Personalization is a dying artform, the marketers say.
With the rapid advanced in deep learning, machine learning, and AI (artificial intelligence) powered technology, businesses should be leveraging every source they have available to drive sales, marketers say.
Enter the new age of hyper personalization.
It's the "future," they say, but what exactly is hyper personalization? How does it affect user privacy? How is it connected to the filter bubble problem?
There isn't a universal, agreed upon definition for "hyper personalization."
However, there are recognizable characteristics of hyper personalization. We can identify these characteristics instead, as they'll paint a good picture of what hyper personalization is.
Key recognizable characteristics of hyper personalization include:
- the use of machine learning or AI to generate and deliver highly "personalized" content
- the use of real-time data
- combining past behavior with present behavior to predict future behavior (i.e. recommendations for products)
Based off the above, we can now reasonably define hyper personalization as:
the use of an AI or Artificial Intelligence and real-time data to generate and deliver personalized content, products, and service information to its users.
Hyper personalization vs Personalization
At a marketing level, personalization is combining personal information to a communication.
An example of personalization would be when a company sends you an email after you just purchased a pair of shoes and the email includes your name and the exact item that you purchased.
Another example would be a shopping app greeting you with a user-defined name (such as a nickname or your first name) upon opening the app. "Welcome back to our store, avoidthehack!"
The key distinction between hyper personalization and personalization is:
- the use of behavioral data (past and present)
- the use of some type of machine learning (it will analyze given data and predict the most likely future behavior.)
For example, a store's deep learning algorithm sending you a text at 5pm on a Tuesday to buy that new Dell monitor you have been searching for, for the past 2 weeks.
This is after:
- You've searched for monitors on the website no earlier than 5pm but no later than 6pm
- You've clicked on Dell monitors more than any other brand, or have a prior history of buying Dell monitors
- You've visited the website/stayed on the website most often on a Tuesday
- You have responded/engaged with texts from this store in the past
The deep learning algorithm combines this information. That text it sent you is its best prediction to actually get you to buy this Dell monitor.
The real question to be asking is what data doesn't hyper personalized marketing collect (or attempt to collect).
Ultimately, they want to know exactly who you are, your habits, your interests, what other devices and apps you use, your location and what locations you frequent, how often you do something, how often you buy something, and... see where we are getting at?
Don't believe it?
Hyper personalized approaches can collect and analyze all sorts of information, such as:
- Data from the sites you visit on your devices (desktop, smartphones, tablets)
- Your current precise geo-location, often within 1-mile accuracy
- Your past location history (ex: ever visited a coffee shop, and then had many ads for coffee on your phone?)
- Information on the network you're currently connected to (are you on your home Wi-Fi or using cell data?)
- User agent string of the device you're using - web browser, operating system, exact software version
- How long you spend on a webpage, where your mouse tracked on a webpage
- Cookies, whether 1st part or 3rd party, stored on your device
- PII (personal identifiable information) such as age, gender, home address
Seems far-fetched and you still don't believe it?
Well, let me use Amazon as an example.
You know Amazon. It's a household name, the e-commerce leader, and all that jazz.
But did you know that alongside the likes of Netflix, they're also a giant in the hyper personalized marketing arena?
This is mostly due to their AI powered recommendation system, DSSTNE. (As of 2016, DSSTNE's code base is open-source, so good for them I guess.)
In fact, it's estimated that their hyper personalization powerhouse, DSSTNE, accounts for nearly 35% of their sales.
The real question is, what data does Amazon collect that may or may not feed DSSTNE?
- Your payment Information
- PII such as your name, address, phone number, and age
- Your location
- your IP address
- Voice recordings (Alexa)
- Images and videos when using Amazon Services
- Content interaction information (downloads, streams, playback details, and your network details when streaming content)
- Your device metrics (device usage, app usage, connectivity data)
- Your product/content search history on the website
- Your wish list on the site
- Web browsing history
This list has been abbreviated. You can view the list in its entirety by reading Amazon's privacy notice.
Now to be fair, some of the information Amazon collects is totally understandable. For example, they need valid credit card information to process payments and a valid address to ship the product(s) to.
However, why do they "need" information about the other devices on my network? Why do they need ridiculous access privileges on my smartphone across all their services? Why are they (presumedly) storing and sharing voice recording data from the likes of the smarthome speaker, Alexa?
How often are they analyzing, updating, and sharing these massive amounts of data they regularly collect?
Even with the pages long privacy policies, there are still a lot of questions left unanswered for us end-users.
In this new age of hyper personalized content and services, we are still left in the dark about how and when our data is used to give us these hyper personalized experiences.
Where does the line get drawn for data collection, storage, and sharing when it comes to both personalization and this newer trend of hyper personalization?
So, as you can clearly see, hyper personalization (or "hyper personalized marketing") aims to collect as much data (information from you) as possible.
At the end of the day, our marketing overlords want to harvest as much information about us as possible in order to feed their massive AI machines, which put shiny stuff in front of us to spend our money on.
It's important to understand that we don't believe there is inherent wrongness associated with machine learning or the use of artificial intelligence.
In fact, we try to be optimistic; we like to believe that there is a lot of good to be done with such powerful technology, especially as it continues to develop.
However, we do believe that the current climate of "hyper personalization" leaves much to be desired.
This current climate actively supports the blatant and habitual abuse of user privacy.
You see, the more information these companies can collect about you, then the more information they must feed these AI engines.
More information generally means that what the AI engines spit out make it far more likely for you to consume spoon-fed content, or spend money.
(In many cases, this cycle goes way too far and you run into the filter bubble problem.)
So, what are these companies going to do?
They're going to spend a ton of resources trying to collect whatever information they believe they "need" or want.
Enter the relentless ad trackers, intrusive fingerprinting practices, excessive telemetry, creation of shadow profiles, and many more tactics we frequently see today.
And their efforts to collect information are dubious at best, and outright abusive (often illegally so) at worst.
This begs the question: how do you protect yourself?
The short answer is, it's hard. Very hard.
There is little you can do you combat directly against the harvesting of data that often accompanies a hyper personalized marketing approach.
However, there are many things you can do that indirectly protects your personal data and privacy.
Some of these things include:
- Using a privacy focused browser with privacy preserving add-ons.
- Addressing all critical leaks, such as WebRTC leaks which can reveal your private IP address
- Sign out of web services or social media when you're not using them. This helps prevent easy cross-linking across different platforms.
- Blocking 3rd party cookies and ad trackers
- Limit use of services that engage in aggressive data collection + hyper personalization, if possible
- Limit what information (outside of the true essentials of a service) you knowingly and voluntarily give to companies
Again, I want to emphasize that the algorithms and the AI isn't necessarily the enemy here.
They aren't necessarily committing the user privacy abuses that have become commonplace on today's Internet.
It is important to realize that the issue with hyper personalization lies with the harvesting (both knowingly and unknowingly) of user information, the sharing of this sensitive information with third parties, and the lack of control such methods of information collection breeds for the end user.
Even with all the privacy policies, cookies policies, or what have you, it's still an informed consent issue.
Hyper personalization can be a neat marketing tool if the big players would play fair, and readily informed us about how they truly use what information goes into these deep learning algorithms.
So, with that said, as always, stay safe out there!