Personalizing online grocery shopping for new visitors

A way to Improve Conversion Rate of New Visitors

Show what customers prefer, not the banners that we would prefer.

Problem Statement

How can the product help first time online grocery shoppers to build their basket quickly?

As a customer — How do you start shopping in an online grocery app with 50,000 products?

We don’t want first time visitors to land on page filled with banners and expect them to explore on their own! Rather we want to ensure they see what they would like to see and not what we would like to show. That’s basically personalizing your products discovery. Now how do we do this was the problem which we solved in Designathon competition held at bigbasket.

Personalization is comparatively easier for repeat customers because we would know their shopping behavior, thanks to data analytics. But how do we personalize the supermarket for first time visitors without asking many questions? Remember more questions means more drop offs or skips.

Personalize — Ask only two questions

Question 1

The first question has to be easy to answer, basically without any cognitive load. How does this question help?

• We can avoid showing meat products to vegetarians, vegans & eggetarians. Rather use that space wisely for showing other relevant products.
• We can improve discovery of niche products, especially in vegan and non veg category.

Question 2

Out of 100+ categories, we arrived at below 7 categories and here’s why —

1. Cut Fruits & Vegetables — Unique to bigbasket. A differentiator.
2. Baby Care — Show the non obvious. Also we don’t want to show diapers ads to bachelors.
3. Cosmetics — A good proxy to determine one’s gender. We didn’t want a question like “Please choose your gender”.
4. Chocolates & Biscuits — One option has to be super easy to choose! :)
5. Organic — A decent indicator to know if customer is price sensitive, because “organic products” is perceived to be expensive. Also to show niche products in the catalog, which would be difficult to discover otherwise.
6. Ready to eat — Common in the supermarket, every household has it.
7. Breads — One of the sticky categories, which would result in repeat shopping.

What next? — The Personalization Engine

The engine will generate a list of products which are the ‘Ideas for your first order’.

As you may know, a product list has 2 main components — The result set & their sort order.

Step 1: Fetch products

Pick top 10 highest selling products from each of chosen category

Step 2: Sort based on category weightage

Assign priority/ weightage to each category to decide which category products should be shown on top

E.g: Food preference > Cut Fruits & Vegetables > Baby care > Chocolates.

Step 3: Use 2–1–2 Product Injection Model

We wanted to leverage the power of existing ‘Smart Basket’. It is from analytics driven logic which gives top selling products in a city (vegetables, atta, dals, etc.,)

Take the list from the personalization engine (Step 1&2). After every 2 products, inject a product from smart basket list.

This will ensure visitor sees a mix of top selling products in the city and products of his interests.

One example —

1. Visitor chose Non veg+Baby care+Chocolates & Biscuits+Ready to eat.
2. Weightage is set in the order Non veg > Baby care > Ready to eat > Chocolates & Biscuits
3. Personalization engine fetches top 10 products from each of the above 4 categories. Sorts it. It also fetches top products from smart basket.
4. List would be — Chicken (non veg), Nestle Nan pro powder (baby care), Onion (smart basket), Idly batter (ready to eat), Dairy milk (chocolates), potato (smart basket), Chicken-boneless, pampers diapers and so on..
5. Similarly, if visitor chose eggetarain, he would see eggs in place of chicken.

20 seconds video of a personalization example

To conclude on the solution —

1. Job here is to get the visitors started! Instead of asking questions in the mainstream way, understand user’s interests and infer persona based on the categories chosen.
2. Banner blindness is reality, so ease the path for visitors to find their products.
3. Highlight categories which are sticky and unique differentiators of bigbasket.

Impact.

Immediate — Increase in visitor to order placement conversion.

Increases average number of items added to basket which increases probability to start checkout. This would naturally improve the % of new visitors who place their first order (because humans tend to complete a task when they have invested time and energy).

Long term — Broaden the brand perception

We are proactively increasing category awareness right from the beginning. A real life scenario — bigbasket is often just associated with “grocery”, it’s been a challenge for us to increase awareness among other categories like gourmet, cosmetics.

Current new visitor on-boarding

Proposed new visitor on-boarding

Few questions which may strike you —

1. What if customer added few items from personalized list now and came back later to add more. Do they have to go through these steps again? — No. There would be a constant widget ‘Ideas for your first order’ on the home page for easy access. This is because in grocery shopping we have learnt that customers build their basket over multiple sessions. (Check the 5th screen in the above image)
2. Why did we skip ‘staples’ in the category selection question? — Staples is a given when it comes to ‘grocery’. Also, staples products are part of the final list because we are showing products from smart basket where top products are rice, atta, dals etc.,
3. Can we scale this system for returning customers? — Yes. If a customer finds this personalization relevant, then we could provide this option under My Account where they can quickly customize product listing pages.

If there are any questions or suggestions, please feel free to comment or reach out to me. Thank you for your attention. :)

Team Lorem Ipsum — Nimit Jain, Rahul (rahul baskaran) & Bhavana A S