Ads.txt: A Web 1.0 Solution To A Web 3.0 Problem

The Sell Sider” is a column written by the sell side of the digital media community.

Today's column is written by Manny Puentes, founder and CEO at Rebel AI.

A certified letter might make its way through a number of post offices as it travels from Point A to Point B, but neither the sender nor recipient care how the letter arrives as long as it arrives in the right place. The same should be true in programmatic advertising.

But in ads.txt, the IAB’s latest tech initiative, the burden is put on the recipient to list all the possible post offices en route so the sender can feel more confident it ended up in the right place. While this might increase the feeling of transparency, it doesn’t solve the core objective and certainly doesn’t simplify it.

The specification sets out to solves a very real problem in the industry – preventing bad actors from making a profit from selling misrepresented or potentially fraudulent inventory – but the implementation in its current form isn’t sufficient to address the complexity of the current supply chain and still leaves channels open for fraud.

Many major technology providers have already voiced their support for the initiative. While it’s a positive development to see the industry coalesce around solving big problems, it’s also important to recognize what the new specification does and doesn’t solve for the industry and the long-term implications it might have.

The False Negative

The ads.txt specification is simple in concept: A publisher must list the companies authorized to directly sell and resell its ad inventory in a publicly available, crawlable text file. Advertising systems, such as demand-side platforms (DSPs), will host a crawler that will cross-reference the ads.txt file to ensure that they are buying inventory from a legitimate source.

As a fraud prevention model, ads.txt collapses when an “authorized reseller” works with a network or supply-side platform (SSP) trafficking illegitimate inventory, either intentionally or inadvertently. As long as the reseller is listed in the ads.txt file, the DSPs will then see the reseller as “authorized” to sell that publisher ID, regardless of where the inventory actually came from. This puts the burden back on the reselling platforms, where it already exists today.

Ads.txt is actually more likely to eliminate other ad companies that are legitimately being called by the publisher page. This is simply a result of an increasingly complex ecosystem where any given ad call travels from publisher to DoubleClick for Publishers (DFP) to any number of platforms (as seen here).

Ads.txt is a tempting way to simplify this complex path by pushing publishers to publicly list their partners, but the existing implementation of ads.txt illustrates that the reality is more complicated.

As an example, here are the current business relationships listed by a real publisher in its Ads.txt file:

The advertising calls on this publisher’s page include calls to different platforms, such as DFP, which calls AppNexus. This is seen as legitimate because AppNexus is declared in the ads.txt file.

However, a typical ad call rarely stops there. AppNexus in turn calls AdForm (as seen below). AdForm isn’t declared within ads.txt as having a relationship with this publisher.

AppNexus is using AdForm as a demand partner, but AdForm could potentially call another DSP to purchase the inventory. If that DSP or even one of its direct advertisers crawls the ads.txt file and determines that AdForm doesn’t have a publically listed relationship with this publisher, they may demand a clawback or take other adverse actions against the platform, even if AdForm legitimately sold the inventory.

The Inevitable Side Effects

Some might argue that this is how the specification is intended to work by putting pressure on unlisted platforms and eventually eliminating them. But these platforms might also be providing better data, higher conversions or other data layers that contribute to the supply chain’s overall value by providing a high yield for publishers and higher conversions for advertisers.

Ads.txt files can also be cached by the DSP for up to seven days. Supply chain partners that may appear to have been invalid at the DSP caching time might be valid at the time of the ad call and vice-versa. Since there’s no historical ledger or versioning, disputes are likely inevitable.

This ends up hurting publishers, which will be the potential victims of clawbacks from ads that were legitimately run on their domain, but through platforms that weren’t listed in the file at the time of caching, through no fault of their own.

If buyers start strictly adhering to the specification as they say they will, this will absolutely drive companies out of business, an effect some are praising. But this will stifle innovation and the opportunity for new companies to compete against giants like Google. And still, nothing is stopping fraud from entering the ecosystem via an authorized reseller or platform.

Net Neutrality For Advertising?

Though the stated initiative of ads.txt is to stop inventory resale, it achieves this by establishing “preferred” channels, which naturally favors the industry’s most influential companies – sort of like a net neutrality for advertising. While it might be tempting to push for simplicity through consolidation, that consolidation will come at the expense of innovation, choice and the open market.

There’s no doubt we need to continue to work together as an industry on these types of initiatives, but we must also make sure the solutions are the right ones for the long-term health of digital advertising.

Follow Manny Puentes (@epuentes), Rebel AI (@Rebel_AI_) and AdExchanger (@adexchanger) on Twitter.

9 Comments

  1. As a Premium Publisher, the issue is that the letter does not go from A to B, but often to P! How it gets there does matter. I would not buy a car from a bloke on the street - i go to a Garage where i know the source of the car.. The media resellers need to come and do deals with the true publishers and get white listed - so ensuring clients investment goes to the Publisher not a technology middleman adding and inflating prices and taking a big chunk from the middle. Ads.txt may still be a 1.0 solution - it is however a simple, easy and potentially revolutionary move in stopping fraud and reducing tech tax in the eco system of Programmatic Advertising.

    Reply
  2. Ramsey M

    +1 - This is a thoughtful, wonky piece. I didn't think the post office analogy worked b/c the post office controls the distribution completely therefore doesn't need to track the hops, but I think the ad call analysis was great. I think v1 of ads.txt has holes as you say but it's a good step towards declaring the distribution and intermediation that's happening to root out bad actors and decrease the costs of executing digital.

    Reply
  3. >If that DSP or even one of its direct advertisers crawls the ads.txt file and determines that AdForm doesn’t have a publically listed relationship with this publisher, they may demand a clawback or take other adverse actions against the platform, even if AdForm legitimately sold the inventory
    DSP would simply stop bidding on that reseller publisher - it doesn't need to take actions against AdForm.

    >specification is intended to work by putting pressure on unlisted platforms and eventually eliminating them.
    Exactly. If SSP is not authorized to sell/resell inventory of a particular publisher, then they should not be allowed to do so. It has nothing to do with curbing innovation.

    Reply
  4. Fantastic perspective - we agree wholeheartedly, this is an ancient way of thinking and basically bullying smaller companies, who provide significant value to the ecosystem, into submission. It was created with good intentions but failed to encompass the many facets of the supply chain.

    This is only going to cause prices to go up significantly for these buyers participating, then the same buyers will delist ads.txt to buy inventory at true market price. Just scan using everything you can, this will solve the challenges. Exclusivity does not mean better pricing nor better performance, inventory quality does!

    Reply
  5. Neal Richter

    Interesting points Manny. However aren't the complexities you highlight in fact good reasons to find simple solutions to ensure the integrity of the supply chain? Advertisers want to show marketing messages to real people on real websites and apps. All else is waste and complexity and 'tax'.

    The entities with the most incentives to fix it are the end points, the publishers and the advertisers.

    I'd encourage you to get involved in the IAB Techlab+TAG collaboration as you've covered ground that's been discussed.

    Note also that ads.txt is the first step, not the last. There's more to come.

    Reply
  6. I am neutral on the idea.
    Obviously, we will NOT implement "hard blocks" on any impression received from a seller not listed on Ads.txt. To Manny's point, the reseller may be legitimate and provide value.
    We WILL however generate alerts based on Ads.txt, and possibly implement a three strike policy.
    Will these alerts be useful? only time will tell....

    Reply
  7. Hi Manny,

    interesting discussion and it does apply in case of Header Bidding / MetaSSP business. But I think that here, the example does not hold as Adform is called as DSP and then delivers directly as an Adserver. There is no call to the Adform SSP. In theory, yes: We could decide not to deliver the impression after winning it with the DSP, but to inject one or our SSP tags and then sell the impression to one of our Demand Partners downstream. However, if we were to do this, this would exactly be the situation that adx.txt wants to limit. Someone selling inventory which they are not supposed to sell.

    For header bidding, you are in general right: AppNexus SSP calls multiple other SSPs which then might not be listed as trusted source (but actually should be), otherwise they should not end up in the header bidding tag on that publishers side.

    So the real effort is at the end at having the adst.xt "right" all the time.

    Reply
  8. Jeff G

    I agree with the Web 1.0 solution to a Web 3.0 problem, but after that the author completely misses the point.

    Publishers don't want ad tech companies they have no relationship with selling their inventory. Ads.txt is a flawed solution to this problem, but it has highlighted how important this is to premium pubs, and for the first time has given them real actionable control over their own inventory in the open marketplace.

    Pubs do not want and are willing to risk the negligible revenue upside from adform buying inventory from AppNexus and auctioning it off to another DSP. What possible value is Adform brining to this equation, when that DSP can just plug directly into one of the 9(!!!!!!) platforms this publisher has already declared working with.

    If there is value, Adform can pitch that value to the publisher and be number 10.

    Otherwise Adform has no business holding an additional auction for that inventory.

    Reply
  9. While I’m all for a solid fair unobtrusive solution to help clean up the fraud, this one seems strange that’ it’s coming fast at a time where theirs allot of new innovations and in housing gaining popularity. also strange that their seems to be a lot of "status quo” firms, news outlets pushing hard for this solution. And after reading this, Ad monsters “ A Look at Ads.txt” articles

    This could be a way to gain exclusivity and wall more gardens and stop the in-housing and innovations. leaving the "status quo” business models intact. the tech and the infrastructure all work and work well, its the people that control the tech who configure it to their own interests, so why would we want a small group of middlemen firms and org’s to have that power exclusively ?

    I’m very interested in learning more of the pros and cons of this as their seems to be a lot of pumping of the pros and squashing the cons PR damage controlling witch is very strange as well.. ..

    Please forgive my english, it’s not my native lingo..

    McCafe

    Reply

Add a comment

XHTML: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>