Bot Traffic IndexThe bot problem, quantified

GIVT and SIVT: The Invalid Traffic Split That Decides Refunds and Reputation

A verification report lands with one invalid traffic percentage, and the argument starts immediately, because that single number mixes the harmless housekeeping of the internet with deliberate fraud, and the two deserve opposite responses.

The definitions, and why they are written like law

The Media Rating Council's Invalid Traffic Detection and Filtration Standards divide invalid traffic into General Invalid Traffic, GIVT, and Sophisticated Invalid Traffic, SIVT. GIVT covers traffic identifiable through routine means: declared search crawlers and spiders, data centre traffic on published lists, duplicated and malformed requests, the background radiation of the automated web. SIVT covers traffic that mimics humans and evades detection: bots on residential proxies with forged device signals, hijacked devices, falsified location and domain data, and sessions engineered to pass the routine filters that catch GIVT.

The split exists because money follows it. GIVT is expected, filtered as standard practice, and generally excluded from billing without drama. SIVT findings trigger the serious conversations: make goods, refunds, and whether a publisher or a network has a fraud problem or a fraud business model. A report that merges the two lets a small amount of SIVT hide inside a normal looking GIVT rate, which is exactly where a fraud operation wants to live.

What each type looks like in a store's own data

Stores meet GIVT daily without knowing its name: search and AI crawlers, uptime monitors, price comparison fetches, the SEO tool auditing a competitor and sweeping your catalogue on the way. It inflates raw session counts, distorts conversion rate denominators, and is a data hygiene problem with a filtration answer. SIVT arrives dressed as performance: sessions with plausible dwell time and scroll depth that never buy, affiliate and lead traffic with fingerprints that repeat across supposedly distinct users, ad clicks from geography the store does not serve, at hours its customers are asleep.

The detection asymmetry is the point. GIVT yields to lists and declared identity checks, the verified bot work. SIVT yields, when it yields, to behavioural and statistical analysis: distributions that are too smooth, device and network combinations that repeat, conversion patterns that break under scrutiny. A store whose bot tooling only runs list checks has a GIVT programme and a SIVT hope.

Asking verification partners the separating question

The practical move is a question, asked of every verification and analytics partner in writing: report GIVT and SIVT separately, state the filtration applied before billing or reporting, and describe the SIVT methodology beyond list matching. The Trustworthy Accountability Group's work on fraud in the advertising supply chain runs on the same premise, that the category definitions are shared precisely so buyers can compare answers. A partner that cannot split the number has told you which half of the problem it cannot see.

Internally, split the dashboard the same way. Routine filtration rates belong in an operations view, watched for spikes when a new crawler or a misconfigured monitor appears. SIVT indicators belong in a commercial view, attached to the channels and partners that get paid on traffic, because that is where the finding converts into a refund claim or a terminated contract. One blended percentage serves neither conversation, and serves the fraud in both.

Related reading on this site: Glossary, About, AI Training Crawlers at Your Store: Allow, Block, or Charge the Question First.

Ask your verification partner this week for GIVT and SIVT as separate numbers, with the filtration method for each. The quality of the answer is the audit, whatever the numbers turn out to be.

Frequently asked questions

What is the difference between GIVT and SIVT?

General Invalid Traffic is identifiable by routine filtration: declared crawlers, known data centre traffic, malformed requests. Sophisticated Invalid Traffic is designed to evade those filters and mimic humans, using techniques such as residential proxies and forged device signals. The MRC standards define the split.

Why does the GIVT and SIVT split matter commercially?

GIVT filtration is standard practice and usually excluded from billing without dispute. SIVT indicates deliberate fraud and drives refunds, make goods, and partner decisions. A blended rate can hide a serious SIVT problem inside a normal GIVT figure.

Can list based bot tools detect SIVT?

Lists and identity checks catch GIVT. SIVT is built to pass them, so detection depends on behavioural and statistical analysis, such as repeated device and network combinations and implausible session distributions. A list only tool leaves the sophisticated half unexamined.

How should a store report invalid traffic internally?

Separately, by type. GIVT rates belong with operations monitoring, where spikes signal crawler or tooling changes. SIVT indicators belong with commercial reporting, tied to the partners paid on traffic, where findings support refund and contract decisions.