The Playbook: The Problem With the Old Model – Too Few Choices, Too Much Money.

We’re doing a series on how Tavern works. Today, we’re starting with why the process isn’t broken; it’s just too slow.

March 26, 2026 | 4 min read

The Process

Political campaigns run on two things: knowing what voters respond to, and turning that knowledge into ads that move people. The tools campaigns use to do both, message testing and ad creation, haven't fundamentally changed in decades.

Here's what the standard process looks like: You're a campaign manager in a competitive House race. You hire a pollster and spend $50,000 testing messages. You come in with a half dozen ideas you think will work, you put them in front of a sample of voters, and a week to ten days later, you get back results showing that two of them polled a few points better than the others. You pick those two. Then, you and your media firm spend another $50,000 shooting ads built around those messages. And then you put them on air.

You just spent $100,000 and several weeks to test six messages and produce two ads. If they work, great. If they don't, you're mostly out of runway to find out why.

That's the model. And everyone involved is usually doing their job well. The pollster is giving you rigorous data. The media firm is producing real creative. The problem is the process: you just didn't get enough shots on goal. 

You tested a small number of messages you already believed in. You found out which of your existing instincts was slightly more right than the others. Then you made two ads and hoped.

The Pace

Campaigns move by the hour. Research moves by the week. Ad production moves by the month. Those speeds don't fit together, and no amount of talent closes that gap.

Think about what actually happens during a competitive race. A national event changes what voters care about. A new attack ad by your opponent doesn't exist on Monday and is everywhere on Tuesday. An issue that was central in December is irrelevant by February. By the time a traditional poll comes back from the field, the race you were asking about may not be the race you're running anymore.

You're not getting a movie of how voters think. You're getting a photograph of a moment that has already passed.

The Tests

There's another cost buried in this model. A significant share of expensive research comes back telling campaigns things they already understood. Voters are worried about the economy. Healthcare is important. Life feels expensive. All true. All things you're likely going to incorporate into your ads somehow.

The process is limited by what your team can imagine, and great teams can do great work, but we’re all constrained by the number of hours in a day and the number of creative thoughts we can muster in a week. 

There's nowhere in the process to surface something no one has thought of yet, or to test a dozen unconventional ideas to see what breaks through. There's no room to test each piece of an idea and understand why something works. You end up with the most broadly acceptable version of your message, which is also often the least memorable one.

That's not the pollster's fault. It's not the campaign manager's fault. The instrument is designed to reduce risk and confirm things you think you already know. And most of the time it does that. 

The Fix

The goal isn't to spend less. It's not to fire anybody. It's to get more out of what you spend and test more than your team could ever write or even think of on their own. 

The same $50,000 that buys one large, slow poll could support multiple faster, targeted tests that build on each other. Each round teaches you something. Each round makes the next one sharper. Instead of picking the best of six messages you already believed in, you explore hundreds of variations, including ones you wouldn't have thought to write yourself, and let the data tell you what actually moves people.

That requires a different kind of pipeline. One where message generation, testing, and creative production aren't three separate vendors working in sequence, but one connected process that moves fast enough to keep up with the race you're actually running.

The tools to build that pipeline exist. That’s what we’re building here at Tavern.

More to come.