Personalization to drive incremental growth: How to and how much

We are surrounded by personalization everywhere today. From push-back notifications on different apps communicating deals and offers to being recommended movies and TV shows by Netflix, we are all living a life being guided by machine learning/ personalization algorithms. Therefore, it can be said, without a shred of doubt that the approach to personalization is starting to spread to marketers everywhere and it has a monumental impact to drive business growth.

A very interesting anecdote has been talked about personalization in the book: "The Everything Store: Jeff Bezos and the Age of Amazon ".  In the early 2000s, Amazon had two broad teams. The first team was an editorial team that worked on writing good product descriptions, curating good content, and then recommending it. The second team called P13N that worked on personalization; they took hard customer data, created algorithms to recommend people what they should buy next. Over time, the revenue generated by the personalization team exceeded that of the editorial team, that too driven by just hard-coded logic and algorithms. The algorithm was called Amabot. Now Amazon, which has always been a data-driven company and focused heavily on cost management, sided with Amabot and laid off the majority of the editorial team or reassigned them to other areas. Later an Amazon employee anonymously put out an ad on Valentine’s Day, 2002 in a local Seattle newspaper that said  :

Dearest Amabot,

If You Only Had A Heart To Absorb Our Hatred...

Thanks For Nothing, You Jury-rigged Rust Bucket.

The Gorgeous Messiness Of Flesh And Blood With Prevail!

While the ad was written with the thought that this personalization will soon be obsolete, we don’t need to say it loud how much that thought was flawed. However, the bigger question still lies, how much of a deep dive should a company actually do into its consumer’s life, so that what appears to be a cool thing right now doesn’t turn out to be creepy. In this article, we discuss that how companies use personalization to grow from X to 10X and how much of it should be actually done.

How do e-commerce companies use personalization to achieve growth?

The basic revenue equation is a function of the amount of traffic a platform is getting (measured by the number of visits, the number of sessions) multiplied by the number of people that convert multiplied by the price point at which they were acquired. These are the three broad goals that most e-commerce companies optimize for to drive growth. However, there is a basic checklist that needs to be taken care of while one proceeds towards personalization.

1.   Defining the objective: Traffic/Conversion/Average Order Value/Cross-Sell: The reason to personalize should be guided by a particular business objective in mind.
2. Level of personalization: 1-to-1 user personalization/customer segments: The homepage can be customized for different customer segments according to the life stage of the respective customer: a first-time user vs a repeat customer vs a very loyal customer. At the same time, there can be different widgets on the homepage which are specific to a particular user based on his/her previous actions (search history, products from favorite brand category) It’s always better to go to a micro-level of personalization, but it also comes with greater cost and effort.
3. Basis of personalization: Short- and Long-term affinity/Demographics/Behavior patterns/Lookalikes/ Actions: Personalization can also be done on the basis of affinity of the user (most searched category vs search once), on the basis of demographics (male or female, zip code), on the basis of certain behavior patterns (filters applied before while searching). It can also be done on the basis of lookalikes or actions done by some other user that has ordered or bought the same thing. (Performed the same thing: you might have seen in your Amazon saying, “people who bought this also bought”)
4.  Frequency: Real-time, static, weekly: Lastly, how often should one personalize. It can be done in real-time like a fresh feed being shown every time the user refreshes it, or it can be done back-end by an algorithm based on what’s trending on a weekly or bi-weekly basis.

Where all do companies personalize?

Personalization for most e-commerce can be done either “on app” or “off app”. The avenues for the same are as follows:

On-App:

• App/Web Homepage
• Sort order-product listing page
• Search
• Coupons/Offers
• Product Recommendations

Off- App:

• CRM (Notifications, Email, SMS)
• Digital Marketing

How much of your data is taken to personalize?

In my B-school days, a friend of mine expressed her concerns over the data that is used to create a personalized feed – how much of it actually is derived from the data we leave behind or give with consent and how much of it happens from the data, we don’t actively remember giving my consent to and information is picked from those sources.

While the concern is genuine, usually most e-commerce sites don’t have much information about us. Even when a Facebook or a Google ad campaign is run, neither of them shares user ID data and only performance data (traffic, views, impressions, engagements). Even, for the data inside the app, unless for a payment app like Paytm, where they have your KYC data and they are able to tie that to your socioeconomic data, your age, occupation, extrapolate those and create several segments. Even if e-commerce apps that have a virtual try-on feature, ask for permission to your photos, they don’t use that data. Mostly, it’s guided by user behavior on the platform combined with algorithms that drive the personalization in India.

How much should one personalize?

Too much personalization can make it boring. It limits the scope of exploring new things. Therefore, one should not be getting into the loop of personalization, thinking that everything should be personalized. There should be some room left for new discoveries.

There has to be a mix of both: personalized spaces and non-personalized spaces.

Ideally, it’s good to have a 2:1 or 3:1 ratio between personalized and non-personalized content.

What other factors should one consider?

UI elements:  It should be kept simple and direct for customers. It is shown to enhance better recall.

“When” is equally important: short-term and long-term affinities are different. It is important to identify the difference between the two and curate content accordingly.

 However, personalization should not be done just for the sake of doing it and the outcomes could be really bad, if not executed properly, just like the picture below.