Welcome back to The Forecast Files. I’m Claire Whitmore, and as always, I’m joined by the ever-insightful Henry Sinclair. Henry, how are you today?Doing well, Claire, thank you. And hello to all our listeners. Today’s topic is one that, frankly, keeps a lot of us up at night—how unpredictable media planning has become. I mean, it’s a bit of a wild west out there, isn’t it?Absolutely. The media landscape just never sits still. Audiences are constantly shifting—one minute everyone’s on a new social platform, the next, they’re off chasing the latest trend somewhere else. And don’t get me started on how quickly those trends can turn everything upside down.Right, and that’s why the old way of forecasting—just giving a single number and hoping for the best—well, it just doesn’t cut it anymore. There’s too much noise, too many moving parts. You can’t just say, “We’ll get 1.5 million impressions,” and expect that to hold up when the ground is shifting under your feet.Exactly. I remember a campaign a few years back—oh, this still stings a bit. We’d done all the usual projections, had our numbers lined up, and then, almost overnight, a new social trend just exploded. Suddenly, our audience was somewhere else entirely, and our forecasts were, well, let’s just say, not worth the paper they were printed on. It was a real lesson in humility—and in the need for a better approach.That’s such a familiar story, Claire. And I think a lot of our listeners have been there too. Which is why, as we discussed in our last episode, embracing uncertainty isn’t just a nice-to-have—it’s essential if you want to stay ahead in this business.So, let’s talk about how PyMC changes the game. Instead of pretending we can predict the future with a single number, PyMC lets us model a whole spectrum of possible outcomes. It’s what we call probabilistic forecasting.Yes, and that’s such a powerful shift. Rather than saying, “This campaign will get exactly 1.5 million impressions,” we can say, “There’s a 90% chance it’ll land somewhere between 1.2 and 1.8 million.” That range is so much more useful, both for us and for our clients.It’s like giving them a weather forecast instead of a crystal ball, isn’t it? You wouldn’t plan your weekend on the assumption it’ll be exactly 22 degrees and sunny at 2pm sharp. You want to know the range, the likelihoods, so you can plan accordingly.That’s a great analogy, Henry. And it’s not just about numbers—it’s about setting expectations. I remember Dana, one of our colleagues, sharing how showing clients the full range of potential outcomes actually helped them feel more confident. They could see the uncertainty, yes, but also the logic behind it. And when things shifted mid-campaign, they were ready to adapt, rather than panicking because the “magic number” wasn’t hit.Exactly. It’s about building trust, isn’t it? When you’re upfront about the uncertainty, clients feel like they’re in the loop, not being sold a fantasy. And, as Dana said, it gives everyone room to adapt if things change—which, let’s be honest, they always do.Now, let’s dig a bit deeper into how PyMC actually does this. At the heart of it is Bayesian inference. I know that can sound a bit intimidating, but really, it’s just a way of updating our forecasts as new data comes in. We start with what we know, and as we learn more, the model adjusts—getting smarter, if you like, with every new bit of information.That’s right. It’s a bit like learning as you go. The more data you feed into the model, the better it gets at narrowing down the possibilities. And what I really appreciate about PyMC is how transparent it is. You can see the assumptions, the uncertainty, and the logic behind every forecast. There’s no black box—everything’s out in the open.And that transparency is so important for building trust. I had a client once—very skeptical, always wanted to know exactly how we’d come up with our numbers. When we started using PyMC, we could actually show them the model’s logic, the assumptions, the uncertainty. It turned the whole relationship around. Suddenly, they weren’t just a client—they were a partner in the process.I’ve had a similar experience, Claire. There was one client who, frankly, didn’t trust any forecast that wasn’t a single number. But when we walked them through the PyMC model—showed them how it updated with new data, how it handled uncertainty—they became one of our most loyal partners. It’s amazing what a bit of transparency can do.Absolutely. And I think that’s the real power of this approach. It’s not about pretending we know the future—it’s about being prepared for whatever comes our way, and bringing our clients along for the journey.Well said, Claire. And that’s all for today’s episode. Next time, we’ll be diving even deeper into how these models work behind the scenes—so if you’re curious, be sure to join us again.Thanks for listening, everyone. Henry, always a pleasure.Likewise, Claire. Take care, everyone, and see you next time on The Forecast Files.

Episodes

Ep.2 - Precision in the Grey