A Clean List Goes Bad on Its Own
Inbox placement for Republican committee mail swings by about 30 percentage points across mailbox providers in an ordinary week. Not during a crisis, and not in the frantic fortnight before an election. In a normal week, the same message to the same kind of supporter has meaningfully different odds of arriving depending on which provider the recipient happens to use.
That comes from a working paper described in the Washington Examiner this month, which puts the cost of two recent filtering episodes at roughly 142,000 first-time donations that never happened.
There is a real argument about what caused it. The paper's authors reportedly decline to separate a technical explanation from a partisan one, there is an FTC letter in circulation, and there is a proposal that providers be required to disclose when they change how they treat a class of senders. That argument is worth having. It will not be resolved by November 3, and even if it were, a notification would arrive far too late to matter. By the time a campaign learns its mail is being filtered, the reputation that produced the filtering was built months earlier. Sender reputation is not something you repair on deadline.
Which is why the more useful question is not who decided. It is how a campaign's mail earns that treatment in the first place — and the honest answer is more uncomfortable than the one most operatives reach for.
The trap that catches you is usually one you built yourself
When deliverability people talk about spam traps, most campaign staff picture the obvious version. An address that never belonged to a human, seeded quietly across the internet to catch senders who scrape or buy lists. Hit one of those and the diagnosis is simple: you mailed people who never asked to hear from you. In September, Brian Krebs reported trap data showing one donation platform's mail hitting spam traps at roughly four times the rate of the other's during the final week of July 2025. Raymond Dijkxhoorn of SURBL, one of the blocklists involved, put it plainly: it is the mechanics, and mail should not be ending up in spam traps.
But there is a second kind of trap, and it is the one that should worry a disciplined campaign, because it catches programs that did everything right.
Recycled traps were real addresses. A supporter opted in, used the address for years, and eventually abandoned it — changed jobs, changed providers, moved on. The provider let it bounce through a dormancy window and then quietly reactivated it as a detector. Industry practice generally puts that window somewhere around a year, though the range runs from roughly six months to two years and no provider publishes its policy.
Read that again with a campaign file in mind. You can hit spam traps mailing nothing but people who genuinely asked to hear from you. No purchased list. No scraped addresses. No vendor cutting corners. The address was clean when you collected it. It expired in your custody, while you weren't looking, and now it is reporting you.
The dormancy math almost nobody runs
Put that mechanism against a campaign calendar and the problem stops being abstract.
A committee that stops mailing after November 2024 and reactivates its file in September 2026 has left that list sitting for twenty-two months. That is not close to the recycling window. That is comfortably past it, on a file where some share of addresses stopped being used by a human somewhere in the interim and have since been repurposed into detectors.
The reactivation send is not a reactivation. It is a sweep across a minefield the campaign laid itself, and the damage registers in exactly the eleven weeks when the program can least afford it.
The volume problem compounds it. Campaigns go from near-zero sending to enormous daily volume in the final stretch, out of infrastructure that has been dormant for a year or more. Providers read a sudden spike from cold infrastructure as abuse, because from the outside it is indistinguishable from abuse. A program that has been sending at a steady, engaged cadence for two years can scale into October without tripping anything. A program that went dark cannot, and there is no version of this where you talk your way out of it in week nine.
Standard guidance for warming a new sending domain on a dedicated IP runs six to twelve weeks, and that range gets quoted as though it were the finish line. It isn't. Warming gets you a functioning sender at modest volume. It does not get you to the point where you can mail hundreds of thousands of people. We tell clients not to raise volume by more than about ten percent a week, and even that is aggressive enough that it needs watching. Run that math from a cold start and the climb to the volume a statewide program actually needs takes more than a year.
Election Day is eleven weeks out.
Maintenance is the whole strategy, and it is unglamorous
None of the fixes are clever. That is what makes them easy to skip.
It means suppressing addresses on engagement signals before they go dormant long enough to be recycled. It means validating the file on an actual schedule — twice a year is the right target, and even once a year would put most campaigns and offices well ahead of where they are today. It means running warming sequences that turn new contacts into engaged ones rather than parking them in a file to age. It means monitoring through seed networks so a placement shift is visible the week it happens instead of after the quarter closes. And it means sending at a cadence that keeps the domain's reputation alive between cycles, even when there is nothing urgent to say.
We took one program from roughly 2,000 engaged constituents to more than 50,000. It took twelve months, and the twelve months is the part worth noticing — not the growth rate. Our new-contact sequences typically convert somewhere in the twenties and have run past thirty percent, which is what earning an address actually looks like when you put a number on it. Deliverability holds above 99 percent, and it holds because the file is maintained continuously, not because of anything ingenious at send time.
The off-cycle year is the cheapest inventory you will ever buy
Here is the part that gets treated as optional and shouldn't.
Off-cycle, acquisition costs less. The inbox is quieter. Complaint tolerance is more forgiving. There is room to test something, get it wrong, and correct without a headline. On-cycle, every input costs more, the channel is saturated, and there is no room for any of it.
A candidate who goes dark between cycles is choosing to rebuild the most expensive asset in the program at the single worst moment to build it — and to rebuild it on a foundation that has been quietly turning into spam traps the entire time. That is not a pause. It is deferred spending with interest, and the interest is charged in inbox placement during the last six weeks.
The same logic runs past Election Day. The engaged audience a campaign spends two years assembling is the same audience the next campaign will need, and the recycling clock starts the day the sending stops. Winners often have an official communications track that legally cannot absorb the file. Losers usually just stop. Either way the asset degrades, and somebody repurchases it at a premium twenty-two months later.
The inbox is not something you win in October. It is something you are either maintaining or losing, every month, in a cycle year or not. The campaigns that will land in November started paying for it a long time ago — and the ones that didn't will spend the fall arguing about whose fault it is.