Ad verification tools vs proxies: DoubleVerify, IAS, GeoEdge and HUMAN

Do you need a verification vendor or a proxy?

Both, for different questions. A tag-based verification vendor sits inside the ad call and measures every impression the campaign serves: was it viewable, was the traffic valid, was the page brand-safe, did the creative behave. That is measurement at scale and it is what DoubleVerify, Integral Ad Science (IAS), GeoEdge and HUMAN sell, each with a different emphasis. A proxy check reproduces one impression from one network location at one moment, with the exit, the redirect chain, the final URL and a screenshot on record. That is evidence, and it is what the vendor's aggregate cannot give you when a client asks "what did a user in Munich actually see?"

The proxies for ad verification page covers the proxy side; this article describes the vendors from their public documentation, explains pre-bid and post-bid measurement, and then lists the five situations where an independent proxy check adds something and the several where it does not. Vendor capabilities change; the descriptions below are general and dated 30 September 2026, and the vendors' own pages are the authority on what each product does today.

• Vendors measure viewability, invalid traffic, brand safety and suitability, and creative security across every impression, pre-bid and post-bid.
• Proxies reproduce a single impression from a chosen city and network type, with evidence attached.
• Use proxies for pre-launch geo QA, re-verifying an alert, disputes, affiliate and redirect audits, and competitor monitoring.
• Do not use proxies to measure viewability, invalid traffic rates or reach; that is the vendor's job and a proxy cannot do it.

What tag-based vendors measure

DoubleVerify publishes coverage of viewability, fraud and sophisticated invalid traffic, brand safety and suitability, and attention metrics, delivered through a measurement tag on the creative and through pre-bid segments in demand-side platforms. Integral Ad Science covers the same four areas (viewability, invalid traffic, brand safety and suitability, and context) with pre-bid and post-bid products and industry accreditations for its measurement. Both are the vendors a media plan usually names when it says "verified impressions".

GeoEdge is focused on creative security and ad quality from the publisher and platform side: detecting malicious and disruptive ads (auto-redirects, malvertising, forced downloads, cloaked landing pages) and blocking them in real time. Its documentation also describes using proxy services to QA how campaigns render in different geographies, which is the overlap this article is about. HUMAN (formerly White Ops) specializes in bot mitigation and invalid-traffic detection across advertising, including pre-bid filtering of bid requests and post-bid measurement of fraud.

All four work at the scale of the campaign: millions of impressions scored by rules and models, reported as rates and segments. None of them reproduces one impression for one visitor, and none of them is designed to show what a specific publisher page served to a specific city on a specific afternoon. The IAB's guidelines for the conduct of ad verification set the rules the vendors operate under, and are worth reading before you decide what your own checks should record.

Pre-bid vs post-bid

Pre-bid measurement happens before an impression is bought. The vendor scores the bid request (the page, the app, the user's network type, the likelihood of viewability or fraud) and the buyer's platform uses that score to bid or not. This is how known data-center traffic, suspicious apps and unsafe pages are avoided in the first place, and it is why a check from a datacenter IP sees a different auction than a household does: the pre-bid filters have already removed most of the demand for that impression.

Post-bid measurement happens after the impression is served. A tag inside the creative reports whether the ad was viewable, on what page, to what kind of traffic, and whether the creative or its landing page misbehaved; the vendor may also block a creative from rendering. This is where alerts come from: "3% of impressions on publisher X were sophisticated invalid traffic", "creative Y triggered a redirect on mobile in Brazil".

A proxy check is neither. It is an independent request that behaves like a visitor from a chosen place, made by you, with the response recorded. It can confirm or contradict what a post-bid alert says happened, and it can test geo-targeting before the pre-bid rules ever run, but it does not see the auction and it does not measure across impressions.

Where an independent proxy check adds value

1. Pre-launch geo QA. Before a campaign spends, load the placements from a residential exit in each targeted country and one excluded country. A geo-targeting mistake costs nothing to catch at this stage and a great deal after. The country-by-country guide has the checklist.
2. Re-verifying a vendor alert. A post-bid alert says a creative redirected on mobile in Brazil. Load the creative's landing page from a residential exit in São Paulo with a mobile profile, record the redirect chain, and you have either a reproduction with a screenshot or a negative result with the exit and time noted. Either is more useful in the conversation with the network than the alert alone.
3. Evidence for disputes. Make-goods and refunds are argued over what was served. A screenshot with an exit reference, timestamp, URL and status, filed in a declared batch with its sample size and limits, is evidence a network or an affiliate program can check. The evidence kit is that file format.
4. Affiliate and redirect audits. Verification tags do not follow your affiliate links. Loading them from each market and recording every hop, the final storefront and whether tracking parameters survive is a proxy job, covered in how to test affiliate links by country.
5. Competitor monitoring. No vendor measures a competitor's campaign for you. The ad libraries list their creatives by region; a residential exit shows where those creatives run and what their landing pages serve locally.

Where it does not

Viewability, invalid-traffic rates, attention, reach and frequency are population measurements and a proxy check is a sample of one. Running a thousand proxy loads does not produce a viewability rate; it produces a thousand samples from your own automation, which is itself non-human traffic and should be kept small and rate-limited for that reason. Brand-safety scoring across a publisher's whole inventory is a vendor's crawl, not a spot check. And a proxy cannot see the auction, so it cannot tell you why an impression was or was not bought.

There is also a limit inside the spot check. A residential exit removes the network-type difference; it does not reproduce a specific user's cookies, device, app identifiers or history, and publishers can still evaluate browser and behavior signals. A placement difference between two loads is evidence to repeat and investigate, not proof by itself. The ad verification proxies page states what a check can and cannot prove, and the evidence template has a field for the limitations of each batch so the caveat travels with the data.

A workflow that uses both

1. Plan: the media plan names the verification vendor and the pre-bid segments; the QA plan names the markets, placements and landing pages to spot-check and the evidence fields to record.
2. Before launch: load each placement and landing page from a residential exit in each targeted market (city or ASN where the campaign is bought that way) and from one excluded market. File the batch.
3. In flight: the vendor tag measures every impression; a small scheduled sample of proxy checks per market catches rotation, redirect and localization problems the tag does not report.
4. On an alert: reproduce it from the market and profile the alert names, record the result either way, and attach it to the ticket with the network.
5. At reconciliation: the vendor report carries the rates; the evidence batches carry the screenshots and chains for anything disputed.
6. For affiliates and competitors: proxy checks only, on a schedule, with the same template.

The proxy side of that workflow costs little. A full render with a screenshot measured 3.8 MB at the median on 30 September 2026, which is about $6 per 1,000 checks on the smallest TrueProxies traffic pack and about $2 on the largest; the cost per check article has the table. City, region and ASN targeting on GB-based plans keeps the spot checks aligned with how the campaign was bought.

FAQ

Do proxies replace DoubleVerify or IAS? No. Vendors measure every impression for viewability, invalid traffic and brand safety. Proxies reproduce single impressions with evidence. Campaigns use both.

Which ad verification companies exist? The commonly named tag-based vendors are DoubleVerify, Integral Ad Science, GeoEdge and HUMAN, alongside platform-side tools such as Google's Ads Transparency Center and Meta's Ad Library. This list is not exhaustive and it is not an endorsement; check each vendor's current documentation.

Can a vendor see what a proxy check sees? Not the same thing. The vendor's tag reports from inside impressions that were bought and served; a proxy check is an independent request you control from a chosen place. Cloaked landing pages, affiliate redirects and localized prices are typically outside the tag's view.

How many proxy checks are enough? Enough to notice a problem, not enough to become one. Ten checks per market per day at different hours, doubled where a difference appears, is a common starting point. Keep the sample small, never click ads, and rate-limit the automation; your checks are non-human traffic too.