Hello and welcome to The Forecast Files. I’m Claire Whitmore, joined as always by the ever-insightful Henry Sinclair. Today, we’re diving into something that, well, might sound a bit technical at first—PyMC. But trust me, it’s far more approachable than it sounds.Absolutely, Claire. PyMC is, in essence, a Python library for probabilistic modeling. Now, before anyone tunes out, let’s just say it’s the tool that lets us make predictions that actually account for uncertainty, rather than pretending we know the future with absolute certainty. Which, let’s be honest, we never do.Exactly. I always like to use the weather app analogy. You know when you check your phone and it says there’s an 80% chance of rain? That’s so much more useful than just a yes or no. PyMC lets us do that for marketing forecasts—so instead of saying, “This campaign will get 1.2 million views,” we can say, “There’s an 80% chance it’ll get between 1 and 1.4 million.” It’s a subtle shift, but it changes everything.It really does. I mean, the world isn’t black and white, is it? There’s always a bit of grey, a bit of, well, unpredictability. And PyMC helps us embrace that, rather than ignore it. Claire, I think you’ve got a story about how this changed your approach back in your early days?Oh, I do. When I first started in marketing, I was all about the single number forecasts. I’d present these neat, tidy figures to clients—no room for doubt. But the reality was, things rarely landed exactly on that number. Once I discovered probabilistic modeling, it was like a lightbulb moment. Suddenly, I could show the range, the uncertainty, and clients actually trusted the forecasts more. It was liberating, honestly.And it’s so much more honest, isn’t it? We’re not pretending to be fortune tellers. We’re saying, “Here’s what’s likely, here’s what’s possible.” That’s a much stronger position to be in.Let’s bring this down to earth a bit. We recently spoke with Alex from the data team, who’s been using PyMC to forecast ad performance. And what stood out to me was how he talked about confidence intervals—not just giving a single number, but a range, with confidence. That’s a game-changer for managing expectations.Yes, Alex said it’s not about being perfect, it’s about being prepared. When you give teams a range, they’re not blindsided if things shift. They can adapt, pivot, and make smarter decisions as the market changes. I think that’s so important, especially in fast-moving industries like ours.I’ve got a campaign story that really drives this home. We were working on a big launch, and the initial forecast—if we’d just gone with the single-point estimate—looked fantastic. But when we ran it through PyMC, the confidence interval was much wider than we expected. That uncertainty made us pause. We scaled back our commitment, and sure enough, the results landed toward the lower end of the range. If we hadn’t accounted for that uncertainty, it could’ve been a very expensive mistake.That’s such a good example. It’s not about being pessimistic, it’s about being realistic. And I think teams appreciate that honesty. It builds trust, both internally and with clients.And it’s not just about the numbers, is it? It’s about how those numbers help us adapt, stay agile, and avoid overcommitting when things are uncertain. That’s the real value.And that brings us to something I’m really passionate about—PyMC isn’t just for the data team. Its approach empowers everyone, from strategy to creative to client services. When you understand uncertainty, you make better decisions, full stop.Absolutely. I’ve seen PyMC-driven insights lead to smarter, more flexible planning across the board. Media teams can adjust spend in real time, people teams can set more realistic targets, and client services can communicate results with a lot more clarity. It’s a win all around.I actually mentored a client services team recently, helping them use PyMC-driven forecasts. At first, they were a bit wary—numbers can be intimidating, right? But once they saw how to communicate uncertainty clearly to clients, it changed the whole dynamic. Clients felt more informed, less anxious, and the team felt more confident in their recommendations.It’s that transparency, isn’t it? When everyone’s on the same page about what’s likely and what’s possible, you get better buy-in and, frankly, better outcomes. And it’s not just a data thing—it’s a people thing, a media thing, a you thing. I think that’s the real magic of PyMC.Couldn’t agree more, Henry. Well, that’s all we’ve got time for today. Next time, we’ll dig even deeper into how PyMC helps us forecast with confidence—and maybe even spot the next big trend before it happens. Thanks for joining me, Henry.Always a pleasure, Claire. And thanks to everyone listening. Until next time, stay curious and keep embracing that uncertainty.Goodbye, everyone.Goodbye!

Episodes

Ep.1 - Embracing Uncertainty with PyMC