You’ve probably experienced it yourself (since you were curious to read this article) – you were just thinking about something a few moments ago and later while browsing the web or using apps, you saw related ads even though you strongly remember not searching or having clicked on anything that might have tipped Google/Facebook off.
No one can LITERALLY read our minds unless you’re hooked up to a lie detector test, which also has been proven can be fooled. But, what you might not know about is that the extent to which these companies collect and process data. In the Advertising ecosystem, there are many participants that enable the entire process and make it ruthlessly efficient as well. Refer the image below for an overview.
Still confused? Here’s how it works – A publisher (i.e. blogs, media websites/platforms) signs up with Supply Side Networks/SSPs, who hold and auction off ad inventory, which is ad space from all the publishers on their platform to Advertisers who operate through DSPs (Demand Side Platforms) via auctions that take place on the Ad Exchanges.
There are other intermediaries in the mix like data management platforms, which collect user data and with the Ad inventory available with SSPs, help DSPs match the advertisers’ requirements with the target audience on specific publishing platforms.
Ever notice a slight delay in Advertisement showing up after the webpage’s loaded? That’s because, the entire process of auctioning, starting from matching the advertiser to publisher as per the advertisers’ specifications, deciding the price to be paid to the publisher for Ad impressions served, to fetching the Ad content from the Ad Server, happens in about 10 milliseconds, starting from when you clicked on a link to load the webpage.
Remember the mention of Data Management Platforms? There’s a high probability that the companies in the image above hold gigabytes worth of data on you. This data is captured from a lot of sources. Imagine all the websites and apps you have ever used and then combine all of them together – you might get an idea.
Now, this data itself doesn’t make much sense. Companies like Google and Facebook have the best engineers working to build computer programs that process this data, or ‘Algorithms’ on a massive scale and makes predictions and decisions for millions of users in a matter of seconds.
Algorithms are used to find patterns in your data, which then provides these platforms with insights on possible user behavior, traits, etc. This, in turn, is used to target advertising content, based on these insights.
One would ask, How accurate could this get? Considering the amount of data out there and how quickly Machine Learning and Artificial Intelligence systems are getting (example, Google’s AlphaGo), pretty accurate.
Perhaps, it’s safe to say that the line between coincidence and an accurate prediction is pretty thick. It should be far from a surprise when something closely relevant shows up on your screen unannounced.
If we were to explain it using simple math, think x + y + 12 = 22. It isn’t really difficult to complete the equation once you have either x or y (the real thing has way more complexity involved).
This is not something that only big companies do. Machine Learning-enhanced tools and services aimed to retarget customers are now available for smaller companies as well, with the entry barriers in terms of pricing set lower (example,
Does this mean that giants like Google and Facebook themselves don’t have much data on you? Far from it – Facebook’s Cambridge Analytica scandal should be enough of a demonstration of how influential their platform is.
Regarding Google, we came across something much more intense in the form of an experiment on
On mentioning the word ‘Dog toys’, he started seeing Advertisements that suggested products that matched the exact keywords or were related, on multiple websites on Google display network.
Do we have a reason to worry? Yes and No.
Yes, because events such as the Cambridge Analytica scandal, where the data fell into wrong hands, might happen again. There probably are more of such bad entities out there trying to get their hands on this data for their own agendas.
Stricter policy enforcement is in place with the Government intervention and public demand, but given how complex the entire system is, it’ll only take so long for someone to exploit a loophole once found. Companies will always collect data to improve their products and the user experience, and its part of their business model.
There are benefits to it that go unnoticed – like the content recommendation systems that are dominantly used on platforms like Netflix and Youtube. Youtube’s suggested videos feature is as powerful as it gets. With 4000 hours worth of content uploaded every minute, the platform suggesting videos which are highly probable to be watched next, from different content creators and not just the ones the user subscribed, is incredible.
Almost disturbingly so, with the rapid growth in network access speeds and smartphone adoption, the number of hours spent by users on Youtube and Netflix have gone up exponentially and have become a cultural phenomenon.
It has caused companies to pivot from traditional media which started falling in popularity really quickly since new content platforms started growing. The end user does benefit the most out of this growth and evolution. So, it’s only fair to let go of the occasional misses along the way.
(Disclaimer: This is a guest post submitted on
Nikunj Thakkar is the Founder and CEO of DataOne Innovation Labs. DataOne is a Business Solutions company with a strong focus on Business Intelligence; Big Data Analytics; Data Processing and Management; Product Development; Machine Learning and Natural Language Processing; and IoT Implementations. Reach out to him at firstname.lastname@example.org, Facebook , LinkedIn or Twitter.