How MGID Detects and Prevents Invalid Traffic (IVT)

Introduction

At MGID, traffic quality and transparent metrics for advertisers and publishers are a core part of the platform. The Fraud Prevention team runs proprietary controls to detect, filter, and remove invalid traffic (IVT).

This article is MGID’s Description of Methodology for IVT: what is filtered, what advertisers see in reporting, and when campaign metrics are corrected. Detailed detection logic is not disclosed.

Invalid traffic is traffic that does not represent genuine user engagement. It is grouped as:

General invalid traffic (GIVT) — identified with routine list- or parameter-based checks (for example known bots, crawlers, and data-center IP addresses).
Sophisticated invalid traffic (SIVT) — identified with multi-signal behavioral analysis and, where applied, independent third-party verification (for example emulated environments or coordinated invalid activity).

The monetizable transaction for MGID native campaigns is the click. MGID does not sell viewable impressions as a monetizable metric.

How MGID Detects Invalid Traffic

MGID uses a multi-layered system: list-based filtering, behavioral analysis, and third-party verification. Proprietary IVT controls run on 100% of monetizable transactions that MGID handles.

1. List-based filtering

MGID maintains internal exclusion lists and applies them to traffic:

IP lists — including the TAG Data Center IP List (checked daily) and MGID internal/dynamic IP lists;
Domain and app lists — updated as new threats are identified;
User-agent lists — used to filter known invalid or non-browser clients.

Traffic matching these lists can be rejected before it is billed.

2. Behavioral analysis

MGID reviews traffic patterns to find activity that does not match legitimate use, including unusual click timing or repetition, device and browser inconsistencies, and other multi-signal checks. This is how SIVT is identified when it does not match a known list.

3. Third-party verification

MGID supplements internal controls with:

Pixalate (MRC-accredited for sophisticated invalid traffic detection and filtration on applicable metrics);
Anura (post-impression monitoring on demand).

Advertisers may also use their own verification tools (for example IAS, DoubleVerify, or Moat). Those tools measure the landing page or the advertiser’s environment and are outside MGID’s billed metrics.

Coverage of Pixalate and Anura depends on the integration (see How MGID reports invalid traffic).

Traffic blocked pre-bid may never appear in reported gross clicks.

How MGID Reports Invalid Traffic

Default campaign statistics in MGID Ads are net of known IVT. A rejected click is not charged to the advertiser and is not paid to the publisher. Rejected clicks are not treated as valid events in MGID Ads campaign totals.

How to read the numbers

Where you look	What it means
MGID Ads campaign statistics (default)	Billed / net clicks after IVT filtration
Rejected clicks in the UI or click report	Clicks identified as invalid and excluded from billing
API `numberOfAcceptedClicks`	Billed / net clicks
API `numberOfRejectedClicks`	Clicks removed from billing as IVT
API `numberOfClicks`	Accepted + rejected (gross of IVT filtration for that request)

Gross (pre-filtration) counts and a GIVT / SIVT split at campaign total are available on request from your Success Manager. Transaction-level detail may be provided on request, subject to privacy rules and to protecting detection methods from reverse engineering.

Coverage by integration

Control	Standard widget code	XML / API feeds
MGID proprietary IVT filters	100% of monetizable transactions	100% of monetizable transactions
Pixalate pre-bid	Sampled 1–100% of eligible ad requests, based on supply-partner risk	Only where the request can be scored in the bid path
Pixalate post-impression	100% of click traffic that can load the tracking pixel	Not available (no third-party pixel)
Anura post-impression	On demand	Not available (no third-party pixel)

GIVT is typically filtered in real time. Additional SIVT rejection can be applied during the calendar day of the click, before that day’s statistics are finalized (and in any case within 24 hours). After finalization, that day’s click status is not reopened.

Measurement pixels on the advertiser landing page (analytics or third-party verification) are outside MGID’s measurement.

When MGID corrects campaign metrics

Identified IVT is treated as material when it is more than 5% of a campaign’s billed clicks and the corresponding amount is no less than USD 5. MGID then applies a correction for the detected invalid traffic, typically for the last 30 days. The lookback can be extended in individual cases.

Advertisers can also request a review through their Account Manager or Success Manager. Corrections are applied through MGID’s internal process. This article does not promise a fixed refund SLA.

If you disagree with a filtration or blocking decision, contact your Account Manager or Success Manager. The Fraud Monitoring team reviews available campaign and click data. MGID does not disclose detailed detection rules.

Industry Recognition and Compliance

MGID is a member of the Trustworthy Accountability Group (TAG) and holds the Certified Against Fraud (CAF) seal. Certification is renewed annually.

Covered party: Intermediary
TAG ID: d4c29acad76ce94f
Status: search MGID or this TAG ID on the TAG Registry

MGID also:

honours publisher ads.txt and app-ads.txt files;
publishes sellers.json;
follows Coalition for Better Ads, Acceptable Ads, and AdChoices programmes.

Privacy and unknown IVT.

MGID processes technical data (such as IP address, user-agent, and antifraud signals) to prevent invalid traffic, in line with the Privacy Notice and applicable GDPR rules, including legitimate interest in maintaining security. If signals are missing because of legitimate privacy restrictions, that traffic is treated as unknown and is not included in GIVT or SIVT rates. If missing or manipulated signals are a supported attempt to evade detection, MGID treats that traffic as IVT, not as unknown. A breakout of unknown volume due to privacy restrictions is available to qualified buyers on request.

Decision Rate and unknown traffic.

MGID reports an IVT decision rate: the share of billed clicks for which enough signals were available to make a GIVT and/or SIVT determination. Clicks without sufficient signals are unknown (MGID’s Blind Zone). They are not labelled valid or invalid and are excluded from GIVT and SIVT rates. Coverage differs by environment: the standard web widget supports full proprietary GIVT and SIVT; XML/API feeds receive list-based GIVT, while behavioural SIVT and third-party pixels are limited. Pixalate and Anura coverage is reported by those vendors and is separate from MGID’s proprietary decision rate. A current Blind Zone summary, including a property-level extract where material, is available to qualified buyers on request.

Material IVT and campaign corrections

MGID’s default advertiser reports are net: known invalid clicks are removed before billing. A rejected click is not charged to the advertiser and is not paid to the publisher.

MRC treats 5% of reported activity as material. For MGID the reported (monetizable) activity is the click.

MGID takes corrective action (credit / make-good of billed clicks later found to be invalid) when both are true for a campaign:

invalid traffic is more than 5% of the campaign’s billed clicks; and
the corresponding billed amount is no less than USD 5.

Below that threshold, invalid clicks are still excluded from net reporting going forward; an automatic historical make-good is not required. MGID may still correct a campaign at the account manager’s request.

A typical correction window is the last 30 days of the campaign (extendable when the investigation requires it). Detection logic, filter thresholds, and reject-reason catalogues are not published.

FAQ

1. Are the clicks in my dashboard already cleaned of IVT?

Yes. Default MGID Ads reports are net of known IVT. Rejected clicks are not billed.

2. Does Pixalate or Anura scan every impression and every click?

Proprietary filters apply to 100% of monetizable transactions. Pixalate post-impression and Anura require MGID’s standard widget pixel, so they do not apply to XML or API feeds. Pixalate pre-bid is sampled from 1% to 100% of eligible requests, depending on supply-partner risk.

3. How do I get a GIVT vs SIVT breakdown?

Ask your Success Manager for a campaign-total report. Rule-level detection recipes are not provided.

4. Does MGID guarantee 100% fraud-free traffic?

No. MGID minimizes IVT; residual risk remains. Material IVT is handled under the 5% / USD 5 correction policy.

5. Can my website be blocked without prior notice?

Yes. A website may be blocked without prior notice if invalid traffic is detected.

6. Can I receive a detailed explanation of why traffic was filtered or a site was blocked?

MGID does not disclose detailed detection methodologies. That is consistent with TAG practice and protects controls from reverse engineering. Your Account Manager or Success Manager can share campaign-level outcomes where available.

7. My website received IVT that I did not generate. What should I do?

Harden the site: keep IP controls up to date (including industry data-center lists such as TAG’s), use additional suspicious-IP sources where appropriate, and consider CAPTCHA or equivalent protection. Then contact your MGID manager so Fraud Monitoring can review the account.

8. Can I appeal a blocking decision?

Yes. Access may be restored if the source of IVT is identified and corrective action is taken. Repeated violations may result in stricter restrictions.

9. How quickly does MGID detect IVT?

GIVT is typically detected and filtered in real time. SIVT may need additional analysis and can be applied during the calendar day of the click, before statistics are finalized (within 24 hours).