The Gumball Machine Is Broken: Jon Miller on What Comes After the MQL

About this episode

Most B2B marketing still runs on a single number: the marketing qualified lead. Jon Miller is one of the few people who can tell you where that number came from, because he helped build the system that produced it — first at Marketo, where he helped create the marketing automation category, then at Engagio, then at Demandbase.

What makes this conversation different is that Jon went back and diagnosed his own creation. He’s not quietly onto the next thing. He’s saying, out loud, what the MQL got wrong about how people actually buy — and he’s careful to credit what it got right before he takes it apart.

The short version: roughly 95% of buyers have built their shortlist before they ever talk to a seller. The MQL was designed to catch the last 5% who raise their hand. So the real question isn’t how to optimize lead capture. It’s what you do with everyone who isn’t ready yet — the 95% the old model was built to ignore.

We get into why buying behaves more like weather than a vending machine, the three-tier model Jon uses instead of MQLs, why he thinks legacy automation tools can’t keep up, and how the best CMOs are quietly rewiring what they report to the board. If you’ve ever felt like you were pedaling into a headwind running the playbook that used to work, this one’s for you.

About Jon Miller

Jon Miller founded Marketo in 2006 and helped define the marketing automation category. He went on to found Engagio, which was acquired by Demandbase in 2020, served as CMO at Demandbase, and is now building Phave, an AI-native marketing automation platform.

Chapters

00:00 Introduction to Jon Miller and his journey
01:24 Diagnosing the MQL model
03:27 The gumball machine / nonlinear buying idea
07:23 What the MQL got right
10:14 The three-tiered model of engagement
14:22 The role of CMOs in modern marketing
18:17 AI’s impact on marketing automation
19:55 The Spotify playlist analogy
22:53 The Peppers and Rogers/one-to-one thread
24:43 Common mistakes moving off the MQL
25:25 The three CMO dashboards
27:25 Advice for CMOs making the shift

A few things worth taking away

• The MQL started as a good idea — a contract between marketing and sales — and got gamed over time as teams chased volume.
• Buying isn’t linear. With six to sixteen people on a buying committee researching in places you can’t even track, “run a campaign, get a lead” no longer describes reality.
• Hand raisers are the gold standard, but waiting for them means you only ever talk to the 5% who already built their shortlist without you.
• Jon’s three tiers — hand raisers, MQX, and MEX — give you a way to work the 95% instead of ignoring them.
• When you move off MQL volume as your headline metric, expect the numbers to drop before quality and conversion rise. Set that expectation early, or you’ll hit a buzzsaw.
• The strongest CMOs report pipeline across all sources to the board and stop fighting over who sourced what.

A few lines that stuck with me

“Put your quarter in, get your gumball out. Put your campaign in, get your MQL out. I just don’t think that’s the way buying works.” — Jon Miller

“If you only wait for somebody to raise their hand, you’re talking to the 5% in market. And they’ve already built their shortlist without you.” — Jon Miller

“You can’t get there with a rules-based system. You just end up with spaghetti.” — Jon Miller

Resources mentioned

• The B2B CMO Project — research on the strategic CMO and the three-dashboard model
• Mike Bosworth, Solution Selling
• Don Peppers and Martha Rogers, The One to One Future
• Kathleen Schaub, Marketing in the Great Big Messy Real World

Transcript

Brian Carroll (00:05) Welcome to The B2B Roundtable, where we go inside the ideas, people, and decisions shaping modern revenue teams and how they actually work. I’m Brian Carroll, and today my guest is Jon Miller.

I first met Jon way back in 2006, when he founded Marketo and helped build the marketing automation category as we know it today. In 2015 he founded Engagio, which was acquired by Demandbase in 2020. Now he’s building Phave, an AI-native marketing automation platform.

Here’s what makes this conversation different from other podcasts you’ve listened to: Jon didn’t just build the next thing and quietly move on, the way a lot of founders do. He’s gone back and started diagnosing the problems with something he previously created. He’s talking about what’s wrong, and why it’s failing buyers today.

And here’s why it matters right now. Before they ever talk to a seller, 95% of buyers have already designed their shortlist. The MQL is built to capture the last 5% who self-identify. What about the 95% who haven’t yet?

So, Jon — when did you first start thinking the MQL model was broken, not just underperforming? How did you get there?

Jon Miller (01:24) It started, more than anything else, during my time at Demandbase. After we merged Engagio and Demandbase together in 2020, the first thing I did was help the product team unify the two platforms. But then in 2021, I took over as CMO.

And I had my playbook. This is how I do it: I create definitive guides, big, rich, meaty pieces of content. You run lots of other thought leadership, like webinars, and you generate leads from all of it. Most of those leads won’t be ready to buy right now, and that’s okay — that’s why you nurture them and score them. You know a little something about that. Then eventually, when they’re ready, you pass them to sales. That was the playbook, and it’s the playbook I ran at Marketo.

To a degree, it’s the playbook I ran at Engagio too, although there we also layered on an account-based motion that we’ll get to. So here I am at Demandbase, running that playbook, and the exact same tactics that worked for me at Marketo just weren’t working.

At Marketo, it felt like I’d had a tailwind pushing me forward, making everything work better. At Demandbase, it felt like bicycling into a headwind. That’s what got me thinking: okay, what’s going on here? Over time, I diagnosed multiple problems — like most complex things in the world, there were many reasons it wasn’t working.

Jon Miller (02:56) But more than anything else, it came down to three things. One, buyer saturation. Two, the fact that the traditional model missed important things like brand. And three, the fact that the MQL is really focused on people, not accounts. We can dive into any or all three of those.

Brian Carroll (03:15) I want to understand what you noticed was broken first. As you’ve reflected on it and done the research — what are we getting wrong about how buyers buy today?

Jon Miller (03:27) Let’s start with the core philosophy behind the MQL: that you can run a campaign and get a meaningful response that’s valuable on the other side. That’s how we thought of it at Marketo. If I needed more MQLs, the natural response was, well, let’s run more campaigns.

It trained us to think of buying like a gumball machine. Put your quarter in, get your gumball out. Put your campaign in, get your MQL out. And I just don’t think that’s the way buying works. Arguably, in the early days of Marketo — simpler buying committees, heavy demand, lots of latent need for our product — okay, maybe you could argue there were elements where it worked then.

But fast forward to today, and buying is much more complex. There are six to sixteen members of the buying committee, not one person. And as you said in the intro, that whole committee is going through a complex set of research — happening not just on our website, but increasingly off it, in closed communities and in conversations with AI agents, all invisible to traditional tracking.

When you have that kind of complexity, the model of marketing as a simple linear gumball machine starts to break down. Kathleen Schaub coined a really good term for this, which connected to my math and physics background. She called it “marketing in the great big messy world,” and she pointed out that marketing is actually a complex, nonlinear process — not a simple linear gumball machine.

Jon Miller (05:19) I studied complex nonlinear processes in college, and it turns out that’s the origin of what’s now called chaos theory. The weather is a complex nonlinear process. The stock market is a complex nonlinear process. And these processes are known, among other things, for their unpredictability — their sensitive dependence on initial conditions. The idea that a butterfly flapping its wings in Brazil can cause a hurricane in Japan. Most people have heard that one.

If you embrace the fundamental idea that buying is just as complex as the weather, then it’s an impossible task to say, “I’ll run this one campaign, and that will lead to buying.”

Brian Carroll (05:46) That’s right.

Jon Miller (06:03) Or, “Where did this deal come from?” “Well, they stopped by the booth at the trade show.” No — it’s a much more complex system than any of those simple explanations can really capture.

Brian Carroll (06:16) The gumball machine analogy hits on something people are really struggling with. Attribution. The MQL has been elevated all the way to the board — board members and CEOs care about it because it’s a visible KPI. And there are a lot of misses in how we think about it, because we don’t actually know how many MQLs become real customers. Partly because of what you just described about how buyers buy.

You wrote something on LinkedIn about a three-tier model — this marketing-engaged layer, where people are consuming content but not showing buying signals yet. Most demand gen teams would say those aren’t worth chasing, because there’s no buying intent yet. Can you make the case for why that’s wrong? Why is that exactly where the fight is being lost?

Jon Miller (07:23) It’s worth starting by saying there was some real goodness in the original concept of the MQL. Specifically, it was a contract between marketing and sales. Marketing said, “I’m only going to pass you things that reach this bar, where there’s strong evidence this is worthy of sales attention.” And sales said, “Okay, I commit to this service-level agreement for follow-up.” That was a genuinely good thing about the MQL.

The problem is that over time it got bastardized. Under pressure to hit pipeline targets, some marketing teams gamed the scoring thresholds. I saw so many companies basically say any responder to any campaign was an MQL. And I’d think, that’s not what it was.

That behavior — driven by the desire for more MQLs — is exactly what led sales to start cherry-picking and ignoring most of them. So what was sales cherry-picking? Hand raisers.

Jon Miller (08:31) I’ve talked to some CMOs who say that’s the only thing they report now: hand raisers. People explicitly asking for a sales connection. And that makes sense — these are people you want to talk to. But I think it’s a hundred percent too passive.

Jon Miller (08:49) And it’s too late. If you only wait for someone to raise their hand, you are by definition only talking to the 5% in market. Those people have already created and aligned on their shortlist without you, which means you’re column B, fighting an uphill battle at best.

Mike Bosworth wrote Solution Selling a while ago, and there’s a lot of wisdom in that old book. The idea of solution selling is that helping a buyer see pain they haven’t prioritized — bringing latent pain into an active evaluation — is really valuable. And if you can be the vendor guiding that process, you help shape the buying criteria. So the question is: how do we make that happen?

One way is an investment in branding. Building a brand that creates urgency around the problem you solve, builds a connection between your company and the ability to solve that pain, and generates positive feelings around that connection — that’s incredibly valuable. We could do a whole other podcast on branding.

But there’s the question of how you tie solution selling together with the goodness the MQL brought to the table. Because solution selling gone wrong just becomes cold calling, and that’s not good for anybody.

That’s why I came up with the three-tier model. The top tier is hand raisers. Let’s all agree that’s the gold standard — it’s what sales wants more than anything, and we should track it. But then there are two other tiers. Tier two I call MQX.

The X is important, because making the X stand for “lead” usually isn’t the right answer — unless you have a low-value, highly transactional purchase that one person can make. Most of the time, it’s a more complex buying committee. So I’d generally say tier two should be MQA — marketing qualified account — or even MQBG, marketing qualified buying group. As a sidebar, MQBG is a mouthful, so I usually drop the M and just call it a QBG.

What these all have in common is some signal that there’s a good chance this account or buying group is actually in an evaluation stage — starting to form their shortlist or consensus, maybe they already have. You’re not waiting for them to raise their hand. If you can reach out to those companies at the right time…

Jon Miller (11:46) …in the right way, it can be valuable. The key is that MQX does not mean this person is ready to buy. It means marketing believes, based on the data, that this account might be in market. There’s an interesting debate about whether we should even call it “qualified,” because it’s not qualified the way a salesperson uses that word.

I like using it because it’s a familiar mental model, but you could also call it a marketing recommended account or a marketing indicated account. If the word “qualified” carries baggage at your company, fine — use something else. Then there’s my third tier.

Jon Miller (12:27) I call it MEX — marketing engaged account. This is really the 95% that aren’t in market. You could cold call into that 95%, but the idea of MEX, as opposed to just your target account list, is that it’s someone from your target list who’s also engaging with your brand and ideas. There’s some level of engagement, even if there are no buying signals. They’re showing interest in your topic, even if not intent to purchase. Odds are that’s a warmer outreach than a truly cold call.

But 100%, do not reach out to that person and pitch a demo or a sales meeting. This is about what Bosworth calls solution selling — helping them understand and quantify the cost of the status quo, and creating hope about what the future could be, maybe by sharing examples of what other companies are doing.

That’s not how typical SDRs reach out today. It’s a one-to-one way of building brand and awareness. And a person with “sales” in their title might not be the right person to do it — which is why some companies have MDRs, market development reps. But that means different economics and different compensation. You can’t pay an MDR on this quarter’s pipeline if what they’re doing is planting seeds for a year from now.

Brian Carroll (14:13) That’s right.

Jon Miller (14:13) So that was a lot of framework. But I think hand raisers, MQX, and MEX make a lot of sense.

Brian Carroll (14:15) A lot was going through my head listening to that. You brought up brand, you talked about engaging differently, and you touched on how SDRs work. The current playbook has an SDR follow up on a scored lead, incentivized one way: get a demo, schedule an appointment. So they’re focused on the last mile, the end of the buying process — but the majority of leads they talk to are nowhere close to that. They’re in the early stages.

You’ve been digging into this through CMO dinners and conversations. How are CMOs responding right now? What are you seeing and hearing, and what do you think teams need to do to get from where they are to where they need to go?

Jon Miller (15:21) I see a wide variance in how marketing leaders are responding to these pressures. This is a dramatic oversimplification, but you can bucket most marketing leaders into either a strategic CMO or a tactical CMO.

The tactical CMOs, whether they want to be or not, work at companies that primarily view marketing as pipeline generation. They’re pressured into the traditional playbook and traditional metrics — MQLs, marketing-sourced or marketing-influenced pipeline. These CMOs will give the right kind of acknowledgment — “I know the MQLs aren’t right, and I’m reporting on these other things too” — but they still feel like, “Yeah, but I have to report the MQLs.”

The strategic CMOs are the ones elevating the role. First and foremost, they’re executives of the company who bring their understanding of the market and the customer, so they’re in the room when strategic discussions happen. It’s the concept of the CMO as chief market officer, not chief marketing officer. Those CMOs are driving conversations about the changing buyer, the importance of brand, and the need to think long-term, not just short-term.

Jon Miller (16:54) That said, they’d all agree that pipeline is permission. Even a strategic CMO, if they’re consistently missing pipeline, doesn’t get to go invest in the big new brand project. They get that. But how they tackle it, and how they talk about it, feels a little different. That’s probably the biggest delta — and it runs through into the dashboards they present, how they interact with their peers, and how often they talk to the CFO and about what.

Brian Carroll (17:29) It sounds like the tactical CMOs are more subject to that gumball-machine dynamic — MQLs, driving demand, generating the numbers. And for the strategic CMO, it’s not that they ignore those things; demand generation still matters. But they’re thinking bigger picture about strategy, like how important brand is in B2B.

I want to pivot to one of the challenges all of us are dealing with right now — not just brand, but the technical environment we’re operating in. I’d like to talk about how AI is changing what’s possible. Whether you’re a strategic or tactical CMO, it has a huge influence. What can you do differently for these marketing-engaged leads, for example? How might we approach that?

Jon Miller (18:35) Part of the problem is that tools like Marketo — which I obviously helped create — were built around the mental model of the MQL, of marketing as a linear, simple buying process. In rules-based platforms, very much if-this-then-that, you end up with static nurture paths. You’re lucky if you have two or three paths for two or three personas, let alone understanding the right thing for each person in each account.

When you embrace the modern buying process, where 95% aren’t ready to buy, you need to create latent pain, stay in touch with that 95%, and then catch the signals when they might be becoming qualified — pre-hand-raiser, but ready to reach out. The legacy tools like Marketo just can’t keep up with that. The nurture tracks are too rigid and too limited for engaging the broad market before they’re ready to buy. And they’re too email-centric for when you don’t have permission for a huge fraction of that database.

The analogy I like is that instead of putting people in specific nurture tracks, we want to create a personalized playlist for each person. I like playlists because people understand Spotify. There’s a whole library of songs that could play at any time. But…

Brian Carroll (20:19) That’s right.

Jon Miller (20:20) …what their AI does is think about me — what I’ve liked and listened to, what I’ve engaged with, what I haven’t listened to in a while — and it builds a playlist for me. Even if you and I both like 80s songs, we’ll get different playlists, because we’re different people who’ve engaged with different things. Every single person on Spotify gets a completely unique, personalized playlist.

Can we use AI to apply that same idea to marketing? For the 95% that aren’t in market but that I need to engage over time, let’s craft a personalized playlist for each of them — based on who they are, where they work, what else is happening across their buying committee, and what we know about them.

If we have opt-in permission, that playlist should include email touches — but it won’t always. Sometimes it’ll involve advertising. It might involve LinkedIn touches. It could be a whole variety of ways to interact, by picking the right offer, the right channel, the right content, and the right time.

So legacy marketing automation is rule-based and list-based. It’s too person-based, meaning you can’t really build playlists that look at the account level and go multichannel into advertising.

Jon Miller (22:01) One other thing, which we haven’t talked about: our whole conversation so far has focused on net-new business — acquiring the new customer, the new account. But a lot of these concepts — hand raiser, MQX, MEX — also apply to post-sale, especially for expansion into new buying committees, product qualified leads or accounts, and product adoption campaigns. The legacy tools don’t handle any of that well either, and new AI-enabled approaches can. Which, not surprisingly, gives you a pretty big hint at what I’m trying to build at Phave.

Brian Carroll (22:28) So Phave — that’s what you’re working on right now. AI-enabled marketing automation to deliver these personalized, one-to-one journeys. It sounds like the promise is kind of like what I was reading way back in Peppers and Rogers’ one-to-one marketing, except now we can actually do it — the way you described with the playlist. Anything you’d add?

Jon Miller (22:54) It’s funny you bring up Peppers and Rogers. In the intro, you said I’m an entrepreneur who keeps coming back and revisiting what I’ve done before. The way I describe my journey — from Epiphany, the company before Marketo, where I wasn’t a founder, to Marketo to Engagio and now Phave — I’ve been on a journey to deliver on the one-to-one future.

Each of those companies would have said we’re trying to do one-to-one marketing. Each got us closer, but not quite. What I’m so excited about, living in 2026 in the age of AI, is that I think we’ll finally be able to deliver on what truly is one-to-one marketing, as envisioned by Peppers and Rogers back in 1992.

Brian Carroll (23:47) That vision has been around a long time, but with every iteration we’ve struggled to get there. There was always some limitation. And it seems like right now, the promise of AI to actually deliver on it is real.

Jon Miller (24:02) You can’t get there with a rules-based system. You just end up with spaghetti.

Brian Carroll (24:05) Yeah. Anyone who’s built very complex nurturing journeys knows there’s an eventual breakdown. The more personalized you want to get, there are only so many if-then statements and branches you can build before it gets too complicated — and it still doesn’t do what you’re trying to do.

For listeners thinking, “Okay, what does this mean for me right now?” — how can someone move their organization away from MQL logic? And what are the most common mistakes you’ve seen teams make when they try?

Jon Miller (24:43) The common mistakes. First, doing it alone, without full alignment and buy-in from the rest of the executive team. The single best thing a CMO can do is enlist the head of sales — and the head of post-sale, if they exist — and propose the new set of metrics as a team. That’s number one.

Number two, when you start changing these things, set the expectation that quantity is likely to go down. The quality will go up and the conversion rate will go up. But if people don’t expect the raw numbers to be lower, you can run into a buzzsaw and get into trouble.

What I recommend is that CMOs be very thoughtful about their measurement architecture. We actually published a research report about the strategic CMO on the B2B CMO Project website, where we talked about three types of CMO dashboards.

The first is what you show the board. At the board level, these should be business outcomes. First off, the best strategic CMOs own pipeline — they report on pipeline created, new and expansion, across all sources. That last piece is the key. You’re not reporting marketing-sourced or sales-sourced; you’re reporting on whether there’s enough pipeline for the business. Because at the end of the day, if there is, the board doesn’t care who sourced it. And if you believe in complex nonlinear buying, trying to track where a deal came from is a fool’s errand.

From there, marketing can share in reporting other key metrics — opportunity win rates, net revenue retention, customer acquisition cost. The last board-level one is marketing efficiency: total pipeline generated per dollar of marketing investment. That’s your top level.

Your second level is CEO- and CFO-level indicators. This is where things like hand raisers and MQAs kick in, along with pipeline quality and brand health. Then you save everything else for the marketing team’s operational metrics — including MEX, that third tier: market development metrics, account engagement, program performance. That’s all important, but it doesn’t belong in your CEO- or board-level reports.

Brian Carroll (27:25) If you could go back and have someone give you advice on what to do differently — what advice would you give a CMO who’s ready to make this shift?

Jon Miller (27:36) Some of what I already said — building a committee, working with your peers — is really important. But if I had to narrow it down to one thing, it’s this: earn your seat by being a business executive first.

Surface problems, in marketing and in the market, before other people do. That builds credibility. Spend a lot of time with your peers — the CRO, the CFO. Understand their challenges. Then you’re in a better position to demonstrate how marketing helps them, not just marketing.

And own the customer voice. I alluded to this earlier, but it’s the single most powerful thing the chief market officer can bring to these strategic discussions. If you’re a strategic CMO who’s established that level of credibility, people aren’t going to question the metrics you report — whether it’s MQLs, MQAs, or anything else — because they know you’re a strategic part of the leadership.

Brian Carroll (28:42) Excellent advice, Jon. We’ve covered what I hoped we would today. This matters so much, because as the buying dynamics have changed, people need to approach things differently. What you’ve shared is going to help a lot of people start thinking about how to address the 95% who aren’t in market.

We’ll share the resource Jon mentioned in the show notes. Jon, thank you for joining us today. I’m excited to see how things develop as you build Phave, and I’m really glad you could be on the show.

Jon Miller (29:15) Thank you. It’s been a pleasure.