What Happens After You Record

You’ve made your plan, headed out on election night (or filed your public records request, or visited the elections office), and recorded poll tapes with the Actual Vote app. Your videos have been uploaded to America Counts.

Now what?

The journey from your recordings to published findings

Here’s what happens on our end after you submit your Actual Vote recordings:

1. Your videos arrive in our system. The Actual Vote app automatically uploads your recordings to America Counts. They become part of the public archive at av.democracycounts.org, where anyone can view them.

2. We transcribe the poll tape data. We read the vote totals off each video — every race, every candidate, every vote total visible on the tape — and record them in a structured form, identifying each poll tape by its precinct, voting location, and voting method (election day, early voting, absentee, etc.).

How we keep up: Actual Vote Assist. When an election brings in thousands of recordings, transcribing every tape by hand simply isn’t realistic — and hand transcription is exactly the kind of slow, repetitive work where tired people make mistakes. So most submissions now flow through Actual Vote Assist (AVA), a pipeline we built that reads the numbers off each tape automatically and checks them against the official results. The tapes that cleanly match — the large majority — are cleared without anyone having to retype a digit. A person still does the parts only a person can: setting aside recordings too blurry to trust, telling the system which contests each tape should contain, and reviewing anything the software flags as uncertain. That frees our finite expert attention for the cases that matter most — the close races and the precincts where a reporting error would have the biggest impact on the outcome. It isn’t a promise that every digit on every tape has been re-checked by hand; it’s a deliberate, and very favorable, trade between throughput and accuracy — one that lets us independently verify far more of an election than hand transcription ever could.

3. We gather the official results. We download the officially reported results for the same county and election from the county election authority’s website. These are the numbers that the county has certified as the official vote totals.

4. We do the comparison. This is the core of an Actual Vote analysis. For each precinct and race where we have poll tape data, we compare the vote totals from the poll tapes to the corresponding officially reported totals. If the numbers match, that portion of the vote reporting is confirmed. If they don’t, we have an initial non-match that needs investigation.

5. We investigate initial non-matches. Not every initial non-match is evidence of a problem. Common explanations include transcription errors (we misread a number from the video), missing poll tapes (we didn’t get recordings from every precinct, so the numbers don’t add up), and undocumented reporting logic (write-in votes or provisional ballots handled in ways that aren’t immediately obvious). We work through each initial non-match carefully to determine whether it has a satisfactory explanation — resolving it to a match — or whether it represents a discrepancy in vote reporting. A discrepancy we can’t explain after investigation is what we call an apparent discrepancy.

6. We publish our findings. America Counts releases a detailed public report documenting the Actual Vote analysis: how many precincts were covered, how many comparisons were made, how many votes were audited, and what we found. You can see a real example in our 2024 North Carolina Primary report.

This entire process typically takes place in the weeks following an election, with the goal of completing the Actual Vote analysis before the election certification deadline when possible — since that’s when findings can have the most direct impact. We’re a small organization, and we may not be able to analyze every submission before certification. We prioritize close races and precincts where vote reporting discrepancies are most likely to matter. That said, an Analysis completed after the certification date still has real value — it contributes to election transparency and helps build a broader picture of how vote reporting is working across the country.

And because everything in the Actual Vote system is public — the recordings, the official results, the methodology — you don’t have to wait for America Counts to do the analytical work. Anyone is free to analyze Actual Vote videos on their own, and some users have done exactly that, comparing their own recordings against the official results independently. America Counts’s analysis is a service we provide, but the data belongs to everyone.

A concrete example: what the analysis actually looks like

Let’s return to our fictional Crow County to see what an Actual Vote comparison looks like in practice.

Recall that Crow County has four precincts, and in this election there was one race: US Senate, between Gina Gold and Sam Silver. Actual Vote users recorded poll tapes at all four precincts on election night.

After transcribing the poll tapes, here’s what America Counts has:

And here are the officially reported results from the Crow County election authority’s website:

Now we compare them side-by-side:

Every number matches. The vote reporting in Crow County was done correctly for the US Senate race, and we now have independent evidence confirming it.

This is the outcome we see most often — and it’s a good outcome. It means the election officials did their jobs, and now there’s a public, independently verified record proving it. That’s a meaningful contribution to election transparency.

But what if something didn’t match?

Suppose that instead of the results above, the officially reported totals for Corvus 03 showed Gold with 38 votes and Silver with 62 — a 10-vote shift in Silver’s favor:

Now we have an initial non-match. America Counts would investigate: Is this a transcription error on our part? Did we misread the poll tape video? If not, we’d contact the Crow County elections office and ask them to explain why the official numbers for Corvus 03 don’t match the poll tape.

If they can explain it (for example, if there was a known issue with provisional ballots at that precinct that required a manual adjustment), the comparison resolves to a match. If they can’t explain it, we have an apparent discrepancy — evidence of a genuine problem in vote reporting — one that, in a race this close (the county total would shift from a 200–200 tie to a 190–210 Silver lead), could change the outcome.

Either way, the public deserves to know. And that’s exactly what Actual Vote makes possible.

What your contribution means

Every poll tape video you record adds to the body of independent evidence about vote reporting in your area. Even when everything matches — especially when everything matches — your contribution matters. It means that the next time someone questions whether the vote reporting was accurate, there’s a public, verifiable record that says: we checked, and it was.

That’s the power of independent election auditing. And it’s made possible by people like you.