Where should a PAC spend in 2026, and how does it know before it is too late to move money?

Why committees allocate blind

The FEC data behind this site's spend tracker shows the pattern every cycle: independent expenditures concentrate in a handful of races and arrive late, because that is when the committee finally knows which races are close. The information that would have let it spend earlier and more evenly did not exist at an affordable price. A real poll in one House district is a five-figure purchase, and a committee cannot poll twenty of them, so it waits for someone else's poll. By the time that poll is public, the campaigns and the other side have read it too.

What a slate measurement looks like

Every candidate in the bundle gets a live program: callers reaching the persuasion universe in that district, identifying each voter as supporting, considering, undecided or opposed, capturing the issue that matters to them, and turning the undecided. The campaign gets a dashboard for its race. The committee gets the roll-up: every race on one screen with the same measures, updated as calls happen.

Across six 2026 primary programs, The Political Group's callers held 40,916 live voter conversations in 36 field days and identified 9,976 voters by position, with 63 percent of decided voters saying support per program and a range across programs from 42 to 82 percent. That range is the point. A committee that sees an 82 and a 42 side by side in the second week knows where the next dollar goes; a committee waiting for a public poll does not.

What it costs to measure

The measurement is not a separate purchase; it is a by-product of the voter contact the committee would buy anyway. A 10,000-household program in each of ten races is a 100,000-household bundle, priced at the deepest tier on the published table, and every one of those hundred thousand households is both a persuasion contact and a data point. Run the slate through the calculator for the exact figure. Where a committee wants a read on a district before committing a program to it at all, HyperPulse gives a synthetic read in hours for a fraction of a traditional poll.

Moving money mid-cycle

The mechanics of reallocating are simple once the measurement exists.

• Set a review cadence, weekly, with the roll-up open. The support share among decided voters and the undecided share are the two numbers to watch by race.
• Where support is high and the undecided share is small, the race is won or lost already; stop persuasion spending and hold for turnout.
• Where support is competitive and the undecided share is large, that is where persuasion money moves votes. Deepen the universe.
• Where support is collapsing, decide early whether to defend or to release the money to a race that can still be won. That decision is the committee's product, and it can now be made in September rather than read about in November.
• Add a race by extending the bundle; the new program is priced at the enlarged tier.

An anonymized example

A committee funding several state legislative races in one 2026 primary cycle saw, in the first week of calling, that one of its candidates was drawing support from well over three quarters of decided voters while another was below half. It deepened the universe in the second race and stopped persuasion calling in the first, shifting to turnout. The names are the client's; the pattern is what the roll-up is for.

The two numbers to watch, and what they mean

The roll-up shows many measures, but allocation turns on two. The first is support share among decided voters: of the voters the callers identified as supporting or opposing, what share are with the candidate. Above 60 percent with a small undecided pool, the race is being won; below 45 percent with a small undecided pool, it is being lost; anywhere in between is a contest. The second is the undecided and leaning share of all identified voters: the size of the pool persuasion can still work on. Across TPG's 2026 primary programs that share averaged 26 percent, but it ranges widely by district and shrinks as election day approaches, which is why measuring in September is worth more than measuring in October.

Read together, the two numbers sort a slate into four boxes: won (hold for turnout), lost (release the money), contested with a large pool (spend), and contested with a small pool (spend on turnout, not persuasion). The committee's weekly decision is which races moved boxes.

What the FEC data adds

The spend tracker on this site shows, for every race, what other committees have already put in for and against each candidate. A race where the other side has spent heavily and the candidate's support is still rising is a race worth defending; a race where the committee would be the fifth outside spender on the same side is a race where its money is marginal. The two datasets together, the committee's own voter contact and the public spending record, are the whole picture, and both are on one site.