FRONTLINE HQ

CASE STUDY ·FIELD INTEGRITY

§FIELD INTEGRITY · FLORIDA · SUMMER 2026

Your field budget is only real if the field log is real.

A Florida legislative campaign paid dozens of canvassers to knock doors in summer 2026. More than 30,000 door records landed in FrontlineHQ.

Most of the log looked like field work. Three people did not. Their records moved faster than a person can reasonably move from door to door. Their support rates were too clean. Their notes were empty.

FrontlineHQ did not just count the doors. It checked whether the work behind them added up.

Suspect records

2,400

Paid hours

~150

Exposed

$3,000-$3,750

Seven minutes. Twenty-nine doors.

Honest field baseline · 15–25 doors/hr

Spaced arrivals · mixed results · notes present

Suspect session · 29 doors / 7 min

One mark ~every 14–15 seconds · all "supporter"

29 doors · 7 minutes · 29 supporters · 0 not-home · 0 notes

EXHIBIT 00 · SESSION RHYTHM

Exhibit rows are illustrative reconstructions of the flagged pattern. No voter data shown.

The dashboard said “cushion.” The audit said “check the log.”

The campaign had roughly 7,700 IDs on paper against a win number of about 6,500. That is the kind of number that can make a field program look safely ahead.

Roughly 1,600 of those apparent supporters came from suspect records. Verified support was closer to 6,100 — what looked like a 1,200-vote cushion was actually a 400-vote hole.

WIN NUMBER · ~6,500

IDS ON PAPER · ~7,700

~1,600 PHANTOM SUPPORTERS

Data-integrity issue, not a vote projection.

Three sessions that did not behave like door work.

One odd number is not proof. FrontlineHQ looked for a stack of signals: entries-per-hour physics, the gap between entries, GPS distance-to-door stamps, result patterns, notes, and whether a record carried a paper-import tag.

EXHIBIT 01 · about 30 doors in under 8 minutes

Twenty-nine records. All 29 marked “supporter.” Zero marked not-home. Zero notes.

“The 36-second door.”

At a hundred doors an hour, a door takes 36 seconds. The flagged sessions ran as fast as double that.

That is an average pace of about one entry every 14–15 seconds. The campaign's honest field baseline was 15–25 doors per hour.

EXHIBIT 01
EXHIBIT 01: illustrative reconstruction of a flagged canvassing session showing entry time, logged result, gap since the previous entry, and a censored note field. No real voter data.
TIMERESULTGAPNOTE
14:02:03SUPPORTER—
14:02:17SUPPORTER14s
14:02:32SUPPORTER15s
14:02:46SUPPORTER14s
14:03:01SUPPORTER15s
14:03:15SUPPORTER14s

Exhibit rows are illustrative reconstructions of the flagged pattern. No voter data shown.

EXHIBIT 02 · close to 50 entries in under half an hour

Close to 50 entries in under half an hour. Over 90% marked “supporter.” In the suspect sessions, about 83–89% of entries were logged less than 60 seconds apart. Real door work has a different rhythm.

EXHIBIT 02
EXHIBIT 02: illustrative reconstruction of a flagged canvassing session showing entry time, logged result, gap since the previous entry, and a censored note field. No real voter data.
TIMERESULTGAPNOTE
09:41:10SUPPORTER—
09:41:52SUPPORTER42s
09:42:19SUPPORTER27s
09:42:41NOT HOME22s
09:43:08SUPPORTER27s
09:43:36SUPPORTER28s

Exhibit rows are illustrative reconstructions of the flagged pattern. No voter data shown.

EXHIBIT 03 · roughly 50 doors in under 20 minutes

Roughly 50 doors in under 20 minutes, nearly all marked “supporter.” No notes. The individual sessions ranged from 100 to 250 entries per hour. The audit also checked GPS distance-to-door stamps, with 75 meters as a flag, and whether the result pattern matched a normal not-home rate.

EXHIBIT 03
EXHIBIT 03: illustrative reconstruction of a flagged canvassing session showing entry time, logged result, gap since the previous entry, and a censored note field. No real voter data.
TIMERESULTGAPNOTE
18:12:44SUPPORTER—
18:13:07SUPPORTER23s
18:13:29SUPPORTER22s
18:13:52SUPPORTER23s
18:14:14NOT HOME22s
18:14:37SUPPORTER23s

Exhibit rows are illustrative reconstructions of the flagged pattern. No voter data shown.

The manager could quarantine the suspect records with one command. The records were soft-deleted, not erased: the audit trail remained intact. The turfs reopened. FrontlineHQ also generated a blind 30-call verification list — 20 flagged records and 10 controls, unlabeled for the caller.

Get a free integrity audit →

CLEARED

The audit also cleared people.

This was not a hunt for the highest door count. The same checks cleared three canvassers whose numbers first looked unusual.

One had three-quarters not-home. That is not a manufactured success story. It is an honest grinder meeting a lot of closed doors.

A useful field audit protects the people doing real work as much as it catches records that do not hold up.

The paid-hours leak was real. The data leak was worse.

About 2,400 entries were suspect. Inside them: roughly 1,600 unique voters marked as supporters — entries and unique supporters are not the same count; suspect work also logged some not-homes. That represented roughly 150 paid hours, or about $3,000–$3,750 in wasted field spend.

But payroll was only part of the cost. Roughly 1,600 phantom supporters changed the campaign's own picture of where it stood. A false supporter record can send a campaign away from voters it still needs.

The audit that caught it costs less per month than one afternoon of fabricated canvassing.

What would a small leak cost your program?

Put in your paid field plan. Move the fabrication estimate. See how many paid hours and dollars would be exposed if part of the log did not hold up.

This is a planning estimate, not an accusation. The point is to measure the cost of not checking the work.

Planned paid hours

0

Planned field payroll

$0

Hours exposed

0

Budget exposed

$0

Enter your paid field plan above to see the hours and dollars exposed at the selected fabrication rate.

A rate setting is a scenario, not a finding. A finding requires the underlying telemetry and verification calls.

Send us your field export.

We will run a free integrity audit in 48 hours. We will show the signals, the exceptions, and the records that deserve a call-back — without naming a canvasser before the evidence is reviewed.

→ Send us your field export — free integrity audit in 48 hours

ANONYMIZED REVIEW · AUDIT TRAIL PRESERVED · NO CLIENT NAMES PUBLISHED

Or start with the signup form.

Questions first? team@frontlinehq.vote