Bad Digital and Other Ways to Lose

The political and public affairs market is opaque. The behavior isn’t. Twenty-five years of disbursement data shows you can spot a losing campaign before its own consultants do.

The single most predictive variable in a digital program isn’t budget.

If you can’t see where the money goes, the next best thing is to watch when it starts moving. The single most predictive variable in a digital program isn’t the budget. It isn’t creative or platform mix. It’s the start date.

Winning digital programs start meaningfully earlier. That head start is the difference between a program with time to build audiences, test, and establish frequency, and a program that shows up late to hot inventory making first contact with persuadables six weeks from Election Day. The opponent who started in June is already on screens before your program exists. Late entrants are buying remnant attention at premium prices and calling it a digital program.

Every media channel in the losing-campaign dataset starts late. Not just digital. Losing broadcast programs start months after the winning campaign.

The candidate who is logging travel in January is not the candidate who loses in November. Campaigns with any travel expense in their first 60 days win at 78%. Inside travel expenses, no surprises: private jet beats air travel beats car rental beats mileage reimbursement. Incumbents expensing high-end travel before their challengers have a nomination usually win.

Campaigns without travel in those first 60 days win at 44%. The pattern holds for challengers and open-seat candidates, not just incumbents, though the gap narrows. The candidate sitting in their living room in January waiting for something to happen is the candidate who loses.

Late entry across every channel is the operational signature of a campaign that made its decisions too late to change any of them. By the time they know they are in trouble, they have been in trouble for months.

When a campaign tells you it’s running digital, it is telling you nothing.

In 2016, 75% of political digital spending in California was untraceable to a final media destination. Eight years later, after platform transparency initiatives… it was 78%.

The money found new pipes. It always does. That number comes from tracing every digital disbursement chain as far as disclosure allows and measuring whether it terminated at a known media platform or stopped at an intermediary. No reasonable platform definition I tested brought it below 75%.

Disbursements filed directly to Meta-affiliated payees in California dropped nearly three quarters between 2020 and 2024. The decline is not in total Meta buying. California captures the first stop, but nobody captures the full route of how an electron becomes an ad.

So when a campaign tells you it’s running digital, it is telling you nothing. Unless you can trace where inventory is actually being bought, you ought to be concerned your dollars are making detours. This was as true in 2017 as it is today.

The Tells

These are the signals that show up in the filings before the campaign knows it’s in trouble. All win-rate stats below are descriptive patterns, not causes.

Heavy spend concentration in the final 30 days. Campaigns spending more than 15% of their total budget in the last 30 days win at 28%. This is the most consistent loss signature we found across more than 7,500 committee-cycles. It’s not that back-loading causes the loss. It’s that campaigns which are already losing compress everything into the end because they have no other option. They have run out of time and money, spending what’s left toward an outcome they can no longer change. If you are reading this in October and your October budget is over 15% of your total spend, you are already in the pattern.

How fast you built the team. The win rate climbs with the number of distinct payees engaged in the first 30 days. Campaigns at the top of that curve, 16 or more vendors, won 86% of the time. Campaigns that started with a single vendor win at 49%. Campaigns with pre-existing consultant relationships don’t need to shop around. They make calls and things happen. Campaigns without them spend the first month figuring out who to call while the opponent’s program is already running.

Vendor density. The optimal zone is narrow: less than 2 vendors per $10k spent wins at 75%. Above 5 vendors per $10k, your win rate is ugly. The campaign with 23 vendors on a $100k budget is not running a diversified program. Nobody has enough money to do their job. The money evaporates and the vendors won’t play well. The same can be true when we look at over-engineered digital media plans.

Staffing. Experience still shows up in the data. First-time staff win at 34%. Seven-plus year managers win at 58%. Post-Covid, more campaigns are being run by people with no field experience. The person running your program in 2026 likely came up through a vendor. They learned the vendor’s playbook. They are running it in your race. That is very different than at any point in American politics.

Self-service is for losers. The biggest tell is whether campaigns buy their own Meta and Google or hand that to experienced buyers. Self-service Meta buyers, the ones going direct to Ads Manager without a managed service layer, win at 46%. That finding is not visible from public FEC data. It requires subvendor-level attribution. We include it because the gap is too large to omit, and because the inability to see it from public data is itself the point: the disclosure system wasn’t built to capture how campaigns buy, only what they report buying. The platform isn’t the tell. The sophistication of the purchase is.

Fine, Here Are Seven More

These matter less than the ones above. They are still real.

What the first check bought. Campaigns whose first check was payroll win at 79%. That number is partly conflated with incumbency. Even in open-seat races, payroll-first campaigns outperform. A campaign that starts by ordering supplies is organizing a garage sale. A campaign that starts by hiring staff is running for office.

The Bob premium. Across 13,500 House candidacies since 2004, first-name length predicts win rates. Three-letter ballot nicknames, Bob, Joe, Tim, Dan, win more than longer formal names like Christopher, Elizabeth, Jonathan, and Nicholas. The gap holds steady when you split by incumbency status: short-named incumbents still beat long-named incumbents (90% vs 85%), short-named challengers still beat long-named challengers (5% vs 3%), same story in open seats.

What the last check bought. Final payment to food and catering: 74% win rate. Final payment to office supplies: 56%. Winners end by feeding their volunteers and supporters. Losers end on paperwork.

Whether anyone got paid more than three times. Campaigns with at least one vendor who received more than 3 payments win at 65%. The eight campaigns in our dataset with no repeat vendor at all went 1 for 8. Small sample, but real. It sits at the end of a smooth curve: more repeat-vendor relationships foretell less friction, higher win rate, at every level. A campaign that never paid anyone more than twice never built anything.

How they describe their spending. Campaigns filing 100 or more distinct expense purpose descriptions won at 90%. That’s professionalism. Campaigns filing 1 to 3 distinct purposes won at 48%. The signal is that campaigns disciplined enough to categorize their spending with precision tend to be disciplined about other things too. Sloppy filings are a symptom.

Sunday checks are kitchen-table checks. Weekday disbursers win at 62%. Sunday writers win twenty points less. No accounting professional is cutting the largest check of the week on a Sunday. Sunday checks are written by candidates and their spouses at kitchen tables, writing checks to people who called them that morning.

Round numbers. Campaigns where more than half of disbursements are round dollar amounts win at 20%. Real campaigns have real invoices. Round numbers mean somebody is guessing.

We Checked Everything

While we were in the data, we checked some other things, and I don’t mean the price of gasoline or GDP.

Mercury retrograde. Full moons. Hurricanes. Whether a blockbuster movie opened the week before Election Day. Whether the S&P was up. Whether your state had a drought. NFL results the Sunday before. College football the Saturday before. Bitcoin price on filing deadline. Nada. Flat lines. The Redskins Rule is dead. Allan Lichtman called 2024 for Harris.

After all that, rule #1 is funded incumbents win. The rest, you need to look harder. The data was always there. The question is whether anyone on your team knew how to look.