Black Box Modeling
The highly unpredictable nature of user behavior makes it really a very tricky case for classification & prediction which is why a black box model could more often than not be a Godsend.
A Product Manager’s way of life!
How do I improve engagement on my platform / mobile app?
How do I drive up revenues over a specific customer segment?
How do I stop users from Churning MoM?
How do I gun for better retention rates over my product?
How do I improve the UX on the app?
How do I improve the overall product?
It is highly unlikely that you are a product person and have never come in contact with any of these questions over your entire stint. That’s totally impossible as these happen to be some of the most common challenges that product people usually face, in fact we are usually juggling these questions / situations one at a time or even worse - all at once!
Double down and analyze each of those questions above and you’d see how it’s all totally correlative to the users. And when one is referring to the users it is essentially their behaviors that become the topic of discussion.
The real challenge
Almost all experienced product people across the world would swear by how this one thing is near impossible to master sans spending years getting seasoned over dealing with various situations gaining & learning from experiences continuously, and that is “gauging user behavior accurately” and to be able to get there over the first iteration / silo itself.
Here’s a quote by Thomas G Stemberg, an Harvard grad & a philanthropist who is more commonly known for his invention of Staples, the office supplies retail superstores.
“Indeed, understanding customer / user behavior needs to be put at the helm, positioning & branding could really wait until after…”
Discussing this case study over my 1:1 sessions with a new mentee I was befuddled when I recited this very quote and it was met with something really cold on the lines of:
- Staples?
- Really?
- Why does a mere office supplies business got to worry so much about customer behavior?
Knowing your customer / users ought to be a given irrespective of what business you’re running, if you’re worried about maintaining a brand image in which case it ought to translate to you thinking seriously about offering a great user experience be it in-store / online.
You really got to hand it to Staples the way they choose to do it over their App which could essentially boil down to these steps:
Step 1: collect enough detail about your customer who signed up
Step 2: use the knowledge of the world to establish common patterns
Step 3: use that understanding estimate & make some really insightful predictions
Step 4: front the users with the products of relevance without entirely disturbing his workflow / web footprint too much
Step 5: Bingo! Make it all look like magic! You have a super-pleased user & you may have increased the chances of retention by 75% just there
But, its true that over some businesses or products, estimating / understanding / predicting user behavior could be a real tough nut to crack.
User’s behavior – A Black box!
User’s wants are really complex and in most cases, may not be known to themselves. To decipher what’s in it and tap into the user’s mind to gauge those wants and needs is the toughest part of the entire exercise. And, that’s why the perception of a black box seems to fit in perfectly well in this context.
“The reference to the black box comes from the parlance of how it is representative of something that could have a reputation of being mysterious, unclear / murky, also owing to how there’s very little knowledge of it in spite of all the advancements around.”
Let’s delve a level deeper & try to decipher this “black box” quantifying & generalizing it as much as possible over an exercise that’s “Black Box Modeling”.
Q: Why do people adopt a few products into their lives?
A: It’s because they want it, they need it. So, it’s about them users trying to satiate their wants and needs.
It’s ok to quote that answer if you don’t have anything at all to do with products / product management per se. But if you do, that is a disastrous answer as it fails to touch upon the psychology of the buyer persona / user behavior.
Wants and needs are totally different!!
- “A Need is representative of things that are absolutely quintessential for living”
- “A Want is representative of things that make life better”
At a broad level, there could be 2 things here that most users consider leading into that Adopt / Reject decision. There are characteristics & idiosyncrasies of the product in question & the process that is used to evaluate them.
Let’s go over each of them.
1. Characteristics
2. Evaluation
And this here is the black box representation as to how all these characteristics come together contributing firstly to a constant evaluation process that may as well take a few days / weeks in some cases and then ultimately triggering the decision of PURCHASE (or) NOT.
Note 1: This is a toned-down version of the complexity of a black box that’s the user’s behavior. And it has been toned down so that one gets a proper understanding of it’s complexity.
Note 2: Talking of building a mathematical model around representing this very complexity of the black box makes a strong case for the use of AI - Fuzzy Logic.
Please remember:
There’s a lot that happens behind that all important decision of a “Purchase” or “Pass (not to purchase)” and just as much as it is crucial for the orgs. building those products, so it is for the users as well.
The black box modeling aims to induce a method to that madness & tame that complexity down to an extent so that the depth of it could be perceived well by everyone - engineers to CXOs / Cofounders alike.