They can
Help you spot simple formatting problems, calculate a stated comparison, and create a consistent starting point for investigation.
Inspect / Calculate
A small toolkit for better checks.
No API keys. Inputs stay in this page. These checks support investigation; they do not verify live integrations or guarantee platform acceptance.
ads.txt syntax checker
Paste seller records to inspect common local syntax issues, field counts, relationships, and duplicate records. This is a text-level check.
Impression discrepancy
Difference relative to system A: (A − B) ÷ A × 100. Positive means B is lower than A.
What these tools can and cannot prove
They cannot
Verify account ownership, prove seller authorization, inspect private exchange configuration, guarantee AdSense approval, or establish why a production discrepancy occurred.
Use the guides with the result
Technical checks are most useful when combined with logs, network traces, platform documentation, and controlled test cases.
Privacy by design
The tools are intended to process entered values in the browser. Do not paste passwords, private audience data, access tokens, or confidential production information into public web forms.
How to interpret tool results
These browser-side utilities are learning aids. They do not access a live DSP, exchange, Google Ad Manager, or AdSense account and cannot establish whether a publisher will be approved.
For JSON validation, first paste a sanitized synthetic payload. Verify the input is parseable, then inspect field meaning against the relevant protocol specification. A syntactically valid object may still violate a partner contract.
For ads.txt checks, verify the publicly served root-domain file, not merely a copied local text snippet. A seller record describes authorization; it does not prove that an ad was served or that the website passed a monetization review.
Never paste production secrets, authentication tokens, personal identifiers, or unredacted bidstream logs into a demonstration tool. Use invented IDs and values in educational examples.
Practical investigation notes and worked methods
How to use the ads.txt syntax checker
Copy the publicly intended ads.txt records into the input field, excluding credentials or private operational notes. Run the check and review each reported formatting issue in context. An ads.txt record typically identifies an advertising system domain, publisher account identifier and relationship type, with an optional certification authority identifier. A syntax checker can help identify malformed separators, missing fields or duplicate-looking records, but it cannot prove that an account belongs to the publisher or that an advertising system recognizes it. Confirm domain ownership and seller identity using the actual platform records.
Example ads.txt investigation
Suppose a publisher expects an exchange to sell inventory using a DIRECT relationship. The file contains a syntactically valid line, but the exchange account actually belongs to a reseller. A local parser may accept the line because its field structure is correct. The investigation must still compare the declared relationship with the contractual and account setup. In another case, two records may look duplicated but refer to different seller accounts; do not delete records based solely on similar names. Always preserve the published file before making edits and retest its public URL after deployment.
How to use the discrepancy calculator
Enter System A as the reference count and System B as the comparison count. The tool calculates the absolute difference and expresses it as a percentage of A. For example, if A reports 100,000 impressions and B reports 94,000, the difference is 6,000, or 6 percent of A. This calculation is descriptive, not diagnostic. Before treating the result as a discrepancy, confirm that both systems count the same event type, cover the same time interval, use compatible timezones and apply comparable invalid-traffic exclusions. If A is zero, a percentage relative to A is undefined and should not be interpreted as a normal percentage.
Build a useful comparison worksheet
Record the two source systems, metric definitions, timezone, start and end timestamps, inventory filters, demand filters, data freshness and any known ingestion delay. Compare a narrow cohort before broadening the analysis. If the discrepancy appears only on mobile web, investigate device-specific rendering and tracking. If it appears at midnight, inspect timezone boundaries. If it disappears after several hours, check late-arriving events and reporting backfills. Keep raw totals as well as corrected totals so that the investigation remains auditable.
Avoid common interpretation errors
A parser warning is not proof that Google or an exchange has rejected an ads.txt file. A successful syntax result is not proof of live authorization. Similarly, a 10 percent difference between auction wins and impressions is not necessarily a tracking defect because those metrics occur at different stages. Do not compare gross and net revenue, requests and impressions, or event time and processing time without explicitly labeling the difference. Tools should reduce arithmetic and formatting mistakes, not replace the evidence required to determine a production root cause.
Privacy and operational safety
The tools are designed for local, browser-side checks, but users should still avoid entering personal data, account secrets, raw consent strings or confidential bidstream payloads. When sharing results with colleagues, use sanitized counts and placeholder seller identifiers. Verify the current deployed JavaScript before relying on privacy assumptions; browser extensions, analytics and third-party scripts can affect what a page loads. Never treat a public demo tool as a secure repository for customer information. If a production investigation requires sensitive logs, use your organization’s approved observability and access-control systems.
A complete worked example
A publisher sees 120,000 impressions in its ad server and 112,800 in an exchange report. The calculator reports a difference of 7,200 impressions, equal to 6 percent of the ad-server reference. The analyst then learns that the ad server counts rendered impressions, while the exchange report includes only events accepted before a reporting cutoff. The next step is to compare the same event definition after the exchange data has settled, not to alter ad tags immediately. If the gap persists, segment by placement and hour, capture sanitized tracking requests and inspect the first divergent stage. The calculation helped quantify the question; the event definitions explain what to investigate.