Skip to content
DocsSearch the docsEnglishBack to siteOpen console

Evidence, exclusion lists and refund claims

The Evidence tab holds the files you need once fraud is found: the click quality report for a refund request to Google, sealed network-level evidence for every other ad network, and the exclusion lists that stop the next spend.

The Evidence tab and the dispute evidence export in each source's drawer are on the Pro and AdPix 360 plans. On Starter the tab shows the upgrade screen.

The Evidence tab at a glance#

In Fraud Protection, the Evidence tab has three cards: What Google accepts, Evidence for the selected range, and Exclusion lists. Its files do two different jobs and are built from two different windows:

File What it is for Built from
Click quality report a refund request to Google the invalid clicks in the selected range
Refund evidence (networks) a network-level refund claim to any ad network the sealed rows in the selected range
Google Ads IP exclusions stopping the next spend on Google the current list, not a range
Generic blocklist a firewall, n8n or another ad network the current list, not a range
Meta network advisory reviewing placements on Meta the current list, not a range

The first two follow the page's date picker; the three exclusion lists always give the state right now. Anything still sitting at Review — meaning AdPix did not have enough evidence to decide automatically — reaches none of them. The file you hand to an ad network must not contain a guess.

The blocks themselves — how they are created, when they expire and how they are pushed to Google Ads or a webhook automatically — are covered in blocks and exclusion lists. This article is about the files.

What Google accepts#

This is the first card on the tab and it is worth reading before any claim. Google does not act on another vendor's verdict, AdPix's included. It reviews raw click logs, so a refund request needs the clicks themselves: when they happened, from which IP and browser, with which GCLID. Google decides on its own evidence, and invalid clicks it already caught are credited automatically, never twice.

In practice: send Google the click quality report, not a summary of verdicts.

The click quality report#

Click quality report exports the invalid clicks in the selected range from the paid-click log — the clicks with a Confirmed or Potential verdict — in the column order of Google's Click Quality form:

Column What it holds
Click timestamp (UTC) when the click landed, in UTC
IP address the click's raw IP, only while the click is still inside the 90-day retention
User agent the browser string
Device the device type
GCLID Google's click ID (gclid, gbraid or wbraid); empty for clicks from other networks
Campaign the campaign
Landing page the landing page
Reason codes the verdict's reason codes, space-separated, such as ip_velocity_10m
Tier the verdict tier: confirmed or potential

Three things to know about this file:

  • The raw IP is kept for 90 days. After that the IP address column is empty for that click. If your claim needs IPs, download the file before 90 days have passed.
  • Every download is written to the audit log, because the file carries raw IPs.
  • It is written in an encoding that opens cleanly in Excel, and one file holds up to 100,000 clicks.

Refund evidence (networks)#

This is the file you hand to any ad network for a network-level claim. There are two ways to get it, and both follow the selected range:

  • On the Evidence tab, Refund evidence (networks) — every network sealed in the range.
  • On the Networks tab, in a network's detail drawer, Download evidence (CSV) — the same file, for that network only.

Each row is one network-and-channel pair sealed in the range:

Column Meaning
asn · asn_org · network_type the network the finding is about
channel the channel the traffic arrived on
severity the verdict tier at the range's latest seal — Confirmed or Potential
window_sessions that network-and-channel pair's sessions in the range
window_from · window_to the selected range
first_sealed · last_sealed · seals the first and last seal day in the range, and how many seals
estimated_wasted · currency the estimated wasted spend and the website's currency
reasons the same reasons shown in the network drawer, in plain language
sample_ips real IPs taken from the sealed evidence
evidence_hash the chain hash of the latest seal
chain_verified whether the website's whole chain was intact when the file was built

The waste estimate uses the wasted spend method: the channel's spend multiplied by this network's share of all of that channel's true sessions in the range, and never more than the cost imported for that channel. Spend is read from the ad cost you imported yourself, in the website's currency; cost rows in another currency are left out, not summed. With no cost imported, the column stays zero.

The file now follows the range

Before this update the file returned the website's whole sealed ledger, and neither the page's range nor the network you had opened narrowed it. Now it carries only the seals inside the range and, when taken from a network's drawer, only that network. For a specific claim, set the range first — and leave the hash of every row you keep exactly as it is, because that hash is what can be checked.

The evidence-chain status#

Under the two range files, one line gives the chain's status. Every time this tab, a network's drawer or the refund file is read, the whole chain is recomputed and compared:

Message What it means
"Evidence ledger verified: … sealed entries, none altered." the chain is intact; the evidence can be sent
"Evidence ledger integrity check FAILED at entry …. Do not submit this evidence; contact support." a row was changed or deleted outside the normal path; the number is the first point of divergence
"No sealed evidence yet. It appears after the first scoring run that flags a network." no network has been sealed yet

The evidence ledger, and what it is for#

Behind the refund file is a ledger, not a report. Every time the nightly scoring run executes, each flagged network is appended to that website's ledger as an immutable row. Nothing is overwritten or merged; every seal is a new row.

Tamper evidence comes from a hash chain: each row's hash is built from the previous row's hash plus that row's own content, and the first row of each website starts from a fixed genesis value. Editing or deleting any past row invalidates the hash of every row after it.

What the chain protects is a deliberate selection: the sequence number, the date, the network and channel, the verdict tier, the session counts, the sample IPs, the reasons, and the window. The model's floating-point scores are deliberately excluded: bit-for-bit reproduction of a float is not guaranteed, and a chain that breaks for no reason is worthless. Sealing and verification are two separate systems producing exactly the same bytes — that separation is what makes the check mean anything.

What it proves and what it does not#

  • It proves the record has not changed since it was sealed — not by you, and not by anyone else. A hand-built spreadsheet cannot say that.
  • It proves each row is tied to one window, one network, a session count and a specific set of reasons, and was not assembled after the fact to fit a claim.
  • It does not prove that the ad network owes you a refund. Each network's policy is its own.
  • It is not a third-party attestation. It is the AdPix ledger with tamper evidence, not the signature of an independent auditor.

The exclusion lists#

The Exclusion lists card holds three files for stopping the next spend. The same card is on the Networks tab, so a website on Starter (monitoring, without the Evidence tab) downloads the lists there. All three give the current state and do not follow the date picker. All three are built from one list:

  • the active blocks on the Blocks tab — manual, real-time and nightly;
  • the IP addresses of Confirmed paid clicks in the last 30 days — never a shared mobile-carrier address;
  • the /24 blocks of networks scored Confirmed in the last 30 days that are data center, VPN, proxy, Tor or residential-proxy networks, or on the known-fraud list — never a residential or mobile network.

A block you lifted with Release, and anything on the Allowlist, never appears in this list; a human made that call.

A file with one column named IP address, each row in one of these forms: a single IP, a /24 block written 203.0.113.*, or an IPv6 /64 block. Those are the forms Google Ads accepts in a campaign's excluded-IP list. Device blocks do not go to Google, because Google accepts IPs only.

Google accepts 500 IP exclusions per campaign, so the file has at most 500 rows: Confirmed first, then most recent, then most hits. The rest is not dropped silently: the download carries an X-AdPix-Blocklist-Truncated header with the number of rows left out — a number that does not appear inside the CSV itself — and the Blocks tab also shows the overflow when the active address blocks exceed 500.

Each a.b.c.* block is 256 addresses

A /24 exclusion closes the whole block, not one IP. AdPix only builds these blocks from confirmed data-center and proxy networks, but the import is your call. If this is your first time, take the Generic blocklist first and look at its type and reason_codes columns.

Generic blocklist#

Every entry on the list, uncapped, for a firewall, n8n or another ad network. Its columns are value, cidr, type, severity, reason_codes, hits and source: the value, the same value as a CIDR, the identifier type, the tier, the reason codes, the hit count and the origin.

Meta network advisory#

Meta has no IP exclusion, so this file is not an IP list; it is a list of the networks themselves, and it is advice, not a list to upload. Each row is one network-and-channel pair scored Confirmed or Potential, with the columns asn, asn_org, network_type, channel, severity, flagged_sessions and recommendation, ordered by session count.

There are two things to do with it: exclude those networks through the campaign's placement and audience controls, or filter events coming from them server-side before you send them to the Conversions API.

One refund claim, end to end#

1
Set the page's date range to the exact period you are claiming for. The click quality report and the refund file both follow it.
2
For Google: on the Evidence tab, download the Click quality report and send it with your invalid-click investigation request to Google. If you need the IPs, do it before 90 days have passed.
3
For other networks: on the Networks tab, read the wasted-spend figure and the Channel × network breakdown, then select the main networks and read their reasons. If you cannot explain the reasons, do not file the claim.
4
Download Refund evidence (networks), or that network's Download evidence (CSV), and check that the chain-status line says verified. Do not touch the hash column.
5
To stop the spend, import Google Ads IP exclusions into the campaign's excluded-IP list — or push blocks automatically, as covered in blocks and exclusion lists.
6
Attach the evidence file, the invoice number and the date range to the ad network's dispute ticket or form.

To dispute a single traffic slice rather than a network, that entity's drawer on the Sources tab has Download dispute evidence. That file carries the verdict summary, the reasons, each detector's score and the network context, and is written in an encoding that opens cleanly in Excel. Cross-business signals appear only as counts and no other organisation's identifiers are disclosed; if your access to revenue data is restricted, the conversions field reads [restricted] instead of a number. Downloading it counts as a sensitive read and is written to the audit log.

When a file comes back empty#

  • The click quality report has only a header row — no Confirmed or Potential paid clicks were recorded in this range. Widen the range, or check the Clicks tab for whether paid clicks are being recorded at all.
  • The IP address column is empty — those clicks are past the 90-day retention.
  • The Google Ads file is empty — there are no active blocks and no confirmed clicks or networks in the last 30 days.
  • The wasted-spend column is zero — no ad cost imported for the range, or the channel names do not reconcile to the AdPix taxonomy. Start with Importing ad cost.
  • The chain is not intact — the ledger was altered or a row deleted outside the normal path. Do not send that evidence and contact support; the entry number reported is the point of first divergence.

Frequently asked questions#

Will Google refund me on the strength of an AdPix verdict?

No. Google does not act on another vendor's verdict; it reviews raw click logs and decides on its own evidence. That is why the click quality report gives Google the clicks themselves — time, IP, browser and GCLID — rather than a verdict. Invalid clicks Google already caught are credited automatically, never twice.

Why does the Google Ads file have at most 500 rows?

Because Google Ads itself accepts at most 500 IP exclusions per campaign. AdPix ranks the list by severity, then recency, then hits, and keeps the first 500. The rest is not dropped silently: the download carries an X-AdPix-Blocklist-Truncated header with the number of rows left out, and the Blocks tab shows the overflow when the active blocks themselves exceed the cap.

What exactly does the evidence chain prove?

It proves these findings have not been altered since they were sealed — not by you, not by anyone. Each row is built from the hash of the row before it, so editing or deleting any past row breaks every later row, and AdPix reports that on read. It is not a third-party attestation and it does not oblige an ad network to refund you; it is an auditable record that makes your claim checkable.

Do the `/24` blocks in the exclusion lists catch real users?

A /24 block only comes from a network scored Confirmed that is a data center, VPN, proxy, Tor or residential-proxy network, or on the known-fraud list — never from a residential or mobile network. Shared mobile-carrier addresses never enter the list either. Each a.b.c.* block is still 256 addresses, though; look at the generic list before you import.

Build with the APIUnderstand where revenue comes from.
Was this page helpful?