Why Your MQL Count Is Probably Lying to You
By Anindo Neel Dutta
MQLs are not qualified leads.
They're often just people who clicked something.
Your dashboard says you generated 847 MQLs last quarter. Sales closed four. Marketing calls that a volume win. RevOps calls it a reporting problem. Both miss the point: most of those 847 never had buying intent. They had activity.
Activity is not intent
Most scoring models work like this:
- Email open: +5
- Content download: +10
- Webinar attendance: +15
- Website visit: +5
- Form submission: +20
Cross the threshold—usually 50 to 75 points—and the lead becomes an MQL. Slack pings sales. The "lead" is a student, a freelancer who will never buy, or someone who clicked your ebook because the title sounded useful.
That's engagement, not intent. Engagement tells you who is curious. Intent tells you who is evaluating a purchase.
When MQL volume climbs but pipeline quality stays flat, the count isn't a growth metric. It's a number that's easy to produce and hard to defend in a pipeline review.
Start with your closed deals
The fix isn't importing a scoring template from a blog post or a HubSpot playbook written for a different ICP.
Your CRM already has the data.
Pull every closed-won deal from the last 12–18 months. For each one, capture what was true at lead creation, not at close—close-time data is contaminated by months of sales conversations:
- Company size, industry, geography, title
- Lead source and campaign
- Pages visited before the first meaningful conversion
- Conversion path (ad → landing page → demo request, etc.)
- Time from first touch to opportunity
- Deal size and sales cycle length
Then pull every lead that hit MQL but never became a customer. Same fields. Put both groups side by side.
Look for patterns that show up in the won column and barely appear in the stalled column. Maybe 80% of closed deals came from 50–500 employee companies. Maybe VP-level titles converted at 4× the rate of ICs. Maybe pricing-page visitors in week one closed twice as often as content-only leads.
A cross-tab in Google Sheets against your HubSpot or Zoho export will surface three to five attributes that predict revenue in your business. Those become your model—not a template from someone else's ICP.
Build two scores
The mistake most teams make is collapsing everything into one number. Opens and downloads get mixed with company size and title. A curious stranger with a Gmail address can outscore a VP at a target account because they clicked more things.
Split the model into two dimensions:
Fit: Does this lead look like your customers?
Intent: Is this person behaving like someone evaluating a purchase?
Score them independently. MQL only fires when both clear a threshold.
Adjust these weights after your closed-won analysis, not before:
Fit (max ~100)
| Signal | Points |
|---|---|
| Company size in your ICP band (e.g. 50–500 employees) | +30 |
| Target industry | +25 |
| VP / Director / Head-level title | +20 |
| Known target geography | +15 |
| Enriched company data (Clay, Apollo, or native CRM fields) | +10 |
| Company size under 10 or over 5,000 | −40 |
| Non-target industry | −25 |
| Personal email, no company attached | −30 |
Intent (max ~100)
| Signal | Points |
|---|---|
| Pricing page visit | +25 |
| Demo or contact page view without form fill | +20 |
| Return visit within 72 hours | +15 |
| Requested demo or trial | +40 |
| Attended live webinar (not replay) | +10 |
| Content download | +5 |
| Email open | +2 |
Example thresholds: Fit ≥ 50 and Intent ≥ 40 before routing to sales.
High fit, low intent: nurture. High intent, low fit: disqualify—a competitor on your pricing page is zero opportunity. Both high: same-day outreach. One composite score hides these cases. Two scores make the handoff legible.
Then validate it against revenue
A scoring model that nobody checks is just a formula collecting dust on a lifecycle stage.
Within 90 days, compare MQL → SQL conversion by score band. Top-quartile leads should convert better than bottom-quartile. Flat rates across bands means your score predicts nothing.
Pull closed-won and closed-lost from the last quarter. What Fit and Intent scores did each carry at MQL creation? If winners and losers look the same, fix the weights.
Track revenue per MQL, not MQL count. A working model often drops total MQLs while revenue per MQL climbs. Rebuild weights every six months—ICP, product, and market all move.
The goal isn't more MQLs
MQL count is easy to inflate—lower the threshold, add points for opens, count webinar replays as attendance. Sales doesn't care about that chart. They care about one question: Who should I talk to today?
A useful model answers it. Fewer junk handoffs. A morning queue sales trusts. A number that tracks closed revenue, not click activity.
If MQL count is climbing and win rate isn't, the count is lying. Stop optimizing for volume. Start scoring for who buys.