Customized database marketing for Indian e-commerce

Most of the players in the Indian e-commerce market are losing money.

My friend, Sanjay Dattatri has blogged about the challenges facing Indian e-commerce (click here). As he points out, Indian consumers are price-sensitive, dis-loyal and ill-behaved. I tend to agree with him.

As Sanjay points in another blog post, pricing sanity needs to prevail (click here). I agree with him on this one as well.

I do think that e-commerce vendors in India need to adopt a more customized approach to marketing and pricing. After all, aren't online marketers supposed to know all about their customers' buying behavior, browsing, email response (and customer care calls).

Why give the discount to all your customers when the reality is that some customers value your brand and superior delivery performance (and couldn't care less about the few rupees extra for that). The vendor can give a golden treatment to these price-insensitive customers by giving differential free shipping terms, and loyalty points. Some players like indiaplaza and healthkart do have a loyalty program.

On the other hand, one cannot lose out on selling to the price-sensitive customers, who will switch on a better deal available somewhere else.

The only way out is customized database marketing.

In my opinion, e-commerce vendors need to stick to the following rules of customized database marketing:

1) Target your discounts at customers who are likely to shift their purchases from your competition to your website, when given a discount. Examples of these customers include:

• Customers who bought a regular purchase like clothes, or diapers, or books, or groceries or nutritional products many months back and have not purchased since. I was delighted to see that healthkart sent me a  customized coupon code for whey protein 6 months after my first purchase. They avoided giving the discount to everyone as they know they have targeted a potential buyer. The key is to send out a customized coupon code which works only with the specific customer's login.

• Customers who have browsed a particular product or a category for a long time and not bought that product/ category. This can be done only when the customer is browsing with his cookie activated, and the e-commerce vendor is smart about tracking browsing behavior. A classic example would be a books-buyer browsing a mobile phone product page, getting a customized offer. My recommended approach would be to price the product at slightly higher than competition for the masses, and to give a discount coupon to customers who browsed and did not buy. (Quick programming note to vendors: Please don't send discount coupons for lingerie to men who are just browsing. But for that you've got to read the next point.)

2) Get to know more about the customer (even though you may know nothing to begin with). I get shocked when online retailers say that they deliberately have no information about their customer other than their user ID, name and shipping address. As a trusted seller you should try to get more information about the customer. Incentivize customers to update their profile online (such as gender, date of birth, marital status, interests, etc). The Payback program (called imint earlier) did that when they transitioned. When you know that you are dealing with a single guy who has never bought women wear before, you can stop sending him lingerie ads and coupons. Also some interesting things you can do with demographic data:

• Promote Gudi Padva to Maharastrians, and Onam to the Keralites and Eid to Muslims (irrespective of where they live)! -- if you can classify names by mother tongue and religion
• Promote kids items to families with kids

3) Send personalized recommendations/ suggestions, by putting yourself in the customer's shoes. Most e-commerce vendors are using formulaic recommendation engines which give recommendations based on the customer's last purchase. Ideally the recommendation engine should be based on the basket of goods purchased by the customer in the last 6 months.

Also, the recommendation logic needs to be altered for capital goods purchases (mobiles, electronics) vs. regular purchases (books). A mobile phone buyer does not need recommendations on what other similar mobile phones to buy! However a book buyer may appreciate recommendations on which other similar books to buy. For the mobile phone buyer, a list of accessories which go with his/her phone may be useful, but a formulaic recommendation engine may not capture that.

More later...let me know what you think!