October 04, 2011


Dear Flipkart, Please disallow guest purchase for registered users

Flipkart has been / is one of my favourite shopping destination. My dozens of purchases with them have been smooth. Until I hit a block last week.

Flipkart allows anyone to buy without being a registered user of the site. This policy itself isn't at fault. If someone wishes to buy quickly, forcing that user to create account adds friction to the sale. So, that is fine.

Here is what happened with my account. A user placed order on Flipkart and provided my email address in the contact field. His order now appeared in my account. I sent a mail to Flipkart to remove that order from my account. Instead, Flipkart folks called up that user, and changed email address associated with my account to his email address. This effectively locked me out and my orders now were visible to him. Flipkart folks promised to fix this in "next two hours." I waited for a week and asked them to fix the problem. This time they fixed it within a few hours as against promise of one day.

I had asked them to fix this in an earlier mail as well and here I repeat it again. Please do not allow a guest purchase if the email address is already registered with you. If the email address is already registered, ask the user to login to the account.

August 03, 2011


I don't care about e-commerce bubble in India

Here is my bullet-point take on the discussion on E-commerce bubble in India.
  • My earliest online shopping experience is from 2003. I had ordered a dozen music CDs from EBay. Broken CDs were shipped (or they got damaged during shipping.) I returned the CDs but never got the replacement. Multiple mails were ignored by seller and EBay. My trust in online shopping dropped below zero. My experiment of shopping online met an early death.
  • In last 6 years, I have booked train tickets only via IRCTC. Only once I went to the actual ticket counter when repeated attempts to book tickets for a group failed on IRCTC (blocking close to Rs 10,000 in the process) and those tickets had to be booked urgently.
  • In last 3 years, I have booked movie tickets on BookMyShow. If I can't book tickets online, I don't watch the film.
  • After hearing ample praise about Flipkart and Infibeam, I restarted shopping online at the beginning of 2010 with them. As you can see, I am not an early adopter in strictest sense.
  • In last 18 months, I have bought dozens of books, movie DVDs, phones, printer, laptop bag, laptop cooling pad, portable hard disk and headphones from variety of online stores (Flipkart, Infibeam, IndiaPlaza, Ebay, and more recently LetsBuy.) Collectively I have spent more than Rs 40,000 on these purchases. Every single item has been arrived in excellent condition. My skepticism about getting the delicate printer (actually, print-scan-copy machine) shipped in good condition were shattered when wonderful folks at LetsBuy packed it nicely and sent with BlueDart.
  • I am not counting ordering Dell laptop on phone as online shopping. The amount of running around required to get a laptop from them is a story of another post.
  • Point is, my shopping habits have changed irrevocably in the same way IRCTC and BookMyShow changed the ticketing habit.
  • Excellent discounts, free shipping and cash on delivery are some really strong USPs with which Indian e-commerce services are attracting new customers. While the companies themselves might be losing money to offer these features, they are changing the consumer behaviour in a big way.
  • I wouldn't be surprised if consumers check the products like phone in a offline store and check the best deal available online before making purchase decision. From my limited experience, as of now, the prices of phones from online stores are not matched by the electronic store in the neighbourhood.
  • There are going to be companies who will capitalize on this new behaviour. Even though, the current online audience is estimated at 100 million, these are among the people who have disproportionate purchasing power. The opportunity is quite clear.
  • Now, all players may not have/get the required capital, execution skills or plain luck to be a long term player. A few players who do deliver and tide over the obstacles will own the market. The investors are probably betting that their investment is going to be the one that will turn out to be among the eventual winners.
  • Frankly, as a consumer, I don't really care if a company gets valued at 1 billion dollar or 3. If some companies get affected when the so-called bubble bursts, I will simply switch to the surviving player. In the worst case, I can go back to shopping the way I used to before 2010, though chances of that happening seem low. And, unless you have material interest in survival of specific companies, neither should you worry about the bubble talk.
  • Rephrasing the same, some specific companies may or may not survive in the long run, but online shopping as an industry is here to stay.
  • Well, I lied about the subject. I do care about success of e-commerce companies. I want them to succeed in a big way. Arguably, this is the biggest opportunity that Indian internet industry has seen till now and many startups are well-placed to ride this wave. A few big success stories will put the startup ecosystem in a new orbit.

November 02, 2010


Computer Science Final Year Project Ideas

If you are looking for project ideas for your final year course requirements.
  • Build stemmer for Indian Languages. Read about Stemming on Wikipedia. To put simply, stemming reduce a given word to its root form. For example, if word is "books", its stemmed for would be "book". This is very useful in search. Anyone searching for "book" will also get results matching "books." Presently, there is a Hindi stemmer available in Lucene. It would be great to have stemmer for major Indian languages.
  • Part-Of-Speech tagger for Indian Languages. Probably this needs ground up work starting from building a tagged corpus. Microsoft has a team working on this. If this is your area of interest, you may land up in that team.
These projects require math + linguistic + computer science skills. So, these are quite challenging.


May 29, 2010


Most-visited Websites from India according to Google Ad Planner

Google Ad Planner published the list global Top 1000 sites. I picked the Indian sites from the list. This original list does not have Yahoo India, Facebook/Orkut/Wikipedia/Live (from India), as they are counted under their global site. But you can get traffic for these sites from India with Google Ad Planner. The global rank assigned for these sites is what it would have been if they were included in the list independently. MSN India doesn't have enough traffic to make it to top 1000.

Google has excused itself from this "beauty-contest" list. So there are no numbers for Google.com, Google.co.in and Youtube.com. Other category of exclusion is pr0n, that are included in Alexa's top Indian sites.

(Apologies for bad colour formatting. I am unable to paste the table from OpenOffice, cleanly. And blogger is horribly broken.)

Global Rank Site Unique Visitors / Month Page Views / Month
108 in.yahoo.com 21 M 250 M
126 facebook.com 19 M 2.5 Billion
167 orkut.com 16 M 140 M
234 rediff.com 12 M 960 M
303 wikipedia.org 10 M 89 M
389 indiatimes.com 8.9 M 330 M
501 in.com 7.4 M 89M
578 live.com 6.2 M 160 M
763 irctc.co.in 5.1M 190 M
796 cricinfo.com 5 M 190 M

Labels: , ,

May 24, 2010


Should you builld custom search for your site?

Google has just announced the revenue sharing structure for AdSense. I suppose, recent announcement from Apple about iAds and making the revenue sharing public forced Google to go public on this. After all, they are a huge proponents of "All Things Open." (tm)

According to the announcements, publishers get 68% of revenues for Adsense for Content. This number stands at 51% for Adsense for Search. Adsense for Search is the ads shown by Google when publishers use Google's custom search. Take a look at custom search on TechCrunch. (BTW, it seems, TechCrunch has silently dropped Yahoo BOSS that they had implemented one and half years back.)

Google keeps additional 17% when you use their search.

Now, if publishers is a big player (think TechCrunch), 17% revenue might translate to a meaningful number. These publishers can build a custom search with Lucene/Solr and use Google Adsense for Content on those pages to show ads.

Google custom search is trivial to implement compared to building and maintaining custom Lucene/Solr search. I don't have numbers at hand, but, few thousands dollars seem to be the tipping point where building custom search is a better idea. Just to put a concrete number, I would say, if revenues from Adsense for Search is in excess of $60,000 (per year), custom search is the way to go.

Labels: , , ,

This page is powered by Blogger. Isn't yours?