From $0 to $415M: The 5 Things I'd Do Differently with AI

From August 2010 to September 2012, I was Regional VP at Groupon. We went from zero to 400+ sellers. We launched in 23 new markets. We generated $415M+ in revenue from my region.

It was the most intense professional experience of my life, and I'm proud of what we built. I'm also old enough now - and honest enough - to admit that a significant amount of what we accomplished through sheer force of will could have been accomplished faster, with fewer mistakes and fewer casualties, if I had the tools that exist today. We ran on caffeine, whiteboards, and collective stubbornness. It worked. It was also wildly inefficient in ways I could only see in retrospect.

This is not a nostalgic post. This is a practical one. Here are the 5 moments where AI would have changed the outcome.

1. Hiring: Screening 500 Resumes Took 3 Weeks. AI Does It in an Hour.

When we were scaling into new markets, we weren't hiring one or two salespeople at a time. We were hiring 15–25 per market, across multiple markets simultaneously. At peak, my recruiting team was processing 400–600 applications per week.

The screening process was manual and brutal. Resume review, phone screens, scoring against our criteria. We had three recruiters working essentially full-time on this and still created bottlenecks that slowed market launches. We launched markets two weeks late more than once because hiring was behind.

With current AI tools, 500 resumes screened against specific criteria - track record, industry background, quota attainment, tenure signals - takes under an hour. You build the scoring rubric, the AI reviews and ranks, you interview the top 20. What took three weeks of recruiter time is now a Monday morning task.

The regret here is concrete: we lost candidates to competitors because our screening process was slow. Fast companies hired the best reps from the same pool while we were still in week two of manual review. We weren't out-competed on offer. We were out-competed on speed. That's the kind of thing that keeps a former VP awake at 2 AM fifteen years later (and the associated insomnia).

2. Territory Design: Hand-Drawing Market Maps vs. AI-Optimized Coverage

When we launched a new city, I had a whiteboard, a Salesforce export, and a lot of opinions about neighborhoods. Territory design was a combination of experience, instinct, and whatever data we had time to look at before the launch date.

We got it wrong a lot. Not catastrophically - experienced operators develop good instincts - but wrong in ways that compounded. A territory that's 15% too large creates a rep who's always behind. A territory that's split awkwardly generates internal conflict and double-counting. We'd often spend the first 30–60 days in a new market restructuring territories we'd just designed. Nothing builds team morale like telling someone on day 30 that their territory is changing.

AI-optimized territory design uses density data, historical win rates by geography, rep travel time, account concentration, and competitive coverage to generate and score territory options. You can run 40 scenarios in the time it used to take to draw one. The output isn't perfect - market knowledge still matters - but it's a dramatically better starting point than a whiteboard and a confident marker held by someone who got four hours of sleep.

3. Forecast: The Quarters Where the Board Meeting Was a Disaster

There were quarters at Groupon where we walked into a board meeting with a forecast that was wrong. Not slightly wrong - wrong in the way that requires an explanation, a revised slide, and the particular silence that happens when a room of senior people are all choosing to be polite at the same moment. Every CRO I know has sat in exactly that silence, nodding confidently while their internal monologue screamed.

I once gave a board presentation where I confidently committed to $5M in Q4 pipeline. We closed $1.8M. The board did not find my optimism charming.

The forecasts were wrong for the same reason forecasts are always wrong: asking a sales rep to objectively forecast their own deal is like asking a parent to judge their kid's talent show. They're going to say it went great. It did not go great. The optimism bias ran down every layer of the organization. By the time the district numbers aggregated to regional numbers, the inflation compounded into something that looked like a plan and wasn't.

Behavioral scoring - days since last activity, economic buyer engagement, stage velocity - produces a more accurate forecast than rep submissions because it doesn't care about feelings. If nobody from a buying organization has responded to anything in 12 days, the AI flags it. It doesn't believe the rep's story about why that's fine.

I can think of two quarters where early AI-assisted behavioral scoring would have changed the board conversation from an explanation to a plan. That's not a small thing.

4. Ramp Time: Six Months to Productivity, AI Could Cut It to Three

Our average new rep ramp at Groupon was 5–6 months before they were producing at quota levels consistently. This was not unusual for the type of complex, relationship-intensive selling we were doing in a new market.

The ramp drag was painful in multiple ways. New market launches had a longer bleed period before they turned profitable. High early turnover - reps who weren't going to make it often washed out during ramp - meant we were constantly restarting the clock in some territories. The financial model for each new market had to account for 5–6 months of below-quota productivity before the math started working.

AI-assisted onboarding addresses ramp through two mechanisms. First, reps have access to always-on coaching - they can get instant feedback on call scripts, deal qualification, objection handling, and account research without waiting for a manager to be available. Second, practice scenarios that used to require a manager to roleplay can be done with AI at any hour, compressing the repetition cycle that builds skill. It doesn't get tired. It doesn't have seven other reps waiting. It hasn't checked its phone once.

The benchmark I've seen from companies deploying AI-assisted onboarding deliberately: 30–40% reduction in ramp time. At Groupon scale, across 23 markets, that's not a small number. That's tens of millions of dollars in earlier productivity.

5. Deal Coaching: One Manager, Many Deals, Not Enough Hours

The fundamental constraint of a sales organization is manager bandwidth. One manager covering 8–10 reps cannot give every rep deep deal coaching on every deal every week. Mathematically impossible.

What actually happened: I coached the biggest deals and the reps who were most at risk. The middle of the team coached itself, which means the middle of the team reverted to whatever habits they'd developed before they joined. Average reps stayed average not because they weren't coachable but because coaching was rationed by necessity. Most companies use the available manager time the way my grandmother uses her iPad - they own a remarkably powerful resource and apply it to checking the weather on the most visible deals.

AI changes this. A Deal Coach tool that every rep can access before every pipeline review - where they paste their deal notes and get immediate MEDDPICC-style feedback - doesn't replace manager coaching. It gives the manager better-prepared reps to coach, and gives reps development resources that don't depend on manager availability.

At Groupon, across 400 sellers, I was getting coaching to maybe 40–60 reps per week at meaningful depth. With AI-assisted coaching available to every rep every day, that number is 400. That's the leverage shift.

The Honest Regret

Here's what I'll actually admit: some of the people who washed out during the Groupon ramp would have made it with AI-assisted coaching. Not all of them. Some people genuinely weren't right for the role. But some of them were right for the role and needed more development support during ramp than the management team had capacity to provide.

We lost good people to under-resourcing of their development. That's on me as a leader, and it's the kind of thing I'd do differently.

The playbook from 2010 to 2012 wasn't wrong. The core principles - clear territory, clear quota, clear expectations, rigorous pipeline review, aggressive coaching culture - are still right. The speed has changed by 10×. The leverage has changed by 10×. The same principles, applied with today's tools, produce faster results with smaller teams and better outcomes for the people on them.

If you're building a revenue org and want to talk strategy, I'm at joepeck.ai.