Who
SNAP directors, QC leads,
procurement, field staff
procurement, field staff
Saldor
Reducing SNAP error rates
Saldor
HR 1 cuts the federal share of SNAP administration and imposes error‑rate penalties on states whose payment error rate exceed 6%.[1] 27 states face $100M+ on the benefit side.[2] Saldor is researching the workflow and building software to reduce procedural error at application, recertification, and change reporting — designed alongside the directors, QC staff, and caseworkers who run the program. If SNAP runs through your desk, we’d like to talk.
A SNAP eligibility determination can involve dozens of data points pulled from pay stubs, bank statements, utility bills, verifications of residence, household composition, and immigration status — each governed by its own rule, each subject to change mid‑case. State SNAP policy manuals routinely run 300 to 500 pages, and a worker is expected to apply the right rule in the right order, every time.[5]
The result, reported year over year by USDA FNS, is a persistent rate of payment error that is neither fraud nor negligence. It is complexity meeting a human operator without the right instrumentation. States then carry the cost: corrective action plans, sanctions, reprocessing, and appeals.
We think this is a tooling problem — and a solvable one. But we want to hear from the people closest to it before we decide what to build.
Policy memoranda, waivers, and state options update faster than training materials and desk aids. Workers apply yesterday's rule to today's case.
Income, assets, and household composition arrive from different sources in different shapes. Transcription and arithmetic errors compound quietly through the case file.
Structured interviews become unstructured under caseload pressure. Questions get skipped; clarifications are recorded as conclusions rather than quotes.
Cases move between intake, eligibility, and QC without a shared record of what was checked and why. Context is re‑derived instead of carried.
Two costs now move together. States owe a larger portion of operating spend, and states above 6%, 8%, or 10% error rates owe a federal penalty on top. Every point of error rate has a dollar figure attached that did not exist before.[1]
A one‑point reduction in error used to be an operational nicety. Under HR 1, it’s a line item in the state’s general fund.Arizona’s SNAP eligibility staff fell from 1,370 in July 2024 to 880 in July 2025. Virginia is running 19% turnover statewide, with Richmond vacancy near a third. Massachusetts answers fewer than half of incoming calls. The question is not whether to train harder — it is whether the next generation of caseworkers should be working with the same desk aids as the last.[6]
We are interested in what a modern “desk aid” looks like when it can read the policy manual with you.Saldor is early and designing with agencies, not for them. We’re building a picture of where error actually enters a SNAP case, what tools workers already rely on, and what software has to look like to be adopted inside a real agency. If we end up a fit, we’ll keep talking. If not, you get a shared memo of what we learned.
hello@saldor.com →