You have no usable won and lost export yet. | Build a first version from your stage definitions only: 10 points for each fit criterion and each intent action, cap fit at 30 and intent at 40, disqualifiers at -50, and label every point UNTESTED. Leave the back-test columns as "not available"; never fill them with estimates, benchmarks or example figures. | - | Heuristic (rule of thumb) | - |
| Your leads are outside the UK and EU. | Say that this method’s legal rules do not cover that country and check its data protection law. Keep the sensitive-traits, opt-out and suppression rules as a minimum. | - | Heuristic (rule of thumb) | - |
| You set the sales-qualified threshold. | Count how many of last month’s leads would pass it. If sales cannot contact them all within a working day, raise the threshold or add a nurture path. If sales would have lots of spare capacity, lower the threshold or have sales call the top of the MQL band, so good leads are not left on email only. | - | Heuristic (rule of thumb) | - |
| Leads come from Google lead forms. | Mark multiple-choice answers that show a strong prospect as qualifying; Google then tags matching submissions as qualified leads. | - | Platform rule | [t1-g-lead-form-create] |
| You want Meta to find more leads like your qualified ones. | Connect your CRM in Events Manager and send each lead-status update (for example qualified, converted) as an event; this is recommended when you use the qualified leads performance goal. Before uploading lead emails or phone numbers to any ad platform, check that your privacy notice covers it and that you meet the platform’s consent requirements. | - | Platform rule | [t1-m-crm-qualified] |
| You want Google Ads to find more leads like your qualified ones. | Use enhanced conversions for leads (since April 2026 one on/off setting shared with enhanced conversions for web) with a qualified or converted lead goal. Leads matched by email or phone can be uploaded up to 63 days after the click (90 days with a click ID), so for long sales cycles send the qualified stage, not only the closed sale. | - | Platform rule | [t1-g-oci] |
| You score people in the UK or EU for direct marketing. | People can object at any time to their data being used for direct marketing, including profiling for it. Once someone objects, stop all direct-marketing use of their data: remove them from scoring, nurture emails, sales outreach lists and ad-platform uploads, and keep a suppression record so they are not re-added. Tell people about this right clearly and separately from other information, at the latest in your first message to them. This is a first check, not legal advice. | - | Legal rule | [t1-gdpr-art21] |
| A score would lead to a decision with legal or similarly significant effects, made with no meaningful human involvement. | That is automated decision-making. Under the UK GDPR (articles 22A to 22D) it needs safeguards: tell people about the decision, let them make representations and contest it, and give them human review. Under the EU GDPR (article 22) it is not allowed at all unless it is necessary for a contract with the person, authorised by law, or based on their explicit consent, and the same kind of safeguards then apply. The ICO’s guidance on this is being updated (consultation, March 2026). Ranking leads so sales calls some first is usually not a significant decision in our judgement, but it can become one if the score alone refuses someone a service, sets their price or targets people at risk; keep a person in the loop for those and do a data protection impact assessment. This is a first check, not legal advice. | - | Legal rule | [t1-ico-adm], [t1-eu-gdpr-art22] |
| Your nurture path sends marketing emails or texts to UK leads. | Check each lead’s email and text permission before enrolling them. Sole traders and some partnerships (for example non-LLP partnerships in England, Wales and Northern Ireland) count as individuals: email or text them only with consent or the soft opt-in. Companies, LLPs and Scottish partnerships do not need consent, but every message must say who you are and give a working way to opt out. If you cannot tell which type a lead is, treat it as an individual. Score everyone you lawfully hold, but only email those you may email. The ICO is reviewing this guidance after the Data (Use and Access) Act. This is a first check, not legal advice. | - | Legal rule | [t1-ico-b2b] |
| You choose criteria. | Never use sensitive traits (health, religion or beliefs, ethnicity, sexual orientation or sex life, political views, trade union membership, genetic or biometric data) or close proxies for them. For consumers, avoid age and gender unless the product legally requires it. | - | Heuristic (rule of thumb) | - |
| A lead has not engaged for a long time. | Let engagement points expire with a timeframe (for example the last 90 days), or use your CRM’s score decay (HubSpot can cut a score by a set percentage after a set period), so old activity does not keep a lead "hot". | 90 days (rule of thumb) | Heuristic (rule of thumb) | - |

## Procedure

Copyable checklist:

- [ ] Define stages: Lead, marketing-qualified lead (MQL), sales-qualified lead (SQL), customer: one line each, with who decides.
- [ ] List candidate criteria: Fit: industry, size, role, location, budget band. Intent: demo or pricing page, form type, email replies, repeat visits, free trial started and key in-trial actions.
- [ ] Back-test: For each criterion, compare its rate in leads that became won deals with its rate in all other leads (lost, disqualified, no response), using only data known before first sales contact; keep the ones that separate clearly.
- [ ] Assign points and limits: Bigger separation, more points; cap each group; add negative points for clear disqualifiers (students, competitors, out-of-area).
- [ ] Set thresholds: MQL and SQL cut-offs sized to sales capacity, checked against last month’s leads.
- [ ] Route: Who gets each SQL, response time target, nurture path for MQLs (only for leads you may email), recycle rule for no-response (for example back to nurture after 30 days with no reply; rule of thumb).
- [ ] Fairness and privacy: No sensitive traits or proxies; objection handling; human review for anything beyond sales routing.
- [ ] Calibrate: Each month, check that won deals scored higher on average than lost ones and log it; change points only after about 50 new closed deals or 3 months, whichever is later.

## Output template

```
## Lead scorecard: [BUSINESS]

### Stage definitions
Lead: [ ] | MQL: [ ] | SQL: [ ] | Customer: [ ]

### Back-test
Data: [export date range, leads, won deals] or "no export yet: all UNTESTED"
Criterion | % of won | % of all other leads | Won deals in sample | Status (tested / UNTESTED) | Keep? | Points

### Fit score (limit [ ])
Criterion | Value | Points

### Intent score (limit [ ], timeframe [ ] days)
Action | Frequency | Points

### Disqualifiers
Criterion | Points

### Thresholds and routing
MQL: [ ] | SQL: [ ] | Leads passing last month: [ ] | Sales capacity: [ ] a day
Routing: [who, response time] | Nurture: [ ] | Recycle: [ ]

### Fairness and privacy
Excluded criteria: [ ] | Objection handling: [ ]

### Calibration
Monthly review date: [ ] | Change log: [ ]
```

## Failure modes

| Mistake | Symptom | Fix |
|---|---|---|
| Points set by opinion, not outcomes. | High scores that do not buy. | Back-test every criterion against won and lost deals. |
| Mixing fit and intent in one number without limits. | A student who downloads everything outranks a perfect buyer. | Separate fit and intent, and cap groups. |
| Threshold ignores sales capacity. | Sales ignores the "hot" list because it is too long. | Size the SQL threshold to what sales can call within a day. |
| Old activity never expires. | Leads stay "hot" for months. | Use a timeframe for engagement points. |
| Scoring on sensitive traits or proxies. | Discrimination and privacy risk. | Remove them; review criteria for proxies. |
| Back-testing only on deals, or on fields sales fills in later. | The score looks excellent in testing but does not predict which new leads buy. | Compare buyers with all other leads, using only data known before first contact. |
| Weighting criteria from a handful of wins. | Points swing wildly at each monthly review. | Label criteria with fewer than about 30 won deals UNTESTED and keep their points small. |
| Objections not suppressed everywhere. | People who objected still get emails or are uploaded to ad audiences. | One suppression flag in the CRM that every list, sync and workflow excludes. |
| Never recalibrating. | Scores drift away from reality after product or market changes. | Monthly check against outcomes. |
| No path for leads below threshold. | Good future customers are forgotten. | Nurture and recycle rules for MQLs. |

## Verification

- [ ] Every criterion shows won versus lost rates, or is labelled UNTESTED.
- [ ] Fit and intent are separate, with group limits.
- [ ] The SQL threshold is checked against last month’s volume and sales capacity.
- [ ] No sensitive traits or proxies are scored.
- [ ] Objection handling and a monthly calibration date are written down.

## Agent mode

If an AI agent is operating a browser or API on your behalf:

**May do without asking:**
- Analyse an exported spreadsheet of leads and deals with personal details removed.
- Draft the scorecard, thresholds and routing rules.
- Read existing scoring settings in a read-only session.

**Must stop before (the user decides and does these):**
- Turning on or editing a live score or workflow in the CRM.
- Bulk-editing, merging or deleting CRM records, or overwriting lifecycle stage or score properties.
- Enrolling contacts in emails, sequences or workflows, or sending any message.
- Editing a live Google or Meta lead form; on Google an edit that needs policy review pauses the campaigns using it.
- Uploading or syncing lead or customer data to ad platforms.
- Exporting or sharing personal data outside the business.
- Accepting CRM or integration terms, or changing billing plans.

**Setup advice:**
- Remove names, emails and phone numbers before sharing exports with an AI tool.
- Build the score as a draft, use HubSpot’s Preview distribution (or a spreadsheet copy in other CRMs), check 20 records by hand, then turn it on yourself.

## Worked example

The figures below are illustrative. Never reuse them: use only the user’s own numbers and facts, and label anything they have not given as an ASSUMPTION.

**Scenario:** An Indian B2B SaaS company selling practice software to accounting firms in India and the UK.

**Input:** 300 leads a month from ads and webinars, 2 sales reps who each have time for about 3 new leads a working day alongside their open deals. Last 12 months: 120 won and 480 lost deals. Form fields: company size, role, country, "When do you need a solution?". Tracked: pricing page visits, webinar attendance, demo requests.

**Output excerpt:** Back-test against all other leads, using only data known at arrival (points = gap in percentage points / 2, rounded to 5): 10 to 50 staff 68% of won versus 31% of other leads (keep, +20); partner or owner role 74% versus 40% (keep, +15); pricing page visit 61% versus 22% (keep, +20); demo request 55% versus 9% (keep, +25); webinar attendance 35% versus 33% (no difference: drop, 0); "need within 3 months" 58% versus 20% (keep, +20). Disqualifiers: students and competitors -50. SQL threshold 70 of 100: last month 64 of 300 leads would pass, about 1.5 per rep per working day (64 / 2 reps / 21 days). Capacity is 2 reps x 3 leads x 21 days = 126 a month, fewer than the 300 leads arriving, so a threshold is needed; the 62 spare calls a month go to MQLs scoring 55 to 69, highest score first, within 2 days; MQL 40 to 54: email nurture to those who agreed to marketing emails, recycle after 30 days. Privacy: the first email to every UK lead states the right to object; anyone who objects is suppressed from scoring, nurture, sales outreach and ad uploads. Review on the first Monday of each month; label all points UNTESTED again if the product or market changes.

## References

### Platform settings

| Platform | Path | Setting | Note | Source |
|---|---|---|---|---|
| HubSpot | Marketing > Lead Scoring | Fit score, engagement score or combined score; groups with limits; score range; thresholds | - | [t1-hs-lead-scores] |
| Google Ads | Lead form editor > Custom questions > multiple-choice question set as qualifying | Select the answers that indicate a strong prospect; qualifying questions cannot be optional | - | [t1-g-lead-form-create] |
| Meta Events Manager | Connect data sources (plus icon) > CRM > Connect | CRM connection for the qualified leads performance goal | - | [t1-m-crm-qualified] |

### Sources

| ID | Title | Tier | Accessed on | URL |
|---|---|---|---|---|
| t1-hs-lead-scores | Build lead scores (HubSpot Knowledge Base) | T1 | 2026-10-27 | https://knowledge.hubspot.com/scoring/build-lead-scores |
| t1-g-lead-form-create | Create lead form assets (Google Ads Help) | T1 | 2026-10-27 | https://support.google.com/google-ads/answer/16726130 |
| t1-m-crm-qualified | Set up your CRM for qualified leads (Meta Business Help) | T1 | 2026-10-27 | https://www.facebook.com/business/help/279369167153556 |
| t1-g-oci | Offline conversion imports FAQs (Google Ads Help) | T1 | 2026-10-27 | https://support.google.com/google-ads/answer/10029210 |
| t1-gdpr-art21 | UK GDPR Article 21: Right to object | T1 | 2026-10-27 | https://www.legislation.gov.uk/eur/2016/679/article/21 |
| t1-eu-gdpr-art22 | EU GDPR Article 22: automated individual decision-making, including profiling (as adopted) | T1 | 2026-10-28 | https://www.legislation.gov.uk/eur/2016/679/article/22/adopted |
| t1-ico-b2b | Business-to-business marketing (ICO guidance) | T1 | 2026-10-28 | https://ico.org.uk/for-organisations/direct-marketing-and-privacy-and-electronic-communications/business-to-business-marketing/ |
| t1-ico-adm | Automated decision-making, including profiling (ICO detailed guidance) | T1 | 2026-10-27 | https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/individual-rights/automated-decision-making/ |

---
Deepstory methods are process guidance for AI tools. Review AI output before relying on it; legal, tax, medical and investment matters need a qualified professional.
