Lead qualification & scoring (the MQL/SQL handoff), explained

Marketing celebrates a record month: four hundred leads, the dashboard glowing green. Sales works a handful, lets the rest go cold, and quietly grumbles that "marketing's leads are rubbish." Both teams are looking at the same prospects and reaching opposite conclusions. That argument, repeated in companies of every size, is the whole reason lead qualification and scoring exist: to replace a gut-feel fight with a definition both sides signed.

The idea in depth: a shared funnel, not a marketing toy

Start with the words, because most of the confusion is vocabulary. A Marketing Qualified Lead (MQL) is a prospect who has shown enough fit and interest, the right job title at the right kind of company, plus behaviour like requesting a demo or lingering on the pricing page, that marketing believes a salesperson should reach out. A Sales Qualified Lead (SQL) is one that sales has looked at, accepted, and judged worth pursuing as a real opportunity. The line between them is the handoff: the single most contested border in any commercial organisation.

That terminology is not folklore; it has a documented origin. The analyst firm SiriusDecisions (acquired by Forrester in 2018) built the Demand Waterfall to give sales and marketing a single, shared picture of how a raw inquiry becomes revenue, flowing down through stages such as inquiry, MQL, sales-accepted lead, SQL and closed deal. Before a shared model like this, the two teams literally counted different things and then argued about the gap. The lesson is to treat qualification as a joint definition, agreed by both teams, rather than a label marketing assigns on its own. A score sales did not help design is a score sales will not trust, and an untrusted lead is an unworked lead.

The older, more granular cousin of the MQL is BANT, Budget, Authority, Need, Timeline. It traces back to IBM, whose sales leaders needed a fast, repeatable checklist for whether an enterprise prospect was real before committing reps to a deal that might take years to close. The four questions still hold up: can they pay, can the person you are talking to decide, do they have a problem you solve, and are they trying to solve it now? BANT is a sales-conversation qualifier; MQL/SQL is a pipeline-stage definition. They are layers, not rivals: lead scoring decides who gets a call (the MQL), and a BANT-style conversation decides whether that call produces an opportunity (the SQL).

flowchart TD
  I("Inquiry: anyone who raised a hand") --> M("MQL: fit + intent score crosses the bar")
  M --> A("Sales accepts the lead (the SLA in action)")
  A --> Q("SQL: BANT-style conversation confirms a real opportunity")
  Q --> O(["Opportunity in the pipeline"])
  A -.rejected: back to nurture.-> N(["Nurture / recycle"])
					

Scoring is the engine that decides who clears the MQL bar. A lead-scoring model adds points for fit (the firmographics, industry, size, role) and for behaviour (the engagement, downloads, visits, email opens), and fires an MQL when the total crosses an agreed threshold. The trap is famous enough to be a cliché in the field: a model that rewards activity mistakes a curious researcher who downloads five eBooks for a buyer, while missing the quiet executive who visits the pricing page once because they are ready to write a cheque. So weight fit heavily, then validate the threshold against history: score your last batch of won deals and ask whether your model would actually have flagged them. If it would not, the points are decoration.

Why the handoff is a contract, not a conveyor belt

Here is the part most teams under-invest in. The qualification model can be elegant and still fail, because the failure is rarely technical, it is organisational. Marketing throws leads over a wall; sales catches the ones it likes and lets the rest rot; nobody agreed in advance what "qualified" meant or what response was owed. The fix has a name: a service-level agreement between sales and marketing, in which marketing commits to a defined quantity and quality of leads, and sales commits to working each one within a set time and logging the outcome.

The evidence that this alignment matters is consistent. HubSpot's research on sales–marketing alignment reports that companies with a tightly aligned SLA see materially faster revenue growth than those without one, while teams working in silos can see revenue stall or decline. Treat the precise percentages as vendor-published and directional rather than laboratory-proven, but the direction is corroborated by the entire reason SiriusDecisions built the waterfall in the first place: shared definitions reduce the friction that quietly loses deals. Write the SLA before you tune the algorithm. One page: the agreed MQL definition, how many marketing will deliver, the maximum time sales has to first-touch one, and a standing fortnightly meeting where both teams review the leads that didn't convert and fix the definition together.

"The new role of marketing and sales is to help buyers progress through their buying journey, not push leads through a funnel or pipeline.", Forrester, on the B2B Revenue Waterfall (2021)

Mark Roberge, who built the revenue engine at HubSpot and now teaches at Harvard Business School, makes the same point from the sales side: the foundational step in aligning the two teams is to define the SLA, and, tellingly, he is openly sceptical of obsessing over the lead score itself. His instinct was to have marketing commit to a perceived value of leads rather than a raw count. The lesson for a leader: the number on the dashboard is not the goal. Worked, converted, learned-from leads are the goal, and only a contract between the two teams produces those.

The honest limitation. The whole edifice rests on a shaky premise, that a buying decision can be tracked through one named individual. It usually cannot. Forrester's analyst Terry Flaherty has argued bluntly for "the end of MQLs", citing data that lead-centric processes convert at under 1%, failing, in his framing, more than 99% of the time, precisely because they score a person while the real decision is made by a group of three or more. Forrester's updated B2B Revenue Waterfall (2021) shifts the unit of measure from the individual lead to the buying group and the opportunity, noting that over 80% of purchases now involve complex, multi-stakeholder buying. So use MQL/SQL scoring as a prioritisation aid, not a law of physics, and if you sell complex deals into committees, start asking how many people from one account are engaging, not just how many points one contact has racked up.

A worked example

Take a 12-person B2B software company selling a workflow tool to operations teams, treat it as a composite and every figure as illustrative. Marketing sends sales roughly 200 "leads" a month; sales works maybe 30 and dismisses the rest as junk. The founder's instinct is to buy a fancier scoring add-on. The better first move costs nothing but a meeting.

Write the definition together. Sales and marketing sit down and agree what a qualified lead looks like: an operations manager or above (fit), at a company of 50–500 staff (fit), who has either booked a demo or visited the pricing page twice in a fortnight (intent). Everything else is a nurture contact, not an MQL. Build the score to match. Fit points dominate, a perfect-title, perfect-size contact starts most of the way to the threshold, and a junior researcher at a too-small firm can download every asset and never qualify. Write the SLA. Marketing commits to 40 of these MQLs a month rather than 200 of anything; sales commits to a first touch within one business day and to marking each lead won, lost-now-nurture, or wrongly-scored. Close the loop. Every fortnight they review the "wrongly-scored" tags and adjust the model, the recycle path in the diagram above made real.

The illustrative result is not "we found a better algorithm." It is that the monthly fight disappears, because both teams are now arguing about a shared, editable definition instead of a feeling. Forty trusted leads that get worked in a day beat two hundred that breed resentment, and the fortnightly loop means the score gets sharper every month instead of calcifying.

flowchart LR
  subgraph Wall["Throw-it-over-the-wall"]
    A("Marketing: 200 'leads'") --> B("Sales cherry-picks ~30")
    B --> C(["170 rot; 'your leads are rubbish'"])
  end
  subgraph SLA["SLA + closed loop"]
    D("Joint MQL definition") --> E("Marketing: 40 qualified MQLs")
    E --> F("Sales first-touch in 1 day, logs outcome")
    F --> G(["Fortnightly review re-tunes the score"])
  end
					

Frequently asked questions

What's the actual difference between an MQL and an SQL?

An MQL is marketing's judgement, based on fit and behaviour, that a prospect is worth a salesperson's call. An SQL is sales' judgement, after looking at or talking to that lead, that it is a real opportunity worth pursuing. The MQL is a promise; the SQL is the acceptance of that promise. The gap between how many MQLs marketing sends and how many sales accepts as SQLs is one of the most useful health metrics you have, a wide gap means your definition is wrong or your handoff is broken.

Is BANT outdated?

BANT is old (it predates the software it now lives inside) but the four questions, can they buy, can they decide, do they have a real need, and is the timing now, are not. Where it strains is on complex modern deals: "Authority" assumes one decision-maker when there are usually several, and rigid "Budget" qualification can disqualify a real opportunity that has not formalised spend yet. Use it as a conversation checklist, not a gate that throws away anyone who fails one letter. Newer frameworks layer on multi-stakeholder and need-depth questions, but they are refinements of the same instinct.

How many points should make something an MQL?

There is no universal number, and chasing one is the wrong question. Set the threshold empirically: score your recent won deals with your draft model and find the point total that would have caught most of them without flooding sales with noise. Then treat it as a dial, not a setting, if sales says the leads are weak, the bar is too low; if marketing is starving sales, it is too high. The fortnightly review is where you turn it.

Should we score accounts instead of individual leads?

If you sell complex deals into buying committees, increasingly yes. This is the core of Forrester's critique: scoring one contact misses that the real signal is multiple people from the same company engaging at once. Account- or buying-group scoring (sometimes called the demand-unit or MQA approach) aggregates that activity. It is more work and needs the data to support it, so smaller or transactional businesses may not need it yet, but if your deals routinely involve three-plus stakeholders, the individual lead is the wrong unit.

Whose job is it to define a qualified lead, sales or marketing?

Both, in the same room, which is the entire point. A definition marketing writes alone will optimise for volume; one sales writes alone will reject everything short of a signed contract. The qualified-lead definition is the central clause of the SLA, and it only works when both teams own it and revisit it together. If you take one thing from this page, make it that: the model is downstream of the agreement, not the other way around.

Related in the Toolkit

Qualification is the entry point to the commercial funnel: it decides which prospects feed the acquisition and activation levers, and a good model is only as good as the customer need it scores against.

• Growth-lever framework (acquisition, activation, retention, monetisation, referral), qualification sits at the top of the acquisition lever; scoring decides which acquired leads are worth activating.
• Growth loops, flywheels & compounding, a tuned scoring model that feeds sales the right leads is a loop that compounds; the fortnightly review is its feedback edge.
• Recurring-revenue metrics (ARR/MRR waterfall, Rule of 40, magic number, CAC payback), better qualification lifts conversion and shortens CAC payback by stopping sales from burning time on bad-fit leads.
• Net & gross revenue retention (NRR/GRR) & expansion economics, scoring for fit, not just intent, is what produces customers who stay and expand rather than churn.
• Upsell, cross-sell & land-and-expand, the same qualification discipline applies inside the base: scoring existing accounts for expansion readiness.
• Customer needs identification & latent needs, the "Need" in BANT; a score is only meaningful if it tracks a need the buyer genuinely feels.
• Design sprints, a fast way to test a new qualification definition or scoring threshold against real prospects before you commit to it.
• Engagement, retention & loyalty programs, the nurture path for leads that don't qualify yet; most "not now" leads are future buyers, not junk.

Where to go next

• Forrester, "The End Of MQLs" (Terry Flaherty), the sharpest critique of single-lead scoring, with the sub-1% conversion data and the case for buying-group thinking. Read it before you over-invest in an individual-lead model.
• "Step 1 for Marketing/Sales Alignment: Define an SLA", Mark Roberge (YouTube), HubSpot's former CRO on why the SLA, not the score, is the foundation of a working handoff; short and practical.
• HubSpot, "How to Create an Effective Sales and Marketing SLA", a concrete, free walkthrough of writing the agreement, with the alignment-and-growth research behind it (vendor-published, so read the figures as directional).
• MarketingProfs, "Navigating the New Demand Waterfall", a clear explainer of the SiriusDecisions model that gave the industry MQL, SQL and the shared funnel in the first place.