Frequently Asked Questions about FouAnalytics

There has been a huge surge in marketers asking me for help to audit digital campaigns since 2021. They've long suspected something was wrong, but could not put their finger on it, given the tools and vendors they were using at the time. Over the years, I've been asked questions like the following. A lot. So I thought I'd gather them in one place for you, in case you had the same questions. If I missed any, be sure to message me and I will address.

I already have analytics on the site, why do I need FouAnalytics?

Google Analytics and Adobe Analytics are great tools for measuring site traffic. But they don't report on bots and other invalid traffic. FouAnalytics was built to detect bots and is used to troubleshoot discrepancies observed in Google and Adobe Analytics. Further, did you know that fraudsters can write false data directly into Google Analytics, to make it appear there is traffic, when there is no traffic at all? They can also easily make the traffic appear to be organic, direct, social, paid campaigns, etc. By having FouAnalytics on the site, you can detect the phantom traffic written into GA, because security mechanisms prevent the criminals from writing false data into FouAnalytics. If you see traffic in GA, but not in FouAnalytics, then that traffic didn't actually occur.

See: How Site-Owners Use FouAnalytics to Troubleshoot Bot Traffic

I already use fraud detection, why do I need FouAnalytics?

Yes many of the largest advertisers have double-paid for fraud detection for years, but it hasn't reduced fraud at all for them. In other words, they pay for it from a budget line item called "verification fees." They pay for it again, unknowingly, because the DSP, the exchanges, and perhaps even their agencies, hide the verification fees in the media CPM. So $2.00 CPM inventory is quietly sold at $2.15 CPM and the 15 cent CPM for verification is hidden in the media cost, a different line item in the advertisers' budgets.

Fraud verification has failed on many levels, and consistently report around 1% IVT. They often don't tell you which sites or apps caused the IVT, so you can't take action yourself (like add sites and apps to block list). They also don't explain why something was marked as IVT, or why something was not marked as IVT. One of the reasons they fail to catch so much of the fraud is that bots are clever and block their detection tags. They have no data with which to mark the bot as IVT. And they happily let you assume that "not IVT" means no fraud. Furthermore, they fail to even look for many other forms of fraud, and fail to detect even bots that mock them and tell them they are bots -- like "fartbot." FouAnalytics gives you a way to check if your fraud vendor detected all of the fraud for you (so you can see just how much they missed).

See: How to use FouAnalytics to Scrutinize Programmatic Campaigns

Can FouAnalytics measure Facebook, Google Search, YouTube, Tiktok?

FouAnalytics tags, like most third party tags, are not allowed to measure the ads directly, on Facebook/Instagram, Google search ads, YouTube ads, and Tiktok ads. So we place the code on the landing pages/sites where the clicks go. That way, we can measure whether the clicks are from bots or humans, and more importantly see whether those users did anything further on the site. Google Analytics may have given you a clue previously, when you see clicks arriving from various paid media channels exhibiting high bounce rates, low time on site, or single pages per visit. Bots typically don't stick around and waste time on your site; they quickly leave so they can do the next fraud thing. The problem is that your legacy analytics doesn't tell you what bots they were or where they came from. With FouAnalytics on the site, you can see both the source of the bots as well as characteristics of the bots themselves -- e.g. which data centers they came from -- and take appropriate action.

See: Digital Campaigns Measured by FouAnalytics

How is FouAnalytics different?

FouAnalytics is similar to Google Analytics ("GA") and Adobe Analytics in that it uses a javascript tag to do the measurement. Virtually all of programmatic advertising involves javascript calling ads, so a javascript tag can be used to measure traffic on a site and also what caused an ad to load -- i.e. bot or human. You can think of it this way -- GA is analytics for your website; FouAnalytics is analytics for your digital media, and your site too. Previously you used fraud verification for your digital media, but that tech was very limited in what fraud it could detect, that tech made lots of mistakes in measurement, and it was all black box, so you could never get straight answers about how they got some % IVT. Adding FouAnalytics tags to your digital media means you can upgrade your tools, so you can see and understand why something is marked as fraud or why something is not marked as fraud.

Do you know any of the engineers who work at the verification tech companies, and do you know when was the last time they updated their detection algorithm? Right. I don't know either. I have been personally tuning the algorithm for the last 10 years, and you can ask me anything and challenge me on the measurements in FouAnalytics. The goal is to better understand it together and troubleshoot campaigns. I am happy to explain to you with all of the available supporting data; and correct my measurements if we find that they are wrong.

See: Cybersecurity Measures Built Into FouAnalytics

Is FouAnalytics MRC accredited?

No. The Media Ratings Council (MRC) sets standards for the digital marketing industry. They also offer "accreditation" to vendors that conform to these standards by showing that they measure what they said they would measure. For example, vendor A submits a spreadsheet with the 3 things they measure; they get interviewed by an accountant from Ernst & Young. If they show they measure those 3 things, they get accredited. If vendor B submits a spreadsheet that said they measure 500 things, and prove that, they also get accreditation. Vendor A and Vendor B's measurements will undoubtedly come out different, even if they measured the same campaign. That is why we have seen discrepancies between two different vendors, both accredited by the MRC, and measuring the same campaign or same site.

To be more specific MRC has no tech of their own to measure campaigns; and they have no "answer key" to know if bots are measured correctly or not. But FouAnalytics is the tech platform that I built to help me audit campaigns for clients. I have looked at data almost daily for the last ten years, and I have "truth sets" of both bots and humans to continuously test and refine the detection algorithm underlying FouAnalytics. I have also measured campaigns and sites against practically every fraud verification vendor over the years and have specific data on what they catch and what they miss. I can't show you that data; but if you add FouAnalytics to your site or digital media, I can show you your own data.

See: IVT Vendors Fail to Meet Minimum Standards for MRC Accreditation (but yet they remain accredited)

If bots are so advanced, why can't they trick FouAnalytics as well?

They can trick FouAnalytics. I always assume I am missing something or my detections are wrong. So I continue to analyze the data and update the algorithm based on what I find. I wrote this article about how bots have evolved over the years. Most bots are able to fake most things correctly; so detection tech that relies on user agent or IP address will not catch most of the bots. While they can easily fake all the javascript parameters correctly, it is actually much harder to fake timings correctly. First of all, the won't know what the "right answer" is, so they have no "goal posts" to shoot for. To be more specific, FouAnalytics asks the browser to complete computations, counts, sorts, and renderings; we time how long it takes for these computations to be completed -- i.e. timings. If a bot lied and claimed it was an Android smartphone but the computations are completed way way to fast (like a server) we catch them. This is called "entropy analysis" and FouAnalytics is in the third generation, specifically because I knew years ago that detecting simple javascript parameters will no longer be sufficient to catch bots. Even detections for automation like puppeteer, webdriver, selenium, headless chrome stealth, etc. are no longer enough.

See: Why It’s Hard for Bots To Avoid FouAnalytics Detection

How do you charge for FouAnalytics?

FouAnalytics remains free for small and medium businesses (under 10 million pageviews per month on-site, and under 100 million ad impressions for in-ad measurement). I also don't charge agencies and consultants that use it to help clients. If there's a large advertiser that wants to use FouAnalytics to monitor all their digital media and/or their site, they pay an annual subscription, much like Google Analytics Enterprise, Adobe Analytics, or Microsoft Office. The annual amount is based on their media plan and is billed up front, so we don't need to settle up to actuals every month (which is a lot of unnecessary work). Even if they go over the counts by 10%, we don't charge them extra. By using a fixed annual subscription fee, we avoid the misaligned incentive of trying to get the advertiser to buy more ads when it is not necessary. Further, I don't charge a percentage of the fraud found, to avoid the misaligned incentive of finding more fraud than there is. This is analytics, and it provides clients better tools to monitor and manage their own campaigns.

I show clients how to use FouAnalytics and what they can do with it themselves. FouAnalytics is analytics for digital media, and that sets it apart from black box fraud verification tech which doesn't provide enough details for clients to understand whether the measurements were correct or not.

See: Tour of FouAnalytics Dashboard and Campaign Checklist

Please let me know if you have any questions not covered above.