How Personalisation is Helping Snapdeal Increase Conversion by 3x- 5x

Find out how e-commerce player Snapdeal is increasing conversions by personalizing its site.

It’s all about the brand experience which creates a community of loyal users. And when it comes to online shopping, the experience becomes much more important as there is no physical interaction. Online portals are walking an extra mile to know more about their users and personalise the whole experience.

Online marketplace recently started personalizing the shopping experience using more than 60 million pieces of user data on more than 10 million items in their catalog. The large combination of data is helping the retailer sell more, a top executive at the company says. Personalisation has added more than 25% to its topline, said the company.

We caught up with Ankit Khanna, Vice President- Product Management at to understand how personalisation helps. Here are some key takeaways.

1. Personalized emails drive conversions

Prersonalised e-mailsPersonalized emails play an important part for drawing audience. Emails should be designed for an individual, on the basis of his/her affinity for any category, products, gender, age, purchase history, browsing path and other parameters.

Snapdeal targets 40% of its database with these user level data emails. As a result conversion rates are 3 to 5 times higher from these 40% of users in comparison to rest. Remaining 60% of the users are targeted with segment level, recommendation and promotional emails.

2. Retargeting is not Personalization

Differentiating re-targeting with rationalization, Ankit explains, “In re-targeting normally the last visit of the user is traced and on the basis of that recommendations are shown. At Snapdeal we use at least last 3 visits such as size preference, color preference and brand preference before recommending user on the basis of it.”

3. Product recommendations and patentsRecommendations for you

If you shop online then surely you might have experienced product recommendations at one or the other site. Things like similar products, people who viewed also viewed this; bestsellers etc. are some examples of product recommendations. Snapdeal claims to process over 200 million rows of data to arrive at right recommendations and has covered 95% of its products barring one category.

Similar ProductsTo refer certain products which are complementary to others, cart recommendations are used. An algorithm supports these recommendations which are visible in the checkout process. The company has not filed any patent as of now and wants to test the algorithms effectively before moving ahead. Although Amazon has had a headstart on patents for personalised recommendations.

4. Evaluation and Impact?

To evaluate  these services and features, the company is doing two types of A/B testing. At the user interface level, it involves changing the position of recommendations etc. But it is not possible for a long time on a live website. Also, it may not be a good idea to tamper with the experience too often.

At the application level, its more about verifying the product. Here, the base is kept constant and variables are varied for a fixed set of users.

To measure  impact it keeps a close watch on two important matrices:

  • Comparison of conversion rates of people who used these recommendations vs conversion of rates of the website.
  • Percentage of revenue from audience based on recommendations

5. Underlying Technology

Snapdeal uses proprietary algorithms based on collaborative/content based filtering and various classification techniques which form the core of the personalized platform, says Ankit. Though these algorithms are open source, Snapdeal has built a system on the top of that. “It’s the accuracy factor that distinguishes us from other portals”, he added.

Java, python and statistical tools like R are used as an underlying technology to enable the product. Cookies are used as a preferred technology to track user behavior on web. But for other devices such as mobile or tablet where cookies do not work, data is mapped on email level taking certain user attributes into account.

Recommended Read: A/B Testing on Savita Bhabhi the Movie

Download NBW: Short news app created for busy professionals like you

Get NBW App - a reimagined fluid short news experience that delivers you clarity and all the important news and trends from your industry. No signup needed !

Download NBW App (Android, iOS)

Sign Up for NextBigWhat Newsletter