A marketing qualified account is a company where firmographic fit, buying intent, and engagement from multiple people all cross a threshold your sales team agreed to in advance. It qualifies the account, not one person inside it.
That single change fixes the problem most B2B teams have been living with for a decade. Marketing hits its MQL number. Sales misses quota anyway. Nobody can explain why, because an MQL only ever described one form fill.
So below you'll find the three inputs that qualify an account, a scoring model you can build this quarter, and the part most guides leave out: why these programs stall after the pilot.
What Is a Marketing Qualified Account (MQA)?
An MQA is an account that has shown enough buying evidence, across enough people, that a rep should work it now.
The definition has three load-bearing words.
Account. The unit of qualification is the company, not the contact.
Multiple. One champion downloading an ebook is not an MQA. It's an MQL wearing a costume.
Agreed. If sales didn't help set the threshold, you've built a reporting metric that nobody downstream will honor.
Where an MQA Sits in the Account Stages
Most account-based teams run a chain like this:
- Target account. Matches your ICP on paper. Nothing has happened yet.
- Engaged account. Someone did something. One visit, one download.
- MQA. Fit, intent, and multi-contact engagement clear the bar together.
- SQA. A rep looked at it, vetted it, and got a real conversation.
- Opportunity. Need, champion, timeline, active process.
- Closed won.
Stages 1 and 2 are where most account lists die quietly. Stage 3 is the one you're building.
Forrester formalized a version of this in its B2B Revenue Waterfall, which tracks buying groups and opportunities rather than individual leads. If you're redrawing your account based marketing funnel, that's a useful reference point for the stage names.
What an MQA Is Not
It is not an MQL with extra steps. It is not an intent spike on its own, because intent without fit just tells you a company you can't serve is shopping somewhere. And it is definitely not every account your ABM platform lit up this week.
MQA vs MQL: What Actually Changes
The simplest way to see the difference is to look at what lands in a rep's inbox.
An MQL arrives as a name. Priya at Acme filled in a form about pricing, scored 71, over the threshold, go.
An MQA arrives as a picture. Four people at Acme engaged in the last three weeks. One is the VP of Engineering, who read two product pages. One is in procurement. The account has been researching your category on review sites since September, and they just hired a Head of RevOps.
Same company. Wildly different brief.
A few other things shift underneath that:
- Duplicate contacts stop being noise. In MQL scoring, the second and third person from a company usually get suppressed. In MQA scoring, they're the strongest evidence you have.
- Anonymous traffic counts. Reverse IP and third-party intent let you score accounts that never filled in anything, which is most of them.
- The failure mode flips. MQL programs fail by producing too much. MQA programs fail by producing too slowly, and by being harder to attribute.
Now, I don't think the MQL is dead, and I'd push back on anyone who says it flatly. Sell a $4,000 tool to one buyer with a company card and lead scoring works fine. Our breakdown of the marketing qualified lead covers where it still earns its place, and SQL vs MQL handles the handoff question.
The MQA earns its keep when a deal needs consensus. That's the whole dividing line.
Why B2B Teams Are Replacing MQLs With MQAs
Three forces pushed this, and they compound on each other.
1. Buying Groups Outgrew the Single-Lead Model
Complex B2B purchases now involve roughly six to ten decision-makers. That figure comes from Gartner and gets repeated by every ABM vendor on earth, which has made people numb to it.
Run the arithmetic anyway. When your form captures one person, you are looking at somewhere between a tenth and a sixth of the group that decides. You then route that one person to a rep and call it a qualified lead.
Forrester reached the same conclusion when it recommended retiring the MQL as marketing's end goal. Signals from multiple group members describe real buying activity far better than one person's score ever could.
If committee structure is new territory for your team, start with B2B buying group.
2. The Conversion Math Stopped Holding Up
Forrester has put inquiry-to-close rates in MQL-led processes at under 1%.
For every hundred inquiries marketing generates, celebrates, and reports to the board, fewer than one turns into revenue.
Channel data says something similar from another direction. MQL to SQL rates in B2B SaaS sit near 13% in some industry datasets and climb past 50% for SEO-sourced leads in others. That spread is the real finding. Nobody agrees what an MQL is, so nobody can benchmark one, so the metric means whatever the person presenting it needs it to mean.
Before you blame scoring for a weak funnel, it's worth checking whether the leak is somewhere else entirely. Sales funnel leakage walks through where deals actually fall out.
3. Duplicate Leads Were Deleting Your Best Signal
Most marketing automation setups suppress the second, third, and fourth contact from the same company. They're duplicates. They clutter the dashboard. Standard hygiene.
But four people from one company engaging inside a month is the clearest buying signal you will ever get. The old model threw it in the bin.
Teams running buying group intelligence do the reverse. Each additional contact raises the account score instead of being merged away.
What Qualifies an Account: Fit, Intent, and Engagement Depth
Every workable MQA model runs on three inputs. Weight them wrong and the whole thing misfires in a way that's hard to diagnose later.
1. Fit
Fit is static. It answers whether this account could ever buy from you.
Score firmographics first: industry, headcount, revenue band, geography, growth stage. Then technographics, which tell you what's already in their stack and what's up for renewal. Add strategic flags last, things like parent company, existing customer, or partner overlap.
Fit should carry the heaviest weight in most models. Engagement from a bad-fit account is noise with good manners. Let intent outweigh fit and you'll spend a quarter chasing companies that were never going to sign.
Build this off your ideal customer profile rather than a wish list, and use firmographics data to pick attributes that actually separate winners from losers.
What to do: Pull your last 50 closed-won accounts. Find the three attributes they share that your closed-lost accounts don't. Those are your fit attributes. Not the twenty sitting in your ICP doc.
2. Intent
Intent is dynamic. It answers whether this account is in market right now.
First-party intent comes from your own properties. Pricing page visits, demo requests, repeat sessions on a product page, webinar attendance. Stronger per signal, but you only see the accounts that already found you.
Third-party intent comes from everywhere else. Research on review sites, surges in topic consumption, hiring for roles tied to your category, funding rounds, leadership changes. Weaker per signal, far wider reach.
You need both, and the split matters more than most teams realize. First-party vs third-party intent data is the clearest way to decide where your budget goes. For sourcing, our comparison of the best intent data providers covers the market.
Most of this activity happens where your analytics can't follow it. That's the B2B dark funnel, and it's larger than your attribution report suggests.
3. Engagement Depth
This is the input that separates an MQA from a well-dressed MQL. It's also the one most teams implement last, because it's the hardest to instrument.
Score three dimensions of it.
Breadth. How many distinct contacts engaged. Two is interesting. Four is a committee forming.
Seniority. A VP opening an email outweighs an intern downloading a whitepaper. Weight by role, not by activity volume.
Recency. Last week beats March. Obvious, and almost universally ignored in practice.
Breadth is the variable I'd watch above the other two. One person can get enthusiastic about anything on a slow Tuesday. Four people can't, not by accident.
Reading buying signals will sharpen what you choose to count, and signal stacking explains why stacking three weak signals beats waiting around for one strong one.
How to Build an MQA Scoring Model
Five steps. Do them in order, and resist every urge to make the model clever.
Step 1: Pick 8 to 12 Attributes
Not fifty.
Every attribute you add is another tuning variable and another thing for a skeptical rep to argue with. Start narrow, prove the model predicts something, expand later if you need to.
A reasonable split is four fit attributes, four intent signals, four engagement measures.
Step 2: Set the Weights
A sensible starting formula:
Account score = (Fit × 0.4) + (Engagement × 0.3) + (Intent × 0.3)
Fit carries more weight on purpose, for the reason above. An in-market account you can't serve is still an account you can't serve.
Revisit the weights after a quarter of real outcomes. If your top-tier accounts don't win at roughly twice the rate of your nurture-tier accounts, the weights are wrong. No amount of extra signal fixes that.
Step 3: Use Tiers, Not One Cutoff
A single threshold creates a cliff. An account at 79 gets nothing, an account at 80 gets a phone call. Tiers create a routing system instead.
On a 100-point scale:
- Tier A, 80 and above. Strong fit, broad engagement, live intent. This is your MQA. SDR outreach inside 24 hours.
- Tier B, 50 to 79. Good fit with partial engagement, or strong intent with thin engagement. ABM campaign plus a sequence. Watch for movement.
- Tier C, 25 to 49. Partial fit, low engagement. Nurture track.
- Tier D, below 25. Poor fit or silent. Passive monitoring only.
Only Tier A is an MQA. Everything else is pipeline in waiting, and calling it anything else is how thresholds drift.
One sanity check I rely on: if more than a fifth of your scored accounts land in Tier A, the bar is too low. Tighten it until sales stops forwarding you complaints.
Routing matters as much as scoring here, so lead routing software is worth sorting out before launch. Tiers B and C need a real lead nurturing strategy behind them, not a quarterly newsletter.
Step 4: Add Decay and Negative Scoring
Two rules almost everyone skips, and both cause the same symptom: scores that look healthy and predict nothing.
Decay. Engagement older than 90 days should count for less. An account that read three blog posts in February and went quiet is not warm. Without a time weight, your score becomes a historical record of who used to care.
Negative scoring. Subtract points for careers page visits, which are job seekers. Subtract for unsubscribes and hard bounces. Subtract for competitor domains. And subtract for contacts who have left the company, which is the one that bites hardest.
That last one deserves a second look. Your account score keeps climbing on engagement from a champion who resigned in June, and nobody notices until a rep calls a dead mailbox. Tracking job changes catches it, and champion tracking turns that departure into a warm opening at their new employer.
Step 5: Write the Handoff Agreement Before You Ship Anything
Get sales in a room and decide four things in writing. What score triggers an MQA. How fast a rep has to act. What context ships alongside it. And what happens when a rep rejects one.
That last question is the one teams forget, and it's the one that determines whether the model improves. A rejected MQA should trigger a model review, not an argument in a Slack channel. Our guide to B2B sales and marketing alignment covers how to run that review without it turning into a blame meeting.
What to do: Run the model backwards on last year's closed-won accounts before you launch. If it wouldn't have flagged deals you actually won, it won't flag the ones you're about to win either.
MQA Metrics That Matter
Volume is the trap. An MQA count on a dashboard tells you nothing by itself, and optimizing for it recreates the exact problem you left the MQL to escape.
Track these instead:
- MQA to SQA acceptance rate. The best single health check you have. Below 50% and your threshold is too loose.
- MQA to opportunity rate. The conversion that pays rent.
- Buying group coverage. How many of the six to ten stakeholders you've actually identified and reached.
- Time from MQA to first meeting. Speed decides more outcomes than scoring precision does.
- Win rate by tier. Tier A should roughly double Tier C. If it doesn't, recalibrate.
- Deal size, MQA-sourced versus MQL-sourced. This is the number that justifies the program to your CFO, so instrument it from day one.
Don't switch off your MQL reporting the week you launch. Run both for two quarters. You'll need the comparison when someone senior asks whether this was worth doing.
Attribution gets genuinely harder once your stages are account-level, which catches teams by surprise. Revenue attribution models is worth reading before your first board deck, and account based marketing metrics covers the broader dashboard.
Where MQA Programs Break
I've watched the same failures repeat across very different teams.
The model scores accounts you can't contact. An account hits 92. A rep opens it. There are four contacts, two have left, one is a role-based inbox, and nobody has a phone number. The score was right. The record was empty.
Stale data inflates everything. Somewhere between a quarter and a third of B2B contact records break every year as people move and companies restructure. B2B data decay explains the mechanics. A scoring model built on decayed records produces confident nonsense, and confident nonsense is worse than no model, because people act on it.
Nobody changed the targets. Marketing still gets paid on volume. So the threshold drifts down, quarter by quarter, until an MQA means what an MQL used to mean. Change the compensation or watch the model degrade inside two quarters.
The model is too complicated to explain. If a rep can't understand why an account scored 84, they won't trust it. They'll go back to working their own list. Simplicity beats precision here, every time.
Anonymous traffic never gets matched. Without reverse IP resolution or third-party intent matching, your model only sees people who already raised their hand. Which is exactly the limitation you were trying to escape.
A 90-Day MQA Rollout Plan
Days 1 to 30. Agree the definition with sales in one workshop, not six emails. Clean your account records. Pick your attributes. Score last year's closed-won accounts retroactively and see what the model would have caught.
Days 31 to 60. Pilot on one segment. One territory, or one product line. Run MQAs alongside MQLs rather than instead of them, and route Tier A to two reps who volunteered.
Days 61 to 90. Check acceptance rate weekly. Adjust thresholds. Document every rejection reason in one place. Expand only after the numbers hold.
Starting small isn't caution. Getting sales to trust the output is the actual deliverable, and trust compounds slowly.
Building this inside a wider program? Account based marketing covers the strategy layer. Signal based GTM covers the operating model once signals drive your routing.
The Data Layer Behind Every MQA Score
Full disclosure before this section: I work at SMARTe. Weigh it with that in mind, then go test the claims against your own records.
Every MQA program I've seen fail, failed in the same place. Not the scoring logic. The data underneath it.
It breaks in three specific ways.
You can only score contacts you know exist. If your CRM holds two people from a ten-person committee, your engagement breadth score is structurally capped. You'll never see the committee form, because you never had the committee.
You can't act on an account you can't reach. A Tier A account with no verified mobile number is a report, not a pipeline.
And scores built on decayed records lie with total confidence, which is the worst property a model can have.
SMARTe runs 289M+ verified B2B contacts and 66M+ company profiles across 200+ countries, with 75%+ US mobile coverage and 50%+ global direct dial coverage. Verification happens in real time at the point of use, not on a quarterly batch refresh. That's what keeps the records under your score honest between cleanups.
On the committee problem specifically, AI agents discover the buying group per account instead of leaving a rep to guess who else matters. That turns engagement breadth from a number you can't trust into one you can. Our guide to platforms for identifying buying group members covers the wider category if you're comparing options.
Enrichment runs at 90%+ match rates, so the accounts in your model carry complete records rather than half-filled ones. CRM data enrichment handles it on a schedule, with Salesforce data enrichment and HubSpot data enrichment both native. RevOps teams report 60%+ less manual work once it's automated.
Bombora intent is built in rather than charged per topic, and funding rounds, leadership changes, and headcount growth feed the same account score. Technographic fit draws on 64K+ technologies tracked.
There's also a Model Context Protocol server, which lets ChatGPT or Claude query verified B2B data live. A marketer can ask for the full buying group at a Tier A account and get real records back inside the chat. No CSV export, no hoping.
Enterprise plans start at $15,000 a year, credit-based, with no per-seat fees. Adding a marketer or an SDR costs nothing extra in license terms. That matters more than it sounds, because an MQA rollout always pulls in more people than you planned for.
See how SMARTe finds verified mobile numbers in your target accounts.
Final Thoughts: Getting MQAs Right
Moving from MQLs to MQAs isn't a reporting change. It's an admission that you were measuring the easy thing instead of the true thing, for years, because the easy thing fit neatly in a dashboard.
The teams that get this right aren't the ones with the cleverest formula. They're the ones who can see the whole buying committee, reach every person in it, and trust the record in front of them.
Everything else is arithmetic on top of guesswork.
