“The internet is not equal (but not in the way you think)”
TDLR / Summary: There are mixed opinions about whether the internet is becoming more varied and different, or more same and equal. The conclusion is that it has become more personalized.
Your “algorithm” and you:
There’s a reason why you’re here right now. It may be that you decided that between your morning coffees and meetings, you would just take a quick peek at your phone. Now, you’ve ended up here. We may think a lot about the different transit paths we take (Travelling from one end of the city to another) but we don’t often think about, or even understand, our every day digital commute. The internet, as we know today, is built on a sophisticated infrastructure of ranking, sorting, and recommending. Recommendation systems, in addition to making information more accessible, and connecting people and resources, have also induced psychosis, echo chambers, and filter bubbles. Recommended content makes up to 80% of all content consumed on Netflix, and, in short form video websites with higher feedback loops (shorter videos + higher screen time = more feedback signals), recommendations have become increasingly powerful. Notwithstanding these impacts, personal content recommendations are a great mirror for self understanding.
The Idea: your “Algorithm”, or what you are personally recommended on a day-to-day basis, is a new, and remarkably telling, frontier to be explored for studies of identity, and consumption.
Consumer researchers have long argued that identity is constructed through consumption. Belk’s 1988 extended self thesis (“you are what you own”) and adjacent identity-based frameworks have shaped decades of marketing thought.
The content that is recommended to you may not only reveal your past behaviour, but, through qualitatively rich content, can reveal who you are. Our early research has shown that while we personally don’t believe that recommended content is representative of our identity, we will go to great lengths to protect our recommendations, viewing it as very private to us. Recommender systems research demonstrates a measurable relationship between recommended content and consumption diversity, and willingness to pay. The 2015 cover story of algorithms predicting and recommending diapers to a woman before she knew she was pregnant, has now become a primitive example of the power of recommender systems today.
The Splintering of the Internet:
For the West, the COVID-19 pandemic was the best experience of this. Your aunt’s internet is not the same as yours, and we don’t all watch the Simpsons at 6pm on Tuesdays anymore. The “Splinternet” theory, by MIT, suggests that a variety of forces, such as local governments play an influence into the fragmentation of the internet. Media-use fragmentation or “splintering” has been discussed for years both in policy and measurement technology circles, and the consensus is that we need better tools to track, and measure, and understand that which people are consuming, and how they are relating to the media (uses and gratifications theory).
If this continues, how do we understand each other?
There have been instances where you see a friend’s social media feed, and may either remark or — by the differences to yours. Fragmentation is not only increasing, it’s increasing personalization. What this means is that the media you are now being shown is more and more representative of your own tastes, preferences, and previous behaviours than ever before. How might we, as well as the average well intentioned organization, be able to better assess what is demanded by the market? How exhausted are we, by Zoboff’s surveillance capitalism, and what are the key requirements to be able to sustainably, ethically, and feasibly understand each other? This has been our primary set of questions since 2021 in developing our patent pending, consumer research technique.
New Ways of Understanding
The rise of the internet has led to a variety of new inventions by market and consumer researchers. For one, co-design practices have become much more common. Digital ethnographies, or “Netnographies” as they are called in industry, study the qualitative meanings created and facilitated by and within online forums such as Reddit. Data analytics platforms assesses a variety of signals, from enterprise, to the individual, across varieties of devices, based on digital behaviour with a variety of learning algorithms, to produce predictive capabilities. Each of these methods contributes something valuable. What we are proposing: builds on these lineages, with a new form of data that we believe in.
What are we proposing?
Our key claim on the market is that if this trend continues, all else held equal, your recommended content is you - or at least the highest resolution behaviuoral proxy of you we will have. It is a rich site for inquiry. It is a structured dataset produced by recommender systems, based on past behaviour. And, provided that privacy and trust requirements are absolutely met, it can be collected and analyzed for insight - both in self discovery and in consumer studies.
Unlike passive scraping or platform-level aggregation, the method is:
- Explicitly consent-based
- Individual-level, item level media recommendations (E.g. that one post you watched this morning)
- Mixed-method (qualitative + quantitative)
- Designed with privacy masking and structured anonymization
Isn’t this just recommended content just a Proxy for real consumer understanding?
Yes, in a way it is. It assumes that day to day recommended content (including ads), is a most representative proxy for your internal states, attitudes, behaviours, and preferences. Of course, it cannot replace existing signals and data points such as consumption data, digital behaviour, and other traditional measures of the individual consumer, such as demographics, or psychographics. It also links existing qualitative methods of data collection. In our feedback with leaders, they have found it to produce the highest qualitative signal, be able to mix into quantitative labels for analysis, for a cost structure similar to that of surveys.
Don’t social media providers have better access to this?
Definitely. As platform owners, they maintain access to the data, infrastructure, and targeting algorithms to serve ads more effectively. However, their business model is primarily on serving digital media buyers, publishers, and end-users, with a revenue model focused on effective ad delivery and measurement. Our primary offering is that to the insights team, strategic planner, and creative advertising team, for the earlier stage development of marketing assets. We believe this data type will find usefulness for the first time ever, producing a holistic understanding of the digital environment to which audience segments of interest spend their time.
Where can I find out more?
.png)
