Methodology

A careful phone deal comparison, explained.

Direct answer: Phone Deal Audit compares the promotional device or trade-in credit described in a carrier offer with the credits shown on subsequent AT&T, Verizon, or T-Mobile bills.

1. You choose the records

Add the offer, order confirmation, trade-in record, and up to six bills. PDFs and images are processed inside your browser. The files, filenames, and extracted document text are not uploaded to Phone Deal Audit.

2. The browser analyzes and sorts the documents

PDF text is read first. Scanned PDFs and images use local optical character recognition only when necessary. Each document is then classified on your device as a bill, promotional offer, order confirmation, trade-in record, carrier email, or screenshot using deterministic wording cues, and you can change any type before the details are reviewed. Carrier-specific parsers look for candidate promotion totals, terms, monthly credits, device payments, dates, and line references.

  • Each candidate has a confidence level and local evidence excerpt.
  • Low-confidence values and uncertain document types are shown as uncertain rather than assumed.
  • If no phone-deal information can be found at all, the audit stops and asks for better documents instead of producing a hollow result.
  • You can continue with manual entry when a document cannot be read.

3. You confirm the important details

You can confirm, edit, mark unknown, remove, or add facts. The comparison runs only after you approve the normalized deal and bill entries.

4. Deterministic rules compare the timeline

The audit uses precise decimal money calculations. It distinguishes a normal activation delay, a catch-up credit, an ongoing match, a missing amount, an amount or line mismatch, a stopped credit, and a term mismatch. It does not use a runtime AI model and does not assign blame.

5. The result stays proportionate

When the deal appears on track or is still within a carrier's published start window, the result is free and no report is offered. When evidence supports a plausible discrepancy, a one-time $39 purchase unlocks a PDF action packet and CSV ledger generated locally on the device. A sample packet built from a fictional case shows what it contains.

How audit confidence is scored

Evidence completeness is a transparent weighted measure, not a recovery probability or legal-liability score. The result screen explains every factor behind the percentage.

  • Promotion value: 16 points
  • Promotion term: 14 points
  • Expected monthly amount: 14 points
  • Line match: 12 points
  • Bill coverage: 18 points
  • Credit-row quality: 14 points
  • Conflict-free evidence: 12 points

Remaining promotional value at risk

When a discrepancy is found and the promotion term is confirmed, the result also shows the remaining promotional value that may be at risk: the most recent bill's shortfall carried across the remaining scheduled credits in the term. Expected credits follow an exact schedule in which the recurring promotional value, after any immediate trade-in value, is divided across the term with the final month absorbing cent rounding, so the credits possibly missing and the remaining value at risk together can never exceed the confirmed promotion. It is shown only when that arithmetic is defensible from the confirmed terms, it assumes the supplied bills begin at the start of the term, and it is never presented as money owed or as a guaranteed loss.

Where the carrier rules come from

The start windows the audit applies, and every carrier rule quoted in the guides, come from the carriers' own published support pages and customer agreements, each linked from the guide that uses it with the date it was last checked. The About page explains how that content is maintained, and the dispute deadlines page quotes each carrier's current dispute clause.

Check your own deal

Use the free scan to compare the promotion and bills you confirm. Files stay in your browser.

Run the free private scan

Free scan. A one-time $39 packet is offered only when a plausible discrepancy is found.