Exchangewire: A Lot of Education Needs to Be Done Around Quality: Q&A with Gorka Zarauz, CQO, Smart
By: Lindsay Rowntree
This article was originally published on exchangewire.com What is a ‘chief quality officer’ and why does ad tech need this new role? ExchangeWire speaks with Gorka Zarauz, recently promoted to the role of chief quality officer at Smart, about what his role entails and how the industry can achieve ‘quality’.
ExchangeWire: You are one of the first people to get the title ‘chief quality officer’. Why is that title necessary, and what does it mean? Will we see more people with that title?
Gorka Zarauz: It means that, at Smart, we’re focusing our efforts and tools on quality. The industry has had a lot of leeway for a while now. It’s time to take things seriously. The bet on quality is obviously the right way to go – and we expect the industry to follow. Yes, I do think we’ll see more people with that title, or a similar one, as the industry better recognises the importance of quality and transparency. This is similar to how the industry came to have ‘chief programmatic officers’ and ‘chief data scientists’.
What does ‘quality’ actually mean for the advertising ecosystem? How can quality be achieved?
Quality, for me, means upholding the highest standards. In our industry, it means maintaining the highest standards for both demand and supply chains. For advertisers, this includes the type of inventory offered, transparency around all transactions, communication, and eliminating any kind of deceptive content. On the other side, publishers need to be confident that no malware or undesired creative runs with their content.
To achieve this, everyone within the advertising ecosystem needs to embrace quality by doing several things; but we need to use common sense in what we’re doing. Historically, the ad tech industry’s lax regulations resulted in some companies loosening their standards. However, things are changing now as advertisers and publishers are demanding strict accountability and those in the middle, such as exchanges, need to make sure that all parties in the ecosystem maintain high standards.
Smart focused on the publisher side for a long time, originally as an ad server. When we entered the world of exchanges and DSPs, that focus remained in our DNA. We knew that we needed to form an independent ‘quality team’ in order to make sure that all such decisions were handled quickly and efficiently.
Fraud is more prevalent on desktop than mobile devices. What’s the reason for this, and are you seeing a rise in mobile ad fraud?
We’ve all heard “this is the ‘year of mobile’” – every single year is the ‘year of mobile’. And each year it’s true, of course. The audience increasingly moves towards mobile devices. Mobile banner CPMs are still low (USD$2, on average) compared to higher video CPMs (USD$6, on average), so the fraudsters are focusing on video right now. Partly this is because that’s where the money is; but it’s also because, from a technical perspective, it’s more difficult to identify video fraud than it is to identify banner fraud. You can add more layers on video to hide where the video is being delivered. People can more easily hide what they’re doing.
However, fraud is definitely growing on mobile – in-app fraud and fraudulent clicks to get people to download apps are both booming. We’ve all read articles about ‘farms of people’ sitting in rooms, downloading apps for a living. Advertisers think these apps are genuinely popular, so they advertise there. But all too often, they’re advertising to a million empty rooms.
You see ads.txt as only a partial solution to domain spoofing. Why isn’t it the panacea some claim it to be? What can be done to improve it?
Ads.txt is a very good first solution for domain spoofing – but some people seem to expect more from it. In a way that’s good, because the expectation will inspire more people to implement it and develop other initiatives. Ads.txt is only a partial solution because fraud will always be there – people will always try to get around the law.
Some publishers are concerned about ads.txt – afraid they may lose some revenue. In the long run, though, it’s better for the industry that ads.txt be widely adopted, because domain spoofing is bad for the entire ecosystem. Smart is pushing really hard to get our publishers to use ads.txt and helping them implement it. It costs publishers nothing to install it – just drop a file in the server and keep it up to date.
Ads.txt is starting to help, but the demand side needs serious involvement. The supply side is focused on fulfilling IAB initiatives, but some demand partners don’t even seem to know what ads.txt is. The advertisers themselves are extremely concerned, of course. They don’t want their brands to show up on spoofed sites. However, DSPs need to make sure they’re delivering to the right sites. If DSPs don’t change their tech to accommodate ads.txt, then they leave in place a risk point at which things can fail. If the trader doesn’t use it, or the DSP does not provide it, ads.txt fails, no matter how many publishers sign up.
The IAB is already discussing ways to improve ads.txt (e.g., for in-app) for the next Open RTB protocol version. For our part, Smart is preparing to make an important announcement soon about ads.txt. Stay tuned!
How does one perform due diligence when choosing a fraud-detection partner? What are the key things to look out for?
At Smart, we looked for fraud-detection partners that didn’t do any sampling. We wanted to hear about real data. If you tell me there are 100 impressions that are not suspicious, that should mean that you thoroughly scanned 100 impressions – not that you sampled 10, found nothing, and decided that the other 90 were also fine. Fraud-detection partners should be deterministic, rather than probabilistic.
We also prefer partners that provide a score on the suspicion level of a given impression. One source may say only “this is fraud” or “this is not fraud”. We prefer fraud detectors that offer a scale of 1-5: meaning “5 is definitely fraud”, “1 is definitely clean”, and the other numbers lean one way or the other. No system is perfect. It’s important to see the full spectrum so you can make decisions based on real data. A 1-5 scale gives us the ability to do a deep dive into the data to find out more.
Here are some other questions we ask: What kind of products do they have? Do they cover pre-bid, or only post-bid? Is the auction being scanned before it comes to you, or only afterwards? Does your partner enable you to block fraudulent IP addresses? Is the information shareable in order to prevent further fraudulent activities? Is the prospective partner accredited (e.g., by MRC or TAG)?
At Smart, we use both external and internal tools in order to tackle as many fraudulent activities as possible; and we share our experiences in order to promote a cleaner ecosystem. Continuous improvement is crucial.
Fraud will always exist, so how does the industry move forward? Will the strategy be to seek out quality, rather than try to prevent fraud?
Well, first of all, everyone needs to be aware of the forms of fraud that occur – and stay up to date. Second, everyone needs to agree that quality is not an option. There’s a lot of education that needs to be done around quality. Being transparent about what quality is would be extremely helpful, and remains all too rare. We should all be telling each other: “Yes, we stopped working with this company/these people, because they don’t meet quality standards.’’ The more people talk about it, the less scary it becomes.
Talking about quality can be scary for some companies, but it needn’t be. When you have high quality standards, you may lose a few opportunities in the short term but, in the end, high standards improve the industry. That draws in more brand dollars, and everyone agrees it’s the right thing to do. Sometimes it’s just a matter of doing it. I don’t think it’s either ‘seek out quality’ or ‘prevent fraud’ – you can, and should, do both.