Advertisers
Checking that their own campaigns ran as booked before they report on them.
AD VERIFICATION
Static residential proxies for ad verification
An ad system that recognises the advertiser may show the advertiser the right thing. Verification means arriving as an ordinary visitor in the market you bought, from an address on an internet provider's network, and recording what was served to that visitor.
This scenario needs no prior approval. Ad verification is named in our acceptable use policy as open by default.
In short: Ad verification is checking that a campaign ran as booked (the right creative, in the right market, on the right pages) by recording what an ordinary visitor there was served. That needs an address in the market that is on an internet provider’s network, has no history on the advertiser’s account, and stays the same between runs: one IP held for the period you pay for.
The Interactive Advertising Bureau and the Media Rating Council describe ad verification as a service for determining whether campaigns were executed appropriately. Their Guidelines for the Conduct of Ad Verification (2012) name its main lines: site context, geo-targeting, ad placement, competitive separation and fraud detection.
Why ad verification proxies need a residential network
The Media Rating Council’s invalid-traffic standard, listed on the same page, names known invalid data-center traffic among the general invalid traffic that measurement filters by routine means. A check arriving from a hosting company’s address can therefore be classed as invalid, and what it is served, or whether it is counted, may not match what a household in the market receives. An address in the market on an internet provider’s network is not caught by a filter that screens hosting ranges. The same standard lists invalid proxy traffic (a proxy that exists to manipulate counts or pass on invalid traffic) among the sophisticated invalid traffic measurers must detect. We read that as a reason to keep a check in your own verification run, reported as such, and out of the delivery you count. It does not make the check invisible, and it does not stand in for a carrier connection on mobile placements. What this section attributes to the MRC rests on its published guidelines; everything else is our reading of how ad systems behave, and we have measured no ad system ourselves.
HOW A CHECK IS BUILT
Three decisions, and the rest is recording
A page fetch proves little on its own. Verification depends on the fetched page matching what a real person would have been served.
Step 01
Buy the market. The catalogue’s unit is the country, so match the country and the provider network the placement targeted; a campaign aimed at a single city needs a vantage point this product cannot promise. The catalogue names the provider behind every address, so you know before you buy which network a lookup of your source address will show.
Step 02
Keep one address per campaign or per market. Verification is comparative: this week against last week, this placement against that one. If the vantage point moves between runs you cannot tell a campaign change from a network change.
Step 03
Record the whole chain. Redirects, interstitials and the creative itself are all part of what was served, and failures such as a wrong redirect or a swapped creative often sit mid-chain.
WHO THIS IS FOR
Who runs these checks
Agencies
Checking campaigns their clients bought before a report goes to the client.
Ad-operations teams
Recording what a campaign served in each market, from one address per campaign or per market.
CHECKLIST
Hold these constant as well as the address
A browser profile that has never signed in to the ad platform or the advertiser’s site.- A fresh address with no history on the advertiser’s account, so the response is the one the audience gets.
- A device type and user agent that match the placement: desktop inventory from a desktop browser.
- A timestamp and a screenshot for every run, kept with the redirect chain.
- A sample no larger than the check needs. Section 5 of our acceptable use policy prohibits generating artificial traffic, clicks or impressions.
WHAT EACH CHECK PROVES
What a verification run can and cannot establish
| Can establish | Cannot establish | |
|---|---|---|
| Creative served | The exact asset and copy returned to your request | That every impression in the flight matched it |
| Geographic delivery | That this market was served at this moment | The share of spend that reached the market |
| Landing page behaviour | The redirect chain and the page as rendered | Whether a user completed anything |
| Placement context | The page the creative appeared on when you looked | Where it appeared for anyone else |
| Brand safety | Adjacency observed on the pages you sampled | Adjacency across inventory you did not sample |
| Compliance wording | Whether required notices were present in your render | Intent, or how a regulator would read it |
RELATED
Other scenarios, and the product itself
FAQ
Questions about ad verification
Why can I not just check the campaign from the office?
Two reasons, and both are structural. Your office network is in one market and the campaign probably is not, so you see the wrong geography. And the ad platform knows the advertiser: a logged-in account, or an address with the advertiser's history on it, may be served differently from the audience. Verification only means something when the request is served the way the audience's requests are.
Do I need a residential address, or would a datacenter one do?
For the geography alone, a datacenter address in the right country will sometimes do. For anything where the placement itself is filtered by traffic quality, it will not. Hosting ranges are easy to identify from public registration data, and the Media Rating Council's invalid-traffic standard treats known invalid data-center traffic as something to filter out. An empty slot then tells you little about whether the campaign was running. An address on an internet provider's network removes that one source of ambiguity. Datacenter
Should the address rotate between checks?
Generally no, for this job. Verification is a comparison over time, and a fixed vantage point is what makes two runs comparable. Rotation is useful when you need many independent samples of the same placement, which is a different question, and each load of a live placement may be counted as an impression, so keep samples small. If you need both, run them as two separate exercises with two separate address strategies.
Can I verify mobile placements this way?
Not properly. Placements targeted at mobile carriers are served over carrier connections, and a static ISP address is not one. That check needs a carrier address, which is not on sale here yet. Verifying a carrier-targeted flight from a fixed-line connection measures something the campaign was not set up to show, and it is better to say so than to file the report. Mobile
How many addresses does a verification programme need?
One per market you are checking, at minimum, because a market is the unit of geography the product exposes. Beyond that it is a question of how much parallelism the schedule needs, remembering that every page load of a live placement can register as an impression. Each address carries 100 concurrent sessions, dropping to 10 past 50 GB in a calendar month (Terms of Service, section 2.4). Terms of Service
Is this allowed under your acceptable use policy?
Yes, explicitly. Ad verification is one of the use cases named as requiring no prior review: you order and start. The list of targets that do need a verification step first is short and specific, and none of it describes checking campaigns you or your clients bought. The same policy prohibits generating artificial traffic, clicks or impressions, which a verification schedule should never amount to. Acceptable Use Policy
Use this to answer one question for each market you bought: was this served, here, at this time? Do not use it to report delivery share or reach, which need the campaign’s own measurement, or carrier-targeted placements, which need a connection that is not on sale here yet. If it fits, buy one ISP proxy per market in the campaign (each priced per IP per period) and run the same check on a fixed schedule.
Sources
- Media Rating Council Standards and Guidelines: the IAB/MRC Guidelines for the Conduct of Ad Verification (2012) and the invalid-traffic standard that names data-center and proxy traffic