When Paula Skaper was writing her book, Rethink, Realign, Reinvent, she used AI as a research assistant. She even set up a separate project just for the citations. But she still checked every citation herself, over several rounds, because every time there was an edit, something could get knocked loose.
Her biggest fear was missing a credit she owed someone, a date typed wrong, the right citation sitting in the wrong place. What she wanted was something she could hand a citation to and ask, “Does this actually say what I think it says?”
That was the topic Paula, Mark Stouse, and I kicked around on our last Causal GTM Leader chat. Last time we discussed ownership: the AI did the work, and your name is still on it. This time we got into “What can you actually check?”
Right now the argument about AI content is stuck on authorship. Did someone or something write it? Is it authentic? That misses the point. Whoever (or whatever) wrote it, the questions are whether it’s real, whether it’s verifiable, whether it’s true, and whether it’s load-bearing.
How many times has an email been sent with someone’s name spelled wrong, or the wrong date?
“Human does not mean absolutely true, right? Human means the truth I think that will get me my answer that I want from you.”
Paula Skaper
There once was a time when we had people for this. Mark used to be a reporter for Newsweek, where having a dedicated researcher was kind of a status symbol. Reporters put “CQ” at the bottom of a story after every fact had been checked, and they were more or less pledging their job on it. My first job out of college was at BC Business Magazine, and our editor could cut a 1,500-word piece to 1,000 without losing the thread.
Most GTM teams have neither. We took away the resources, increased our expectations, and reached for an easy button, first in Sales with CRMs and now in Marketing with AI.
Mark is in the throes of renovating his house, so load-bearing walls are on his mind these days. Knock out a wall that isn’t holding anything up and you get a bigger kitchen. Yay! Knock out the wrong one and the second floor ends up in the kitchen. Yikes! Knowing which is which before you start sledgehammering will save you a heep of trouble.
Go-to-market works the same way. “This campaign will drive pipeline next quarter” sits on a few walls: buyers are in market, the budget holds, the time lag is shorter than a quarter. Pull one out and see what happens to your kitchen.
“For this to be true, for this assertion to be true, what else has to be true and remain true?”
Mark Stouse
You don’t need a tool to run it. Take one number from your next forecast or slide deck and write it at the top of a page. Underneath, list what has to be true for it to hold. Then circle the ones you’ve actually checked. If there aren’t many circles, you know which part of the roof comes down first.
Mark is also testing a new AI tool called Proving Ground. Built with Claude, it can fact-check any document and hands back a list of questions that you answer. It’s a bullshit detector. I tried it on my website and it didn’t pass. What was missing, I should have seen myself. I felt like a dumbass (which is not the intent; the intent is to be less wrong).
I fixed what I could and ran another stress test. Better, but still with gaps tied to recent work I could point to but can’t publish until later this fall. I’ll run it again in a couple of months.
Do you have a red team on your side? If you don’t, are you asking AI the right questions, or are you just believing whatever it gives you?
I see the second one a lot. People blindly accept whatever gets spit out in the first round, never stress testing three or four times, and that answer becomes their narrative. Most of what we know reaches us the same way, secondhand and passed along because it cleared a smell test. By the time it lands on your desk the provenance is gone, nobody can be held accountable for it, and the claim itself is the only thing left to test.
You can also be the red team for your customers. Paula has been that for her clients for years. One client would go quiet for a week every time she sent back 20 or 30 questions on a new idea, then admit the math didn't work. (His last pitch before their contract ended came with a paragraph begging her not to do “that thing” she does.) Now she runs vendor reports through Proving Ground too, asking whether they hold water.
“The one who says, ‘Oh, that’s a great idea!’ is going to get you in a lot of trouble.”
Paula Skaper
Mark goes to a lot of GTM events. The keynotes get half an ear, and the tables after hours get full attendance, with marketers swapping stories that sound strikingly alike.
CFOs have their own version of that table, where they compare notes on what their GTM teams told them. They feel they’ve been sold a pig in a poke, after putting their own reputations on the line with investors to back those claims. Some of that distrust is self-inflicted. To a lot of finance people, sloppy budget management reads almost as a moral failing, and most GTM teams walk into value conversations oblivious to it.
That doesn’t make CFOs 100% right and marketing 100% wrong. But it’s why the trust gap exists, and a CFO who understands marketing but can’t check the numbers won’t take them on faith. GTM has to stop arguing from what was true five or ten years ago and answer two questions instead.
“What is working right now? What is a bet that anybody will fund?”
Mark Stouse
Step one of the Causal 12-Step Program is admitting you were wrong. Two years ago Mark told me to stop correlating. I’d spent most of my career doing exactly that, like almost everyone in marketing. (I’ve written about what else I got wrong.) There’s a sign in my house that asks, “Do you want to be right or do you want to be happy?” At work, swap happy for effective.
Step 1B (Mark’s addition) is making a list with everything that isn’t working, for you or for anyone who depends on you. It won’t be a happy list. But don’t stop there, because stopping at the list leaves you worse off than never making it. You go through it and come out improved, not perfect.
Not only that, but “Wrong” is so loaded that people can treat it as fatal.
“Given what I knew in the moment, this is what I believed. But what new information do I have now?”
Paula Skaper
You can ask that about a forecast or a vendor report without anyone losing face. Whether most GTM teams will ask it before their CFO does is another matter.
Missed the session? Watch it here.
By the way, Paula is now a regular, so we decided to change the name of our event to The Causal Revenue Show because GTM is all about revenue.
If you like this content, here are some more ways I can help:
Cheers!
This article is AC-A (“Authenticity Commons, Assisted”): AI helped produce it, with me in the lead. Here’s what that means.
Achim’s Razor is also published as a LinkedIn newsletter.
