Topline · 20 Aug 2026 · From the week of 17 August
SPOTLIGHT: He Missed Quota 5 Times in 25 Years. Here's the System Behind That.| Mike Carpenter, CEO & Co-Founder @ Xfactor
These are notes on the conversation, checked against its transcript. The episode itself has the full discussion.
In brief
Mike Carpenter, CEO and co-founder of Xfactor and former president of global sales and field operations at CrowdStrike, walks through the sales and operating methods he used across his career and the Xfactor AI software platform he now sells. He covers how he found measurable differentiators he called X-factors, how he staffed large operations teams, how warming territories cut rep ramp time, and why he is critical of MQLs. His central argument is that AI run directly on CRM data gives confident but unreliable answers, and that causal modelling and human subject-matter checks are needed.
For founders
- Mike says that when his team ran Claude on the same dataset Xfactor uses, it hallucinated answers and delivered them with certainty, so he argues causal AI with human checks is needed.
- Xfactor gives findings it cannot fully back up a 60 percent probability rating so users test them before acting on them.
- Mike argues an operating plan must be always on and adjustable, since his spring-built plan broke after the CFO reset targets in October.
- Mike says Xfactor now shows customers its calculation methodology so they can check results, as part of working on consumer confidence, which he called the biggest challenge.
For revenue leaders
- Mike says warming territories before a new rep starts, then placing a seasoned rep into them, moved his ramp time from about a year and a half to 90 days.
- Mike released territories to new hires only once their virtual pipeline reached about 3x, built through marketing outreach.
- Mike says the common six-month onboarding answer from CROs ignores whether territories were warmed before the hire started.
- Mike stopped judging marketing by MQLs and counted touches instead, estimating it normally takes around 16 touches to warm an account.
- Mike says reps who chose their own X-factor experiments picked the easiest ones and hit 10 for 10 without changing how they sold.
What was said 21, most useful first
Warming a territory before a new rep arrives cut his ramp time from about a year and a half to 90 days. Listen
Mike said the theoretical six-month ramp was really closer to a year and a half in his experience. He moved ramp to 90 days by warming accounts through marketing outreach so they understood the problem before a seasoned rep entered. He said the accounts had not been physically touched or talked to, but they understood the problem and could resonate with the issue.
“That would move my ramp time from the theoretical six months was probably more like a year and a half to 90 days.”
Listen to the episode Sales team, hiring & comp Link to this
He released a territory to a new hire only after its virtual pipeline reached about 3x through marketing outreach. Listen
Mike said he ran a combination of marketing outreach into a set of accounts, treating the warmed set as a virtual territory. Once that virtual territory reached 3x pipeline, he cut it and ranked which territories could take new hires.
“Once I got a patch to a 3x pipeline that was a virtual patch that was being done through a combination of marketing outreach efforts, then I'd cut that patch and I'd start ranking which patches could take my new hires.”
Listen to the episode Sales team, hiring & comp Link to this
Talk-to-listen ratios from call data showed correlations with daily movement from stage two to stage three. Listen
Mike said his team ingested conversational intelligence and ran AI against CRM, outreach and SDR pitch data. They looked for keywords and for the balance of time spent talking versus listening. He said they found correlations between these measures and daily stage two to stage three movement, which he called one of the big breakthroughs.
“trying to find the key words and how much time on offensive talking versus time on defensive listening”
He argues MQLs are the wrong metric because a marketing source cannot be reliably traced to a single campaign. Listen
Mike said he is not a fan of MQLs and described a marketing team paid a spiff to tag deals as marketing-sourced, which he stopped because it did nothing for lead generation. He argued a lead could have come through friends, website visits, conferences or calls, so attribution to one email campaign is not knowable. He instead counts how many touches it takes to qualify and warm an account.
“Can you really tell a marketing qualified lead came in through an email campaign? You can't because it could be a friend they talked to”
Listen to the episode Pipeline & demand generation Link to this
He estimates it normally takes around 16 touches before a prospect recognises a problem. Listen
Mike said he measures how many touches it takes to get an account to recognise the need, and that normally this is around 16. He stressed these are correspondences rather than just emails. He said the number is a lot and that his team used the analysis to direct marketing dollars toward accounts where that recognition was building.
“It's not 16 emails, it's 16 correspondence.”
Listen to the episode Pipeline & demand generation Link to this
Xfactor drops findings it cannot verify to a 60 percent probability rating so users test them before acting. Listen
Mike said Xfactor runs many checks on each finding, and if it cannot back a result with certainty it drops the rating to 60 percent. He said that rating tells the user to look at the finding carefully and test it slowly. He said removing statistical error from the system was one of the company's biggest issues.
“if we can't back it up with running tons of different sum checks about it with certainty, we'll drop it to a 60 % probability rating”
The common six-month onboarding answer from CROs is misleading when territories are warmed before a hire starts. Listen
Mike said nearly every CRO gives the same six-month answer for how long it takes to onboard a rep, regardless of company. He argued that figure does not hold once territories are warmed in advance. In his own approach, ramp moved to 90 days once warming came first.
“It's like the common lie. They all say, how long does it take to onboard a rep? Six months it's the same answer for everybody no matter what”
Listen to the episode Sales team, hiring & comp Link to this
Static annual operating plans break when mid-year targets and budgets change. Listen
At CrowdStrike, when the business was about a half billion dollars, Mike started planning in May and submitted in September. The CFO returned in October asking for 15 percent more revenue and 10 percent less budget, and spreadsheet errors cascaded through the plan. He said budget came in December, so hiring did not start until February.
“I'd submit it around September. And then by the end of October, the CFO would come back and say, we need to do 15 % more and you need to cut the budget 10 % more.”
Hiring for international markets has to start about 90 days before the seat is needed, because candidates must leave their current employers. Listen
Mike said selling in Europe requires a 90-day window to find a person, have them leave their company and join. Since his budget did not arrive until December, he would have needed to start hiring in October to reach full international capacity. He said his hiring slipped because it could not begin until budget was approved.
“if you're selling in Europe, you need a 90 -day timeframe when you find a person for them to leave the company and come to join you”
Listen to the episode Sales team, hiring & comp Link to this
Reps who chose their own measurable experiments picked the easiest ones and hit 10 for 10 without changing how they sold. Listen
Mike said reps treated X-factors like a quota, adding a little grit and tracking to make them easy to hit. He said this left the game unchanged. He spent years trying to teach reps that the goal was a different way to win rather than arguing the quota.
“So they'd go 10 for 10 and I'm like, this isn't changed in the game.”
Listen to the episode Sales team, hiring & comp Link to this
He ran measurable experiments in 90-day cycles, dropping those that did not work and scaling those that showed a trend. Listen
Mike said each X-factor had to be measurable, so activity goals like running a set number of lunches did not count. Every 90 days he crossed off what was not working and failed it fast. When an experiment trended positively he shared it with the team and had everyone doing it.
“they all had to be measurable. And every 90 days, I'd cross ones off the list that weren't working, and I'd fail those fast.”
He hired a former Goldman Sachs financial analyst to scan sales numbers and activity for anomalies. Listen
Mike said he recruited an MIT graduate who had been a financial analyst at Goldman Sachs into a sales organization. Her job was to go through all of the team's numbers and every activity to find anomalies. He said the anomalies were what separated making quota from missing it and growing pipeline from failing.
“her job was to rip through all of our numbers and every activity we did and find anomalies.”
He staffed operations groups well beyond the size of comparable teams at other companies because he said the analysis drove results. Listen
Mike said his operations groups were abnormally large compared with the rest of the organization and with peers at other companies. At CrowdStrike he said he argued with the board constantly about how many analysts he needed. He credited this staffing with the difference in results over his career.
“I had really large ops groups abnormally large compared to the rest of the organizations that existed around me”
He tied product usage data to churn and fed the results back into onboarding and enablement. Listen
Mike said he used Heap to track how users used the product, then measured churn against usage frequency and the features used. He said they also looked at how customers judged return on investment. The findings were tied back to the first sales call and to enablement, which he said affects how quickly new reps can be onboarded.
“understanding churn rate against that. How often are they using it?”
Listen to the episode Retention & customer success Link to this
In his team's test, Claude run on the same dataset as Xfactor hallucinated answers and presented them with certainty. Listen
Mike said his team ran Claude on the same dataset they use and it completely hallucinated an answer. He said it did not simply drift but delivered the answer with certainty. He called this confident wrong answer his biggest competitor right now, pointing to companies that put an LLM on top of their CRM.
“we've taken Claude and just thrown it on the same dataset we have, and it completely hallucinates an answer, but it doesn't just hallucinate or drift an answer. It gives it to you with certainty.”
He predicts companies running an LLM on their CRM data will pivot on wrong answers and only notice when numbers fall. Listen
Mike said an AI that gives answers with certainty will lead companies to pivot their business in ways they do not know are wrong. He said they will not find out until numbers start dropping. He described most of the calls his company gets as companies that added a cloud model to their data without a data lake or tests.
“you're going to pivot your business in a way that you're not going to know what's wrong until your numbers start dropping”
He argues causal AI that maps cause and effect, with heavy human involvement, matters more than putting an LLM on the data. Listen
Mike said a general LLM will index Slack, email and documents and learn wrong data. He said the approach needs a co-pilot or digital twin that maps potential links, with a large human intervention piece. He said Xfactor has staffed up its intelligence work so the system can learn those cause-and-effect relationships.
“You need causal AI, and you need a co -pilot, a twin that starts mapping potential links.”
He says LLMs generate new query code each time and do not keep what they learned, so rephrased questions get different answers. Listen
Mike said systems like Claude generate new code every time they run a query without pulling results back into memory to learn from them. He said a slightly different wording can produce a different answer, delivered with certainty. He presented this as a reason the output cannot be trusted without checks.
“It's not pulling anything back into memory and learning from it”
AI's persuasiveness and certainty are not accuracy, so subject-matter experts must challenge outputs. Listen
Responding to Mike, Sam said the primary feature of AI is persuasiveness and certainty, not necessarily accuracy. He said that is why subject-matter experts are needed to say when an output does not work or is not correct.
“The primary feature of AI is persuasiveness and is certainty. It is not necessarily accuracy.”
Xfactor publishes its calculation methodology to customers, who often first doubt results against their own spreadsheets. Listen
Mike said the first question from clients is usually that the results do not match their spreadsheets. He said Xfactor now puts its methodology in front of customers so they can manually check how the results are calculated. He said building consumer confidence was the biggest challenge.
“But now we put it in the methodology so they can actually go manually look at the methodology that we use to calculate those results.”
He argues an operating plan cannot be built once and forgotten. Listen
Mike said micro and macroeconomic changes blow up plans, leaving leaders flying blind. He described Xfactor as building an always-on operating plan that adjusts and surfaces hidden revenue opportunities. He said this is the problem he set out to solve after leaving CrowdStrike.
“An operating plan can't be built once and forgotten.”