Transparency

Combatting ad fraud on the wild wild web

Combatting ad fraud on the wild wild web

Films are made about drug cartels and stories are told of famous bank heists. But why aren’t we talking about one of the biggest organised crimes in the world: online ad fraud? It may not seem as dramatic as many other crime stories, but the elusiveness of digital fraud is one of the many reasons it’s not stopping and we think it’s worth talking about.

By 2025, the World Federation of Advertisers (WFA) estimate that ad fraud is likely to exceed $50 billion, making it second only to the drugs trade in terms of income. The web is turning into the wild west; every advertiser, publisher and adtech company for themselves. Ad fraud is relentless and jeopardising free online content.

But not to worry, there’s a new sheriff in town.

Coull have come a long way since our humble beginnings, we quickly realised how huge ad fraud was and we had to change this. Since then, we have been trying to tackle fraud in the wild wild web.

First of all, what should we all be looking for?

Automated traffic: Otherwise known as non-human traffic (NHT), it is the most common form of ad fraud. These bots can come from software applications which run automated tasks over the internet to simulate human activity.

Invalid traffic: Traffic running through a domain being undesirable, this does not mean the domain itself is bad and therefore would not require blocking. However, the source of the traffic needs to be found and removed.

Ghost sites: Ghost sites are made to resemble real web sites, but have no value and instead host a multitude of advertisements.

Proxy traffic: A proxy allows anonymous access to the internet and can browse the internet without leaving a footprint. This means all ad requests will go through a proxy, so for those monitoring the requests, the only thing visible is the proxy – there’s no way to know who’s behind it.

Cloaked domains: This is when an imbalance between the domain where the ad appeared and the referring domain. This practice enables undesirable properties such as pirate and adult portals to sell inventory under a high-CPM category such as cars or travel.

Spoofing: A malicious party impersonates another device (or user) on a network in order to show ad requests from more reputable sources.

Ad injection: The ad is loaded by ad injection software, often bundled with other software like games and toolbars. This artificially inflates the number of ads on a page and can lead to negative user experience.

With a free and open Internet dependant on ad revenues, it’s important that the entire chain, along with industry associations, work together to ultimately strike out the risk of advertising fraud.

Here’s Coull’s advice and ethos:

Transparency

Being able to be completely open and honest with everyone in the industry means that, things like invalid traffic and ghost sites are easier to detect. Without transparency, it’s like finding a needle in a haystack…in the dark.

Coull strives to be a transparent and trusted company to work with, which is why we’ve put a huge amount of time, effort and investment into eradicating invalid traffic from our platform. For example, we have added features to our publisher dashboard to enable our publisher partners to see when we detect any invalid traffic coming from them and our compliance team stamp it out.

Team work

We don’t have to tackle this crime alone, using the best third party vendors to verify traffic is much more effective. At Coull, we work with MRC accredited 3rd party verification tools to track all inventory and act accordingly.

Also, we have our very own fraud detecting hero, Nicola, Coull’s compliance manager. Every day, Nicola manually scans traffic and domains to cut out the pesky bots and *inappropriate* websites. This enables us to have a multi-level process targeting ad fraud and eliminating it from our platform.

We have introduced our Traffic Quality Assurance program to help publisher partners reach the quality required to partner with Coull and to help our advertisers buy media with confidence.

No double standards

Much like the wild west, online advertising doesn’t have many set rules, turning the internet into a western shootout – fraudulent traffic coming from every direction. One way to stop fraud is by measuring genuine ad impressions and true viewability.

Although industry bodies like the International Advertising Bureau (IAB) and the Trustworthy Accountability Group (TAG) are setting guidelines, not everyone follows them. There are a huge number of ad tech vendors running their own measurement of these standards. This means each vendor’s results will be different, affecting expected CPMs, creating a lack of expected inventory and mistrust.

Coull has been working on pre-bid viewability technology. We can detect where the ad unit is on the page before it’s served, enabling advertisers to decide what inventory to purchase based on whether their ad would likely be in view. The biggest advantage is that this minimizes wasted ad spend, giving demand partners real-time data to help them make the best buying decision.

Educate

Coull’s queen of compliance, Nicola says, “Educating people about the different types of fraud is one of the most important things at the moment. Unfortunately, law enforcement is still behind on tackling ad fraud, so we need to learn how to defend ourselves.” So that means, helping publishers recognise any fraudulent traffic and the different forms it comes in. Also, helping buyers achieve efficient and valuable return on campaigns by evading traps and not buying blind.

Direct partnerships

With the hundreds of partners, networks and exchanges out there it makes it easier for fraud to creep in. Whereas, having direct partnerships can eliminate the risk. According to Integral Ad Science, nearly 9% of digital ads delivered via programmatic channels are fraudulent, compared with only 2% of ads delivered through direct deals with publishers.

Coull cut out the middlemen by hosting our own exchange, connecting demand partners directly to publishers’ ad servers. And our formats, OverStream and Double:UP are direct publisher integrations, for a simpler, diluted environment.
What now?

Many companies are working on anti-fraud techniques, particularly the buy side. However, this year will hopefully see more supply side and exchange take the lead.

Publishers: Fraudulent activity can compromise your business model and can damage the brand’s reputation. You need to be able to identify the different forms of invalid traffic and be transparent about inventory.

Advertisers: Fake views on your online campaign is wasting money and creates inaccurate data about the ad’s performance. Make sure you know exactly what inventory you’re buying to protect brand image and have a more valuable return on campaigns.

Ad tech suppliers: If fraud is being hosted by your technology, you will be liable for rebates or refunds to your advertisers and their agencies – and may even be removed from media plans. Work on keeping up standards and abiding by guidelines. Also, direct relationships with partners result in more trust and transparency.  

It’s all about teamwork and education to banish the fraud cowboys from the wild wild web, for good.

Posted by Naomi Sandercock in Coull news

The importance of targeting

Targeting in your platform is usually a fairly simple operation – it’s easy to setup and you expect it to work based on the rules you implemented. However…

  • How many of you audit your targeting – do your publishers send you the correct information?

  • How many of you are speaking to your demand side to ensure that your targeting is matching their targeting?

  • Are you seeing misaligned CPM’s and wondering why?

The chances are that your targeting rules are misaligned. Most of the tags that we send out are price matched against either domain lists, player sizes and/or geos. I personally do not see much misalignment against domain lists or geos, but I do when it comes to player sizes. The publisher is expected to send a certain size via a particular tag. And due to the strict targeting we apply on our demand side, you’re probably losing revenue and decreasing your fill rate.

Audit

Recently,  I completed an internal audit across our supply chain – I wanted to know the amount of requests we received, where we were unable to detect/receive the following:

  • page_url

  • player_width

  • player_height

In a 24-hour period we received 180k requests where we were unable to pass required information to our exchange. It’s a small percentage of our overall traffic, but add that to misaligned pricing channels and it starts to add up – especially when you sit in a chain of other ad servers.

At Coull, we’re happy to audit your traffic and let you know where sales are going amiss. Equally, we have another amazing option – it’s called multi-price floor targeting. We are not the first to use it. But it does make sense!

We only need to supply you with one tag:

  • Tell us what CPMs you expect and we’ll handle all the necessary targeting.

  • Use subID’s only to identify sub-publishers rather than price points.

  • We will never undersell your inventory.

  • All targeting criteria is immediately lined up with our demand stack, ensuring the best connections.

We care about the quality of our traffic and continue to develop technologies and communications that improve the transparency of inventory. If you would like your traffic to be audited, please reach out to your account manager. We are more than happy to help!

Posted by simonholliday in Coull comment

RTB 2.5 – new features that affect video advertising

DPS2-880x495.jpg

With the release of the draft spec for RTB 2.5 there are some super interesting new ideas around how to describe video adverts. Here we’ll take you through a few and how we see them being used.

1 video.placement

This addition allows publishers and SSPs to describe the type of placement that an ad is being requested for. We’ve all seen 300×250 ad placements come through, and we all know they are in banner requests – now we have a way of saying that explicitly.

2 Data Encoding

You can now specify a data encoding header that should be handled by the bidder. A good example of this would be specifying gzip encoding of the bid request – this simply compresses the traffic over the wire from exchange to bidder and back, saving on bandwidth and ultimately money.

3 Bid Changes

There are a few changes to the bid object. A bidder now has the ability to provide a Billing Notice URL (burl) and a Loss Notice URL (lurl).

For Coull, this added layer of transparency is something we’ve been passing to bidders for a while now, albeit relying on our own tech to make that possible. It’s an important inclusion as 2017 will be the year brands and agencies demand more clarity about what they’re bidding on and the result of the auction in real time. We allow the bidder to see if they’ve won or lost, and what the winning price was, which only helps optimise the whole process.

The addition of these 2 features introduces a subtle but important change to the data a DSP can get from an auction. The win notification can now be thought of as just that – the price you offered was enough to win the auction but it doesn’t guarantee anything.

The burl is a great addition, as it’s stage further on from the current win notification. The burl will provide a more accurate way of tracking spend based on delivered impressions, as it’s connected to the impression – it’s saying this impression cost $x.  Splitting these 2 things up enables DSPs to track things like failed impressions and possibly partners that may have issues with their player.

The loss notification adds another dimension to this information. It enables the DSP to immediately know that the spend that they had assigned to the auction is now free – there will be no impression. Coull has been offering loss notifications since the introduction of its Exchange and we’re pleased to see this finally make it into the RTB specification

4 Source

The new Source object lets the exchange pass on some data about whether or not there will be a decision made upstream from the exchange. Header Bidding is the obvious example here, but more and more Ad Managers are holding client side auctions to increase the amount of demand an opportunity sees, and it’s normal now to see sideways connections from exchange to exchange, again to increase the amount of demand available in an efficient way.

There are some more changes, little and big, to the spec but I’ll leave it there for now. The above represents what we believe to be the most interesting ideas in the new spec. It’s great to see some positive changes have been made in the is latest update, and again we’ll look for yet for improvements come version 2.6.

Posted by simonholliday in Coull comment

How to identify and avoid invalid traffic fraud

Recent reports from the IAB show that digital ad fraud is second only to the opium and cocaine trade in terms of crime rates and revenue. That’s a pretty devastating stat for anyone working in the ad tech industry, and anyone buying media.

But it’s not end of days, in fact programmatic is lifting its game and those who can’t or won’t comply to providing better will find themselves ousted come 2017. The most important thing we need to do right now is help publishers understand what forms invalid traffic takes and how to recognise it. And to help buyers ensure efficient and valuable return on campaigns by evading traps and buying blind.

There are many players in the programmatic ad world, many pipes connecting many different suppliers to advertisers, agencies and brands. At the end of that supply pipe exists many different forms of fraud. Understanding what types of fraud exist is important, and we are taking steps to educate our partners about exactly what they are.

Here are some of the most common types of invalid traffic fraud and how to catch them out:

Ghost sites

We can prove that ghosts exist because we have an abundance of evidence – here’s how to spot a cyber spook:

Ghost sites are made to resemble real web sites that host a multitude of advertisements but when scrutinised further it’s easy to tell a real site from a false one.

Ghost sites will usually use a standard blog template, meaning that they often look exactly the same as each other and appear to be legitimate sites.. The domain name of a ‘ghost site’ will nearly always describe a contextual category (e.g. food, automotive, fashion) that appeals to advertisers.

Some basic signs of a ghost site are:

  • Ghost sites referral traffic often comes from an unsafe location such as a porn site but will be camouflaged with another url – usually one that doesn’t link to anything because it’s not real.

  • Links within content don’t work

  • Social media links don’t work or direct you to a page that’s pretty much empty

  • Videos within the content will take an age to load because the player is requesting as many ads as it can. Ads can appear behind the page itself, be hidden in the page or start to load  a ridiculous amount – often that you’ll never see.

  • About us and contact pages will have content that is scraped from other sites

  • Try emailing the support or sales team – the email will invariably bounce

  • The page will have incredibly high bounce rates

  • Real, quality site domains will be copied and the same url with a different path such as .tv will be purchased.The site will look very much the same as the .com version but the content won’t change much. Company addresses will be false and you will often find the layout of ghost sites have many similarities or are exactly the same template.

  • Copy is scraped from other sites so if you search for it, you’ll it appearing in other places and will eventually find the source.

Got time for a quick test your ad fraud knowledge activity?

Both these sites are blacklisted by Coull  – can you identify any of the above ghost site tell tales?

Gardengirly.com

http://fashionitch.com/

Domain Fraud

Domain fraud relates to a issue with the actual domain being sent through to us, this will often require a supply partner to block said domain. Domain fraud can come in many different forms, there could be a mismatch between the domain declared and where the ad is actually placed for example.

Domain spoofing can be one of the most difficult to detect as well as prevent and can therefore be the most lucrative form of invalid domains. Those spoofing domains declare inaccurate domains in order to make advertisers believe that invalid, or low quality domains, are reputable and often highly sort after.

Watch-Movies-Online.cc → Changed to show → usatoday.com

IVT (Invalid Traffic)

Invalid traffic relates to the traffic running through a domain being undesirable, this does not mean the domain itself is bad and therefore would not require blocking. However the source of the traffic needs to be found and removed. Here are types of invalid traffic:

Proxy Traffic

A proxy allows access to the internet anonymously and can browse the internet without leaving any kind of footprint. This means all ad requests will go through a proxy, so for those monitoring the requests, the only thing visible is the proxy – there’s no way to know who’s behind it.

No proxy: User → Domain

Proxy used: User → Proxy → Domain

People using proxies are usually trying to hide themselves – although there can be a few legitimate reasons to use proxies they are more often used to hide malicious activity.

IP Reputation

IP reputation means that the detected IP has historically been shown to be high risk due to being associated with characteristics of fraudulent activity.

Automated Traffic

Automated traffic refers to malicious bots or non-human traffic, designed to generate false ad impressions, or serve hidden, unseen ads, all while avoiding detection. Often they take over a user’s computer, running in the background whilst the user is unaware of the issue.

Spoofing

Spoofing is a practise where a user’s browser and/or device are manipulated to resemble a different browser and/or device. The malicious party impersonates another device (or user) on a network in order to show ad requests from more reputable sources. This is often used to simulate traffic from multiple visitors and therefore increases the amount of potential impressions running through an exchange.

Transparency is key to fighting cyber crime

We’ve put a huge amount of time, effort and investment into eradicating invalid traffic from our platform and have added new features to our publisher dashboard to enable our publisher partners to see when we detect any invalid traffic coming from them. We alert them to any problems and help them to clean it up so the value of their inventory and relationships with agencies, brand and advertisers is always credible, and buyers transact with confidence.

Having an ethos of transparency is key to fighting cyber crime in all its formats and we believe in providing unique creative offered at programmatic scale, that’s trusted and efficient.

We have introduced our Traffic Quality Assurance program to help publisher partners reach the quality required to partner with Coull and to help our advertisers buy media with confidence, making the most efficient use of their budgets. We’ve developed a stringent program in line with the MRC’s Invalid Traffic Detection and Filtration Guidelines and we detect and block against (GIVT) and (SIVT) as defined  by the IAB, MMA and MRC.

Our platform gives publishers the opportunity to see when they are sending us invalid traffic and work with us to ensure it stops. This strategy has been enforced to ensure we don’t allow invalid media to exist in our platform, helping our publishers increase the value of their inventory and providing a trusted market for brands and agencies to buy valid opportunity to reach their audience.

To find out more about the benefits of working with the Coull Platform, get in touch or visit our OverSteam page to see our video as formats in action.

Posted by simonholliday in Coull comment

Stop forcing your ad agency to buy fraud

Agencies and marketers are taking the reins of their ad spending and campaign monitoring, but there is still a lot of confusion over what KPIs best capture how well a digital campaign has performed.

But does it really matter?

The short answer, yes.

Marketers’ choice of KPI’s can actually be having a massive impact on the overall success of a digital campaign.

Often the metrics and benchmarks marketers set can indirectly encourage ad buyers to turn to fraudulent options as a means of hitting unrealistic targets. Marketers and Advertisers soon find themselves trapped in the number-chasing cycle.

So, how can we break the cycle?

Mastering the Metrics: Viewability

Marketers and Advertisers need to understand ad metrics and read between the numbers. As Digiday reports, GroupM doesn’t care about the time spent on an ad, but rather whether an ad is 100% in view. Metrics such as viewability rate, fill rate and impression rate can offer detailed information on this, but aiming for an unrealistically high % can pose an array of risks:

  • Pushing for a large amount of impressions may compromise the ad’s placement online.

  • Ad buyers may turn to ad servers and/or players that partake in malpractice such as:

    • counting an impression before the ad has loaded

    • rotating multiple ads in a player to count multiple impressions with no ad being properly served

    • placing an ad on a range of fraudulent websites.

  • Ad buyers may be encouraged to ‘cookie bomb’ a page, by dropping lots of cookies onto a page with the idea that at least a small percentage will convert.

  • Highly viewable traffic is not always highly viewed by humans. Unusually high viewability rates can be an indicator of high fraudulent activity.

Transparent Reporting

Marketers, advertisers & publishers need access to impartial campaign performance  measurement.

Here at Coull we recognise the industry’s need for a third party fraud-free “enforcer”. By using an array of fraud detection tools, such as Forensic, the IAB’s approved cybersecurity service and manual vetting, we are able detect the most sophisticated patterns of fraudulent behaviour & work closely with both the publishers and advertisers to relay this information back and (most importantly) take action across our marketplace.

The duopoly, Google and Facebook, have often been accused of having a much less objective approach to reports due to the fact that some of their ad metrics have yet to be verified by third-parties. The UK Business Insider has likened this to “marking their own homework”.

Realistic Targets

Marketers & Advertisers need to be realistic with their KPI benchmarks. As Venture Beat reports, new video formats such as click-to-play pre-roll, native and out-stream, and video within social content are much more fraud-resistant. However, “ad-buyers continued to expect the high  completion rates the overall industry had been promising”.

       

Many have suggested that the need for a new reporting metric is the real answer to the dilemma. But the reality is, numbers will always be subject to potential manipulation. The responsibility lies with the marketers to tackle digital ad metrics head on, without turning a blind eye to the potential risks. Numbers on fill, viewability and completion on high quality fraud free websites are undoubtedly going to be lower than those generated in the realm of bots, dodgy players and fake websites.

Ultimately, elevated completion rates like those projected by both Facebook and Google in recent months are simply unrealistic for in-feed ads. Both marketers and advertisers need to agree that a potential dip in the graph is a small price to pay for real human views on high quality sites.

Posted by simonholliday in Coull comment

2015 predictions for the Adtech industry

The adtech industry has come a long way this year, with the ‘death of display ads’ a drunken rambling we’d rather take back. Albeit rebranded as ‘native’, display advertising is thriving again – banner ads are everywhere, from your news feeds to your search results, 2014 has seen a resurrection of the format.

Another notable change has been the increase in ad spend through programmatic, with an estimated 137% growth this year (eMarketer) and investment into video advertising at an estimated 59% increase from last year (IAB).

So with that said, it’s been a positive year for us. But what does next year hold? Here’s what I believe the digital advertising industry need to think about for a successful 2015.

  1. Mobile is just another form of digital

2015 will be characterised by the continual increase in, and shifting of, advertising dollars from more traditional media to online. As part of this, the lines between mobile and desktop will continue to blur and marketers will need to view the mobile web for what it is – just another form of digital.

  1. We need more transparency around data

And I don’t say ‘need’ lightly. As the industry consolidates, adtech companies will need to be more honest about where money is being spent and whether it actually delivers value back to advertisers. Those that don’t will get found out.

The industry isn’t doing a great job of policing the use of data, so I think there’s a high possibility of government intervention and regulation within the next year or two as a result. I also strongly suspect that Google will adopt a ‘no-cookie’ policy within the next 18 months, and there will be a lot of pressure on the government to regulate.

  1. Innovation will come from the publisher space

At the moment, I don’t really see a huge amount of innovation among adtech companies. In 2015 I think we’ll see this innovation coming from the publisher space, but not from traditional media companies. We’ve already seen disruptive publishers and distributors such as NDN and Perform that are able to adapt quickly and create new revenue streams through their use of programmatic. These guys are already competing in the top 10 of comScore having come from virtually nowhere.

  1. We need to take the reigns on programmatic

I’m sick of attempts to define programmatic. We’re doing a dreadful job as an industry trying to explain what it is and it’s not helping the education process for media brands. It doesn’t help that there is a lack of best practice case studies out there – publishers should be more comfortable in explaining how they are using these models.

Industry bodies such as the IAB and AOP also need to do more to provide leadership and education within the industry. At the moment we have a situation where programmatic – and it’s definition – is being led by tech vendors rather than the industry as a whole, and that’s skewing the picture in favor of those companies.

  1. Viewability is a dealbreaker

Adtech companies that can use viewability as a metric and therefore provide real measurement and value, will be the ones that are ultimately successful in 2015. As the market matures and the industry becomes more transparent, we will be begin to see the companies that don’t add any real value begin to fail. Their shortcomings will be fatally exposed.

  1. Demand for talent increases

Programmatic has risen so rapidly that it’s difficult to find people with the ready-made skillsets necessary. The biggest challenge we will have, shared by other companies in the industry, is the recruitment of suitable talent. The types of expertise needed for programmatic advertising are completely different to those of traditional advertising. We’re in the market for data scientists and mathematicians rather than copywriters, and there’s a huge battle for that type of talent. We need to convince people with a degree in maths, for example, that there are real career opportunities for them in advertising.

I don’t have a magic ball and I can’t give you any groundbreaking predictions; I think we simply need to do what we’re already doing, but better. A lot better.

Posted by simonholliday in Coull comment