Click on image for the detailed report.
This post was first published in the Wall Street Journal – to see it click here
In Defence of Trading Desks
The World Federation of Advertisers report on programmatic trading, issued last week, has set the online ad industry abuzz. I am pleased to see clients taking a stand on transparency and some of the other issues surfaced in the report, despite being one of the purported programmatic culprits.
When Publicis launched Audience On Demand in 2008, we decided to create it as an alternative to the very murky services that were dominating the marketplace at the time, such as ad networks, that operated in the dark and sometimes pocketed triple digit profit margins in the process.
Six years later we are standing firm on our early decisions, and reports like the one issued last week suggest the market is moving in our favor.
But the fact is not all agency trading desks are created equal. And while the WFA report inaccurately tries to paint us all with one color, I encourage every marketer in the industry to take note of the questions in the report that relate to issues such as arbitrage and data. Don’t just ask these questions of your agency trading desk, however; Ask them of every programmatic provider you might be spending with today.
If a programmatic provider is working in a marketers’ best interest it should not be arbitraging inventory, it should be buying audiences and inventory transparently in real time. It should be protecting marketers’ data (it’s their data, and they should honor it as such, unless given permission to blend it). It should have a rigorous vetting process to evaluate all data and technology partners to be sure that protection extends across the ecosystem.
It should also be tireless in pursuit of viewability and quality, and it should show you how it is trying to protect your ads from fraud. I submit that an in-house option or managed service demand-side platform that buys on a marketer’s behalf will provide less brand safety than an agency trading desk. It simply costs too much to deliver extensive black and white lists, tech vetting and human vetting at a client level.
Finally programmatic providers should make it entirely clear what percentage of marketers’ ad dollars are actually spent on ad space, and it should be far, far greater than 40% as the WFA report suggested. That number is ridiculous. Candidly, a fair amount of the math cited in the WFA report is peculiar.
In general, the WFA report steers marketers toward setting up an in-house solution. It’s a viable, though difficult and limiting proposition to pursue. An in-house operation is not going to resolve all transparency issues. It might give marketers complete control, but it also results in limited visibility once the campaigns go out the door, and you are only as good as the technology you tie yourself to.
Meanwhile, if marketers outsource to an ad network, managed service DSP or non-disclosed trading desk, you have little control, less visibility and no ownership.
I hope I get a chance to meet the WFA. I would love to talk to authors of this report about their findings, where the insights were obtained and how the calculations were done. So much of the report is spot on in terms of what questions to ask, but the bias and inaccuracies need to be corrected.
I am passionate about how we cannot let this scycle of ever decreasing cpms continue and I lay the blame at all our doors, agencies, advertisers and auditors. Exchangewire covers this topic with contributions from me. Original article here.
Click fraud has undoubtedly been one of the topics of conversation in the programmatic advertising sector in 2014, with Google’s purchase of UK-based security specialist Spider.io just one of a number of industry moves underlying its growing importance.
Last week Rocket Fuel was fingered in a FT article highlighting its prevalence in the industry (of course it was quick to rebuff the article’s assertions), but the entire advertising – from client-side marketer to third-party ad tech vendor – must accept their role to play in allowing it to continue.
This comes on the back of other articles in mainstream press – for instance a Wall Street Journal article claiming that up to a third of all web traffic is “bogus” – pressing the issue further for the online advertising sector to improve transparency over media buys taking place via automated channels.
Moves to tackle the issue of click fraud (or bot traffic) began to gather pace last year when the IAB’s US chapter established the Traffic of Good Intent (TOGI) Task Force, in a move demonstrating that programmatic ‘media trading’ sector was maturing as braces itself to become a mainstream player, as opposed to an emergent force.
In fact during the last two weeks alone Dstillery announced it was received a patent for its fraud detection technology from the US Patent Office, this follows the news that the similarly named Distil Networks’ bagged $10m in Series A funding just last week.
More recently, the Alliance for Audited Media (formerly the Audit Bureau of Circulations) announced it was absorbing fellow auditing service ImServices.
So while it is clear that there is a near universal intention to wipe out such practices, but it’s probable that the fraudsters will always be on step ahead of the industry’s security brigade byinventing new ways to game the system (some fraudster techniques are quite comprehensively discussed here).
How can individual parties minimise the impact of click fraud?
But that’s not to say that measures cannot be taken to minimise the impact of online ad fraud, and with this in mind, every tier of the industry has to take their share of the blame in letting this happen.
As discussed in previous articles in ExchangeWire everybody has their part to play in minimising the detrimental effects of elements of the ‘bad internet’, and if parties are proactively taking measures to improve things, then they’re part of the problem.
Speaking previously with ExchangeWire Dr. Thomas Servatius, IPONWEB, head of client services, identified that the rise of the programmatic industry had allowed fraudsters to thrive online, with the scale of web traffic allowing rogue players to put sites which generate traffic by non-human means on ad exchanges.
“The problem is that when an advertiser buys traffic on a fraud site, it usually comes very cheap – much cheaper than human built sites [thus opening the opportunity for arbitrage from third-party players and media agencies] – and it has good click through rates.
“So if you have fraud in your advertising mix, what you see as an advertiser is that for a small amount of money, you get a good number of clicks,” he explained.
Explore what will KPI’s look like in a post-click fraud market?
He went on to further relay anecdotal evidence of the internal dynamics that encourage brand-side marketers (the people who are ultimately being ripped off here), from concealing the issue.
Indeed Cameron Hulett, Undertone, executive director, EMEA, further explains that such is the scale of the problem that most campaign benchmarks after a “post click fraud market correction” would be largely redundant.
For instance, most marketing KPIs, such as reach and traffic are drastically inflated by bogus web traffic as it currently stands, causing problems for parties on both the buy- and sell-side alike, contends Hulett.
Hence, it is in the interests of a lot of parities to let this white elephant in the room to go unaddressed, according to some.
Prioritise quality over cost-cutting
Marco Bertozzi, President Audience On Demand EMEA and North American Client Services at VivaKi, argues that the entire industry is incentivised to prioritise lower CPMs (ergo poorer quality inventory, or even bot traffic through long-tail exchanges and networks) instead of quality content (where prices are higher).
“I think educating marketers on the importance of paying more for quality inventory will need to happen because the buy and the sell side are chasing KPIs determined by said client who may be calling for lower CPMs versus quality interactions,” he says.
“If the only metric focused on by auditors and advertisers is lower cpm, then that’s where everyone will focus – turning a blind eye to the lack of quality and transparency but being happy that a lower CPM was achieved.”
Auditing has not kept up with the pace of change in the Ad Tech space. The industry still clings to CPMs and not the value of the impression and what it can deliver, according to Bertozzi.
“If you look at Search, if standard auditing metrics had been applied to search advertisers would not use it and spend would be non existent as agencies would be told to suppress the CPC. The same now applies to display, it is an auction environment and yet still they want to drive down on cpm,” he adds.
Explore alternatives to CPM pricing and last-click attribution
Meanwhile, Julia Smith , a partner at consultancy firm 614 group, and acting MD of Evolve Media, argues that exploring alternative pricing models to selling media on a CPM basis, can make it easier for advertisers and their security partners to detect non-human generated traffic.
“A lot of people are all about the click, and in particular its a problem with the long-tail of sites [meaning non-premium ad networks and exchanges are a particular problem in this regard].
“We can start looking at alternative Using a cost-per-engagement [pricing] model could play an important role in combatting this. While it’s not perfect it can make it harder for click farms to replicate human behaviour.”
However, as mentioned earlier in this piece, fraudsters are just as industrious in their attempts to stay ahead of the security elements of the ad tech industry, with their techniques growing evermore sophisticated.
Sources consulted by ExchangeWire also argued that one fundamental flaw in the ad tech sector that lets poor quality traffic be traded on ad exchanges and networks is the prevalence of the last—click attribution model , which incentivises the entire industry to chase the last click.
Adit Abhyankar, Visual IQ, executive director, says: “Incentives drive behaviour. this is common sense. So if flawed attribution leads to flawed allocation of performance credit, which then leads to incorrect incentives, you can bank on the fact that, it will also lead to bad decisions.”
Meanwhile, Marco Ricci, Adloox CEO of content verification firm Adloox, argues that looking at at specific domains on ad exchanges and networks, for statistics such as CTR per domain and by publisher, is a more sophisticated method of detecting bot traffic.
AOD’s Bertozzi adds: “Attribution, econometrics, understanding business impact will all go a long way to removing an obsession on lowest cpm. It will also focus on the fact that advertisers should be challenging media partners to show where they are advertising line by line. If you have to be transparent about the media placement, you are less likely to buy the long tail.”
Employing sophisticated vetting techniques
Those ad tech players looking to perform blacklists [of sites that are known to have traffic generated by non-human traffic] should perform check such as clickthrough rate (CTR) per domain and by publisher, CTR vs conversions, and CTR vs IP addressees are all useful metrics, according to Ricci.
“We check clicks made in less than one or two seconds we can catch fraud – blink and you’ll miss it. Essentially our clients want a more granular level of transparency than the majority of the market offerings today.”
Bertozzi also argues that those players on the buy-side need to do more to improve the reputation of the sector. He adds: “We provide a rigorous vetting process called VivaKi Verified, which thoroughly evaluates media, data and tech partners to ensure that they meet our standards when it comes to brand safety, consumer privacy and client data protection.
“Rather than buy in the murky pool, we use means to avoid the problems, don’t buy in the murky pool at all.
“We have also a proprietary Quality Index that combines the [safety] signals from partners like comScore, Google, Integral Ad Science and Vindico to all the URLs we have in the AOD marketplace creating our own score.
“Metrics and standards aren’t there yet and adoption needs to happen on a larger scale, but the cost of viewable ad impressions will go up and we need to be prepared to pay them to ensure that better brand-to-consumer interactions are happening. If the only metric is cpm, we are opening up the business to gaming the system.”
So the fact is, regardless of which statistics parties in the ad tech industry subscribe to, as to the extent of the problems of bot traffic, it remains clear that more can be done to address the issues of click fraud.
Those that choose to ignore the problem (for whatever means), are helping to propagate it.
My piece on Digiday outlining the threat of Ad tech disintermediation. First posted here.
I remember sitting with a founder of a well-known demand-side platform a few years back (feels like a lifetime), and he was warning me how the evil Google would disintermediate us all and destroy the agency trading desk business if we were not careful.
The irony now is that the worst culprits of all are the new, up-and-coming tech vendors who are chasing the direct-to-advertiser relationship at any cost.
As an agency, allowing a DSP or real-time bidding ad network to control all the programmatic spend may seem the same as giving an insertion order to an ad network, but it is far from that. The rules have changed with the rise of ad tech. Our whole business is based more and more on data. We need to manage, explore, test and learn with data, and the data needs to be held at the hands of the agency running the wider business, or remain in the advertiser’s hands should they choose to take the process in-house.
To release tens of millions of dollars to a managed service DSP is to release all of your intellectual capital to an external company where the same rules expected of an agency may or may not apply. We see clear benefits when we are able to apply the agency learnings to all the programmatic opportunities. Whether we are looking at cross-channel attribution, econometric modeling or online and offline synchronization of media spend, we can make activity work so much harder in that context — and tie it back to the advertiser’s own data whether on or offline. A third party, or siloed business, simply cannot do the same.
Agencies take heed: This is no longer just a question of outsourcing some digital buying but rather the outsourcing of your agency role and intellect to a third party. You may not recognize the danger, given the modest level of programmatic spend relative to massive TV budgets. But when this spend drifts away, a little bit of control goes with it. Not a good situation given the projected growth of programmatic.
Take a lesson from search. Two things happened in search that made it one of the biggest battle grounds of the agency world through the mid-2000s. The first was that the agencies ignored it when it launched, and the second was they fought tooth and nail to get it pulled back into the agency when it had grown into the mammoth beast that it is today. Today’s DSPs are yesterdays search villains.
An agency digital lead should fight to keep the programmatic business close. Yes, I am biased toward a relationship with an agency trading desk — not just because data-driven, programmatic buying will be the lifeblood of the future media agencies but also because the right agency/trading desk relationship is better for clients.
An advertiser might be attracted to cheaper options. A siloed, third-party provider might “feel” unbiased. But what happens when the market moves (which is does every day), and that marketer is tied to a single provider. They move at the the speed of the provider. Or they pay the significant switching cost. Yes, DSP technology evolves. But their lack of access to the ideal marketplaces may leave an advertiser handicapped. And how will the marketer know? It’s hard to measure performance without any comparison or opportunity to swap (short of making an extensive investment).
The agency relationship should give clients cross-platform, open access to all opportunities — and objectivity. Trading desks should deliver the benefits of relationships, learnings and experience with all of the best DSPs, plus perpetual evaluations of new and evolving partners. They must be able to provide the brand safety, starting with basics like full disclosure on where ads are appearing and how much of your budget was spent on media. It is fascinating to me that Rocketfuel discloses 60 percent margins and there are no concerned glances from advertisers. Really? 60 percent?
I have been warned all my life that Google is the bad guy, but it is becoming clear that as the story unfolds, we are seeing a very different picture. The VC-fueled pressure cooker we are in at the moment is creating disintermediation on a grand scale or at least the potential of it. And agencies and advertisers should both see that there is a major role for their partners in helping them steer through this time so that we don’t walk blindly into a repeat of 2001-2008, an era that both agencies and advertisers regretted longer term.
Discussion of some big issues of the day around video. The role of auditors, advertisers and agencies in Ad fraud / quality issues, how programmatic video can now be a creative way to continue dialogue through re-marketing and publishers ultimately will benefit from the fact tech is starting to reveal how poor many Ad network sells are to clients.