Topline · 28 Jun 2026 · From the week of 22 June
The Godfather of Modern GTM: This Is How To Measure AI Impact | Mark Roberge, Co-Founder @ Stage 2 Capital
These are notes on the conversation, checked against its transcript. The episode itself has the full discussion.
In brief
Mark Roberge, founding CRO at HubSpot and co-founder of Stage 2 Capital, joins hosts Sam Jacobs (CEO at Pavilion) and AJ Bruno (CEO at QuotaPath) to discuss how to measure AI's impact on go-to-market teams. The conversation covers the revenue velocity formula, selling time as the variable AI could most realistically double, ICONIQ data on AI and funnel conversion, and how to frame AI metrics for a board. Roberge's central argument is that teams should stop claiming to be AI-enabled without measuring output, then work back from that output to specific levers such as selling time. The episode also includes a segment in which Roberge argues that fast followers beat first movers more often than not, and that high valuations can trap first movers in ACV jail.
For founders
- Roberge says AI gains in go-to-market have been far smaller than step changes so far, and he expects more evidence by the end of this year.
- Roberge says the research he reviewed shows fast followers winning most of the time, which he averages at roughly two-thirds fast followers to one-third first movers.
- Roberge says first movers raising at very high valuations can be stuck in ACV jail, unable to lower prices because they must grow into the valuation.
- Roberge defines product-market fit as delivering promised value measured by retention, and go-to-market fit as the later proof that customers can be acquired and served profitably.
- Sam Jacobs defines product-market fit as the relationship between customer acquisition cost, retention and gross margin.
For revenue leaders
- Roberge says selling time is the revenue velocity variable AI can most realistically double, moving from a historical best-in-class level of about 30% toward 60%.
- AJ Bruno says QuotaPath has been spot-checking calendars to see whether top performers are spending a growing share of their time with customers.
- Roberge says he would not tie compensation to selling time because it is easy to game, and would use sales contests instead.
- AJ Bruno suggests reading revenue per head together with the share of reps hitting quota, since they show where the go-to-market team is heading.
- Roberge says consistent tenured performers are one of the best indicators for telling a sales quality problem apart from demand gen, macro or competitive causes.
What was said 24, most useful first
Roberge's revenue velocity formula multiplies active opportunities worked, ACV and close rate, then divides by sales cycle length. Listen
He illustrated it with a rep working 20 opportunities at a 10% close rate on $100,000 ACV with a two-quarter sales cycle, giving $100,000 of productivity per quarter. He used the formula to break an output metric down into four variables that can each be targeted.
“It's the number of active opportunities that our rep is working, times the ACV, times the close rate divided by the sale cycle.”
Listen to the episode Metrics & finance Link to this Report a problem
Selling time is the revenue velocity variable AI can most realistically double, since best-in-class teams have historically spent about 30% of their time selling. Listen
He defined selling time as the share of a week a rep spends face to face with a prospect or existing customer. He said he sees a lot of evidence that AI can push that to 60% or more. Holding ACV, close rate, sales cycle, territory and ICP constant, moving from 30% to 60% would algebraically double productivity.
“best in class has historically been about 30%. And I see a lot of evidence that today's AI can push that to 60 % plus”
AJ Bruno cited ICONIQ data showing AI-native companies converting trial to POC to close at 56%, compared with 32% for other companies. Listen
He presented this figure from the same ICONIQ study as the funnel data.
“AI native companies convert trial to POCs to close one at 56 % versus 32 % for everyone else.”
Listen to the episode Sales process & deals Link to this Report a problem
He could not find rigorous studies showing that first movers win most of the time, and that the studies show fast followers winning most of the time. Listen
He said he searched for studies on first mover versus fast follower and found the opposite pattern. Averaging the findings, he said it is about one-third first movers and two-thirds fast followers. He said he looks into this because, as an investor, it is important to know where you have conviction against the consensus that first movers win.
“they all say fast follower wins most of the time”
Listen to the episode Strategy & market Link to this Report a problem
First movers that raise at very high valuations can be stuck in ACV jail, unable to lower prices because they must grow into the valuation. Listen
He said first movers spend heavily figuring out the product and evangelising the category, and can then raise at a very high valuation. That pressure forces them to maximise revenue per deal rather than optimise price.
“where you can't optimize your price because you just have to maximize the crap out of it to grow into that valuation”
Listen to the episode Pricing & packaging Link to this Report a problem
Roberge describes HubSpot splitting customers into six representative cohorts so it could run a monthly NPS study while surveying each customer only twice a year. Listen
He said this worked in HubSpot's first ten years because it was transactional with thousands of customers. Each cohort mirrored the whole base for size, industry and region, so monthly scores reflected new patterns of promoters and detractors. The top detractor reasons went straight to the roadmap, and the team could see whether they moved.
“We split the customers into six cohorts, all of which were representative of the entire base”
Listen to the episode Retention & customer success Link to this Report a problem
Roberge warns that doubling ACV to lift productivity could move a company into a different market. Listen
He said close rate and sales cycle depend on buyer behavior, which teams cannot change easily. He said the number of opportunities a rep actively works is the lever a company fully controls, and the one he thought easiest to double.
“If I 2x ACV, it'll float me up into a different market potentially. So be careful.”
Listen to the episode Sales process & deals Link to this Report a problem
Vendors are far more motivated than buyers to speed up sales with AI, so buying cycles will not speed up at the same pace yet. Listen
He said buying processes will be dramatically enhanced by AI over time, but not at the right speed right now. A host built on the point with the example of a large restaurant brand still buying a POS system through a committee and an RFP, as it did ten years ago.
“I think that's something I'm feeling very much right now is the vendors are highly motivated to accelerate sales actions with AI. The buyers are not feeling that same pressure.”
QuotaPath has been spot-checking calendars to see whether top performers spend a growing share of their time with customers. Listen
He said the analysis covers the go-to-market team, including new business sellers, the solutions team and account managers. He described the trend as qualitative. He said he would discuss it with his CRO before any board conversation, and would not take it to the board right away.
“the top performers are getting more and more, this is again qualitative, more and more selling time as a percentage of their calendar, and the low performers are getting less and less.”
Listen to the episode Metrics & finance Link to this Report a problem
He would not tie compensation to selling time because it is easy to game, and would use sales contests instead. Listen
He recalled reps he hired in 2007 telling him that when phone time was used for comp, reps passed around 1-800 numbers to inflate their phone time even when no customer was on the line. He said his instinct is that it is dangerous to put comp on the metric.
“I'd run sales contests on it potentially, but I think that would be my instinct is like it's dangerous to put comp on it.”
Listen to the episode Sales team, hiring & comp Link to this Report a problem
Roberge's board approach starts from the end goal, breaks it into the revenue velocity variables, and ties AI investment to the variables it is meant to move. Listen
In his hypothetical board pitch he would attack three of the four variables: active opportunities, close rate and sales cycle, leaving ACV alone. He would attack opportunities through selling time, then list the AI initiatives likely to move it and report the impact each quarter.
“I'd start off with like, what is the end goal?”
Listen to the episode Metrics & finance Link to this Report a problem
AJ Bruno suggests reading revenue per head together with the share of reps hitting quota, because together they show where the go-to-market team is heading. Listen
He said revenue per head should rise naturally as sales teams slim down, while the share of reps at quota should also rise as low performers and non-adopters are managed out. Roberge responded that these lagging indicators are extraordinarily hard to attribute directly to AI.
“as sales teams have slimmed down your efficiency, so your revenue per head should be going up naturally.”
Listen to the episode Sales team, hiring & comp Link to this Report a problem
Consistent tenured performers are one of the best indicators for diagnosing whether a sales quality problem exists or the cause lies in demand gen, macro or competition. Listen
He gave the example of a rep who has produced about $250k a quarter for years and then jumps to $400k after AI investments, which he said would suggest something is happening. He later said the board might track the share of reps hitting quota instead, as a more consistent measure.
“you look at your tenured consistent performers to see what's happening. It's one of the best indicators.”
Listen to the episode Sales team, hiring & comp Link to this Report a problem
QuotaPath's new business per rep has grown from about $140,000 last year to about $170,000 today. Listen
He said this is partly AI but not all of it. He said quotas have been raised over the past year and will keep rising, which is also driving ARR per rep higher.
“We were a hundred and forty thousand dollars of new business per rep the first new business on the sales team.”
Listen to the episode Sales team, hiring & comp Link to this Report a problem
About 70 to 80% of QuotaPath's reps hit quota in a given quarter, which he considers close to best in class. Listen
He said this comes alongside quota increases over the past year. He noted that quota capacity, financial plans and incentives all shape the figure, so it is a complicated measure to read.
“we're probably like 70 to 80 % of our reps hitting quota on any given quarter, which I would say is probably close to best in class.”
Listen to the episode Sales team, hiring & comp Link to this Report a problem
AJ Bruno cited ICONIQ data finding that AI lifts lead-to-MQL conversion by about 11% and MQL-to-SQL by a further 8%, while barely moving active deal cycles. Listen
He presented the figures from ICONIQ's state of go-to-market study released in January 2026. He noted that the study used 2025 data, and that AI tooling has changed quickly since then.
“AI clearly lifts top of funnel, 11 % on the lead to MQL and MQL to SQL another 8%, but barely moves active deal cycles.”
The copycat argument that AI makes product development about 10 times faster, so copying a product is much easier. Listen
He presented this as the case made by the fast-follower camp. He said it is one of three or four dimensions of the case, alongside first movers facing higher valuation multiples and recent startups building on architecture that is already outdated.
“product development cycles are 10x faster because of AI. So copying product is just a lot easier.”
Listen to the episode Strategy & market Link to this Report a problem
Harvey's moat is the trust and brand it has among large law firms, which makes it harder to copy on features or price. Listen
He said law firms care less about price and feature optimization and more about what firms like Morgan and Morgan use. If Morgan and Morgan trust Harvey, other firms will buy it regardless of competition. He tied this to Porter's brand moat, where buyers purchase because of the brand.
“your moat is that there's this recognition of Harvey as being a trusted product for all the big law firms.”
Listen to the episode Strategy & market Link to this Report a problem
Roberge describes a hypothetical in which a $100,000 ACV company that raised at a high valuation would struggle to cut its price to match a copycat charging $15,000. Listen
He described an incumbent with about $100 million in revenue at a $100,000 ACV, facing two founders offering a similar product for $15,000. He said it would be very difficult for the incumbent to drop its ACV by 70%.
“It would be very difficult for them to drop their ACVs by 70%.”
Listen to the episode Pricing & packaging Link to this Report a problem
Roberge defines product-market fit as creating the value promised to the market, measured by long-term retention and leading indicators of retention, before unit economics are considered. Listen
He said product-market fit should be proven first, then go-to-market fit should show the company can acquire and serve those customers profitably. He said he prefers founders to focus on product-market fit in an unscalable way rather than on quotas and pricing.
“I talk about as creating the value that you promised for the market as measured by retention in the long term”
Listen to the episode Strategy & market Link to this Report a problem
Sam Jacobs defines product-market fit as the relationship between customer acquisition cost, retention and gross margin. Listen
He said payback period and LTV to CAC determine product-market fit. He argued an AI-native company with 30% gross margins could get many referrals and still be out of business, so referrals alone do not establish fit. AJ Bruno said he did not think gross margin belonged in product-market fit, while Roberge said both answers were correct depending on how the scope is drawn.
“I think product market fit is the relationship between customer acquisition cost and retention and gross margin.”
Listen to the episode Metrics & finance Link to this Report a problem
Sam Jacobs describes the main AI use case in go-to-market as administrative work such as call recording, note-taking and follow-up drafting. Listen
He listed pre-call prep, post-call follow-up and CRM updates, saying pre-call prep and after-call follow-up could become much quicker, maybe doubled. Roberge said the framing made sense to him and that these administrative gains feed into selling time.
“So what I hear you saying is the AI use case and go to market is the administrative work.”
Boards should be shown measured AI impact rather than a general claim that the sales team is AI-enabled. Listen
He said he has grown tired of board meetings where management says the team is AI-enabled, and when asked how it knows, the answer is often that reps write emails with ChatGPT. He argued that companies need to measure it.
“I just got fed up of going to these board meetings this year”
He has been disappointed by how little AI has moved go-to-market performance so far across his portfolio and the wider ecosystem. Listen
He said that when operators such as Kyle Norton put out their AI work and he calls them to see it, even they admit some of what they describe is still aspirational. He said he thinks more evidence should appear by the end of this year.
“It's a disappointment of across our entire portfolio and it's not I think it's not our portfolio It's just the entire ecosystem.”