Tech CRO Says AI Makes Forecasting Stronger — And CRO Accountability Harder To Avoid
George Storm explains how AI strengthens revenue forecasting by improving visibility and decision-making while making human accountability more important than ever.
George Storm
CRO at N.Rich
Forecasting: AI revenue forecasting replaces reassuring historical estimates with continuous, regime-aware probabilities that expose shortfalls before quarters end.
Accountability: AI identifies patterns across revenue signals, but CROs must make deal, pricing, staffing, and board communication decisions.
Calibration: Early models showed dangerous confidence, proving leaders must challenge assumptions and test outputs against messy real-world data.
Workflow: A weekly AI workflow unifies pipeline, usage, support, calls, and internal signals, then routes risks to owners.
Limits: AI accelerates existing demand and surfaces problems, but cannot repair poor data or replace enterprise judgment.
George Storm has been in sales for 23 years. He's currently CRO at a B2B ABM platform called N.Rich, where he leads sales, marketing, customer success, partners, and product support.
We sat down with George to learn how he rebuilt forecasting and reporting with AI. Here's what he told us.
Owning the entire funnel
I'm George Storm, CRO at N.Rich, a B2B ABM platform. We're transitioning from a single-product ABM company to a platform play, which we market as Composable GTM. Our current revenue model is an annual SaaS subscription with attached services. We are introducing a modular motion, per-module subscriptions, and outcome pricing as we launch GTM OS.
Create a free account to continue reading this article.
Want more from the CRO Club? Sign up to get free access and join modern revenue leaders using proven playbooks, peer insights, and AI-driven strategies to build predictable pipeline and scalable growth.
Have an account? Log In
I oversee the full revenue surface, end to end: sales, marketing, customer success, partners, and product support. I own demand creation through net revenue retention — the entire funnel, not just a slice. Our GTM structure includes a lean, EMEA-weighted sales team; a marketing organization that drives demand and ABM; a CS organization handling renewals and expansion; and a concurrent partner motion.
We primarily serve European B2B mid-market to enterprise companies: specifically, revenue and marketing leaders who run account-based go-to-market strategies.
How AI transformed revenue forecasting accuracy
I've spent 23 years in sales, and the last stretch has been harder than the first two decades combined.
I didn't arrive at AI through curiosity. I arrived through a miss. I was forecasting the way every CRO forecasts — off historical conversion — and the model kept telling me what I wanted to hear right up until the quarter it didn't. The mechanism of self-deception was the model itself. And I was the one who built it.
So, I rebuilt how we forecast, end to end. We shifted away from a point-estimate model based on historical conversion. It had a Monte Carlo wrapper that made it feel rigorous. It wasn't. A point estimate is a point estimate even when you average three of them.
We moved to a regime-conditional, four-input probabilistic model. Win rate, capture, NRR, and cycle length, each calibrated against external priors from WEF, IMF, and McKinsey. AI runs the distribution continuously instead of me rebuilding a spreadsheet once a quarter. To do this, we use Claude for macro signals, plus Fathom, Hubspot, and our own platform.
Regime-aware. Probabilistic. Continuous. That rebuild is what pulled me into AI properly, not as a productivity layer bolted onto the seat, but as the thing that lets one operator run the math, read the market, and act inside the same week. That's the work now.
Thanks to this change, the forecast became a read, not a report. It surfaced the gap below commit before the quarter turned, not after. The board stopped being surprised. And that's the whole point.
Why AI reads but the operator must act
As far as what AI should handle and what humans should handle, AI informs the read:
- Regime forecast distribution
- Pipeline prioritization
- Account health scoring
- Churn risk flags
- Weekly cross-functional read across Sales, Marketing, CS, and Partners
AI now powers all of that because it involves pattern recognition across more signals than a human can hold in their head.
Explicitly human activities involve the response:
- The call on a specific deal
- Pricing on a strategic account
- Who goes in which seat
- The narrative to the board.
The reason is simple. AI widens what I can see, but we cannot delegate accountability to a model. Influence as model input, not as decision-maker. The model reads. The operator acts.
How AI improves workflows but risks overconfidence
The first models were overconfident…Adversarial calibration turned out to be the job, not a step in it — red-teaming my own model with a senior-analyst prompt whose only job is to tear the optimism out. I could have saved a quarter if I had learned that sooner.
Here are the good results:
- The weekly CRO report that used to eat the better part of a day now assembles itself, and I spend the time deciding instead of collecting.
- Health scoring catches at-risk accounts one to two quarters before the lagging metric moves.
- The forecast holds up in the boardroom, which is the only accuracy test that matters.
But it's not all good. The first models were overconfident. One came back at 97% likelihood and was structurally too optimistic, because I'd fed clean assumptions into dirty inputs.
Adversarial calibration turned out to be the job, not a step in it — red-teaming my own model with a senior-analyst prompt whose only job is to tear the optimism out. I could have saved a quarter if I had learned that sooner.
AI didn't save me that work. It moved it.
A real-world reporting workflow
Let's run through the weekly read workflow.
On Monday, AI pulls from HubSpot for pipeline and renewals, Amplitude for product usage, Pylon for support, Fathom for call sentiment, and Slack for client and internal signals. It creates one governed data layer. It scores account health, flags the at-risk accounts, fires Slack alerts straight to the owners, and refreshes the regime-conditional forecast distribution.
Then, I read it, decide the response, and we close the loop by Friday. Track, read, act — on a one-week loop with one owner who is deliberately not the manager. The system does the collecting and the scoring. The seat does the deciding.
The forecast feeds the loop. The loop recalibrates where capital and effort go. And the whole org reads and responds on a one-week clock instead of a ninety-day one. The forecast is the read. The plan is the response. The loop is the operating system.
How AI makes revenue leadership harder
AI hasn't generated new demand. It compresses and accelerates your existing pipeline, but it doesn't manufacture intent in a buyer that's not in-market. Anyone selling AI as a top-of-funnel miracle is selling you the wrong thing.
It also hasn't replaced judgment in late-stage enterprise deals, and it hasn't fixed dirty data. Dirty data doesn't get better with AI. It gets louder.
AI has also made my life harder through role compression on the buyer side. As companies consolidate three jobs into one seat, the champion who bought us last year may have changed roles, lost ownership, or left the organization. AI created that dynamic. It doesn't solve it for me.
Where CROs should start — and what they should avoid
Here’s my advice: Don’t bolt AI onto the seat as a productivity tool. Redesign the seat around it…And be in the math yourself. The CROs who hand this to a tool and look away will get the answer the tool was built to give them. Make sure you set it up to tell you when you’re wrong.
Here's my advice: Don't bolt AI onto the seat as a productivity tool. Redesign the seat around it.
Start with your forecast, because that's where the board judges you, and the bar changed in 2026. You don't get fired for missing the number anymore. You get fired for the surprise. A model that can't see the miss coming is the liability, not the miss.
And be in the math yourself. The CROs who hand this to a tool and look away will get the answer the tool was built to give them. Make sure you set it up to tell you when you're wrong.
Follow along
You can follow George Storm's work on LinkedIn and georgestorm.io.
More expert interviews to come on The CRO Club!
