Ask five people at the same company what an MQL is and you get five answers. Ask who is allowed to change the marketing qualified lead definition and the room goes quiet. That silence is the real problem. The number gets reported every month, the two teams argue about it twice a year, and the sentence has never had an author.
Here is the version that holds up. An MQL is a promise marketing makes to sales: these are the people you have to call. A promise nobody can refuse is not a promise. So the definition only means something if a rep can hand a lead back and say why.
Marketing writes the score. Nobody writes the promise.
Most teams build the model and stop there. Someone weights a few fields in the marketing platform, sets a threshold, and the threshold quietly becomes the definition. Seventy points is an MQL because seventy points is where the line landed.
Nobody in sales agreed to that line. They were shown it. There is a difference. It surfaces about six weeks later, when a rep opens a lead from an eleven-person company with no budget and stops trusting the queue.
Write the commitment first, in plain words, before anyone opens the scoring tool. It sounds like this. A lead is qualified when the company is one we can serve, the person has some say in the decision, and something they did this month says they are working on the problem now. Build the score to approximate that sentence. The score is the implementation. The sentence is the contract.
The subtraction that never reaches the report
Watch what happens to a month of leads. The arithmetic below is a worked example, chosen so the ratios are easy to follow. It is not measured from any account and it is not a benchmark.
Four hundred form fills come in. A hundred and eighty pass the score and get flagged as MQLs. Sales works ninety-six of them. Sales accepts forty-one. Twelve turn into opportunities.
Data table
| Step | Leads |
|---|---|
| Form fills | 400 |
| Flagged as MQLs | 180 |
| Worked by sales | 96 |
| Accepted by sales | 41 |
| Became opportunities | 12 |
The monthly report shows 180 MQLs. That count is accurate. It also hides two different failures that look identical from the outside. Sales never opened 84 of those leads at all. Of the 96 they did open, sales accepted 41.
Those are two separate problems and they need two separate fixes. The 84 sales never opened is a coverage problem: capacity, routing, or a queue reps have stopped opening. The 41 out of 96 is the definition itself, tested against the leads sales actually worked. Most teams report neither and publish the blended figure instead. If your B2B marketing funnel has a stage called MQL, these are the numbers that tell you whether the stage means anything.
Who signs it, and why marketing cannot sign alone
A definition written by marketing alone is a forecast of what sales will accept. A definition written by sales alone asks for people who are already ready to buy. That is a very short list.
Three people have to be in the room. Whoever owns demand, whoever runs the reps, and one rep who actually works the queue. That last seat gets skipped constantly, and it is the one that changes what gets written. A frontline rep needs a minute to tell you which form fields are noise and which ones make them pick up the phone.
Then write down who can change the sentence later, and how often. Quarterly is fine. Never is not. The market moves, your product moves, and a definition you have not opened in two years is describing a company you no longer are. This is the same negotiation that our post on sales and marketing alignment covers at the level of the whole relationship. Here it fits on one page.
The rejection loop is the mechanism
Everyone builds the road into sales. Almost nobody builds the road back.
In most systems a rep with a bad lead has three options. Work it anyway and lose the hour. Leave it in the queue. Complain in a meeting. None of those produce data. A rejection loop replaces all three with one action: hand the lead back, pick a reason from a short list, and it returns to marketing with a label on it.
Keep the reason list short enough to use. Wrong company. Wrong person. Not now. Already a customer. Bad data. Five reasons, one click, no free text box for reps to ignore. After a quarter you can sort rejected leads by reason, and the list tells you which part of the definition is wrong. If not-now dominates, the definition is fine and your lead nurturing is the thing that needs work. If wrong-company dominates, the fit rules are broken.
Then pick your denominator on purpose. In the month above, sales accepted 41 of the 96 leads it worked, about 43 percent. That is the definition being tested. Measured against all 180 flagged it is 23 percent, which is the number most teams report and the one that tells you least, because it silently blames the definition for a coverage gap. Measured against all 400 form fills it is about 10 percent, which is what the demand programme actually produced.
Data table
| Measure | Value |
|---|---|
| Accepted, of the 96 leads sales actually worked | 43% |
| Accepted, of all 180 flagged as MQLs | 23% |
| Accepted, of all 400 form fills | 10% |
Scores that measure the wrong thing
Two failures turn up again and again when we open a scoring model.
The first is scoring effort instead of intent. You give ten points for a webinar, ten for an ebook, five for an email click. Add enough of those and a competitor doing research on you outranks a buyer who read your pricing page once. Activity is not interest. In your model, one visit to the page that names a price probably deserves more weight than four downloads.
The second is scoring one person when the decision belongs to a group. A B2B purchase usually involves several people, and the one who fills in the form may not carry much weight among them. A score that only ever describes an individual keeps flagging the analyst. It misses the account where three colleagues opened the same page in a fortnight. Account-level signals catch that. Another five points on the individual does not.
How to rewrite it in one meeting
You do not need a project for this. Book ninety minutes and bring last quarter’s leads.
Pull twenty leads that closed and twenty the reps rejected, and read them out loud one at a time. What did the good ones have in common that the model does not currently look at? Write the sentence together in the room. Set the threshold last, and set it tighter than feels comfortable. A small queue that sales trusts beats a big one they have learned to ignore. Then agree the five rejection reasons, and who reviews them each month.
Put the sentence somewhere both teams see it. A slide in the monthly review gets looked at. A page in a shared drive waits to be looked for.
What changes when the definition holds
The queue gets smaller. The accept rate should go up, and if it does not, the new definition is not better, it is only stricter. On a dashboard built around MQL volume that trade looks like a loss, so put the accept rate on the same dashboard from day one.
The quieter change matters more. Rejection stops being a complaint and becomes a number. The argument between the two teams turns into a monthly review of five reason codes. That is a much better meeting. It is also the point: not a tidier definition, but a shorter distance between what marketing sends and what sales will work. Shortening that distance is a large part of what we do at COSEOM® as a B2B lead generation agency.
FAQ: marketing qualified leads
What is the standard marketing qualified lead definition?
There is no standard, and a borrowed one is worse than a rough one you wrote yourself. The usable shape has three parts. The company fits what you can serve. The person has some say in the decision. Something recent suggests they are working on the problem now. Your business decides what fit means, how much say the person needs, and how recent counts as recent. No template can do that for you.
How does an MQL become an SQL?
An MQL is marketing saying a lead is worth a call. An SQL is sales agreeing after making the call. The handoff between them is where the value leaks, because in many companies nothing records the disagreement.
Who should own the definition?
Both teams, with a named person on each side who can approve a change. Marketing drafts it, sales signs it, and one rep who works the queue reads it before anyone agrees. Without two names on it you are back to a threshold.
Is there a benchmark for MQL to SQL conversion?
Nothing portable, because every company draws the line somewhere else, so a rate from another business is measuring a different definition. Tighten the definition and the rate climbs, with the marketing underneath it unchanged. Compare it against your own previous quarter. A rate that rises while volume falls often means the definition improved, but check the lead mix and the routing before you take the credit.
Should we drop MQLs and report pipeline instead?
Report pipeline as the outcome. Keep the MQL as the handoff, because reps still need a queue and somebody has to decide what goes in it. The failure is treating the count as a goal. It tells you how many people marketing sent over. That is all it tells you.
When should the definition change?
Review it quarterly, and change it only when the rejection reasons point at something. If one reason dominates two quarters running, start with the rule it points at, then rule out routing and sales capacity before you rewrite anything.
COSEOM Team
COSEOM is a multilingual B2B growth agency. Since 2008 the COSEOM team has run international SEO, generative engine optimization, paid media and demand generation programs for B2B software and technology companies, in English, German, Spanish and Portuguese. This account publishes our research, playbooks and product guides. More about COSEOM