September 27th, 2012 by khushnood
Not sure, if you knew that you can also search International flights on 90di. The feature has been there for a while, but we did not mention it too much, as we were reliant on a single provider for that. But now, we have improved it a bit, and so perhaps time to tell you all about it.
There are lots of Airlines in India which travel international with one end of the journey in India. So we provide flights from Jet Airways, JetKonnect, Indigo, Spice Jet and Air India Express directly from these airlines.
Also for these (inbound/outbound flights to/from India) and all the other flights of the world we show flights from Expedia.
Some example queries:
1) Some popular routes from India – in these cases we show flights from the Indian carriers and also from Expedia
Delhi to Dubai flights on 20th
Chennai to Singapore flights on 22nd and back on 24th
2) Pure International flights – these we show from Expedia
SFO to Dallas flights on Sunday
3) Some other routes connecting smaller Indian towns with places abroad – Jet Airways and Air India Express do a good job of covering these routes
Lucknow to dubai flight on sunday
dubai to amritsar flights on next friday
Please do try it out. And give us your feedback. Last but not the least: please do remember to review the choice carefully before booking.
June 15th, 2012 by khushnood
Yes, 1,021,789! That’s the number of travel searches you did, dear users, on 90di.com in the last month.
We are ecstatic to reach the million search mark. And thank you all all for it. We love you for it. And hope that you celebrate with us, by doing more searches on 90di.com, and by spreading the word about us to your family and friends.
We believe in our model, and think its the best model to exist for travel bookings using the Internet.
We would also like to thank our partner Airlines on this occasion.
Indian Railways, we would request you to help us more, by being open. Give everybody an API. It will allow innovation to happen in this field. And people will love you more. Also allow us to make some commercial use of it.
Our Bus partners, we thank you.
Also open source software communities, we love you. We thank:
1. Linux: Where we develop, host and run our stuff.
2. Java and its eco system with lots of other open source libraries, created by the erstwhile Sun. Oracle, please-please don’t screw it up.
3. Apache web server – You never crashed in some 4/5 years of run! Need anything be said in your praise. We love you.
4. Tomcat and Jetty: We have used both of them, in various mixes at various times.
5. GWT: Thanks Google and GWT team for building such a cool piece of software. That said, the roadmap does look uncertain, and we will be forced to move to JQuery, or such, if you don’t dispel the doubts with actions in terms of some solid releases.
6. Demo Recorder: Its a brilliant piece of software to record stuff on linux. And it has helped us saved the day, during one presentation, when Internet connection died.
7. libevent: In our experience, libevent provides an excellent base to serve requests, if you want to code in C/C++, with perhaps 1/3rd the memory requirements of Java and twice as fast.
8. Eclipse: The IDE we use to generate our Java code.
9. vi: The best editor in the world, if you conquer it. For those, who have not had the pleasure of using it – Its the editor for coding C++, and shell scripts, and what not!
10. Memcached: The fastest way to cache some objects, and any thing. With the lowest possible memory & CPU footprint.
11. subversion(svn): For help managing our source code control (I know, I know, some people might say “You are not on ‘XYZ’ yet!”. The answer would be ‘will the change the shoe if it pinches me some where, not because some celebrity said, it pinched him!’)
12. MySQL: The best open source RDBMS, in the world.
Also, although not open source, we would like to thank Amazon AWS, for providing the best cloud services in the world. Which allows guerilla fighters, to more effectively compete with the corporates with armies!
March 23rd, 2012 by khushnood
Important announcement for flight travelers in India: JetLite name has been changed to JetKonnect, effective Monday March 26, 2012.
Apart from the above change, some of the flights which fly as ‘Jet Airways Konnect’ will also be called ‘JetKonnect’ flights.
So as far as Jet Airways group overall, they will be having flights under two brands: Jet Airways as a full service carrier and JetKonnect as the low cost carrier.
Here is the full press release, they have shared with their partners:
Jet Airways-Press Release-regarding JetKonnect
And this is new JetKonnect logo:
August 14th, 2011 by khushnood
Travel is a very hotly fought category in the search space and all colors of SEO are at work! We have been by and large a very reluctant participant in this game. Here is an opinion piece on state of search, earlier this year. But at the same time, we will not restrain from articulating our observations and analysis of any suspected malicious “SEO” activities, when it possibly could affect us.
The past two days data shows some strange user behavior from a category of users coming from Google search engine (which not-coincidentally happens to be the target game of most SEOs).
First lets see the data:
1) Visits from Google for the past month:
So far so good. The visits from Google have increased in the past 2 days. When the direct and referral traffic have relatively decreased, due to the independence day weekend in India. But more traffic. Surprised but Fine! We take it.
2) Now lets look at ‘Average time on Site’ (for users coming from Google):
Whats happening here? Why has the average time on site gone down? And by 2 minutes or so!
Bad. But we can’t do anything about it. May be the new people coming on our site, are not liking it as much. As we can’t think of any reason, why existing users would behave in such a way (and that too en-masse).
3) Bounce rate (for users coming from Google):
Ditto with Bounce rate. High by some 14% or so! in the past two days. So new users are not really liking it. Seems kind of weird but Okay, what can we do.
4) All the above might have made sense … until we see the ‘% of new users’ graph (again for users coming from Google):
Shockingly & Mysteriously, this is down by roughly 14% in the past two days. While for the previous 29 days, it has remained fairly straight. So what’s happening here? What the hell is happening? And where are the new users? And if there are less new users, then why is the total number of visits high?
After some thinking, yesterday and today, we came up with a theory. That these graphs are easily explained, taking into account, the behavior of people doing “SEO” labor. Explain it in a minute.
The data shows that there are actually less new users in the past two days, from Google. And it would be very strange for existing users to suddenly start typing random travel queries on google and start bouncing our site.
But however strange that may be, the data clearly suggests that it is some of the existing users who are doing this. Now who, then, could be these existing users?
What, likely, is happening is that for a lot of travel keywords, e.g. ‘chennai-mumbai train’ those existing users, are just searching google, clicking on 90di.com and exiting it immediately (i.e. bouncing it). And all this in a bid to decrease the good parameters on 90di.com (and possible other competing sites as well).
And at the same time, those existing users, are also perhaps searching for the same query on Google and clicking on <the-website-paying-for-his/her-work>.com and then spending a decent amount of time on that site. All this activity, and likely paid work, in a bid to increase the ‘avg-time-spent’ parameters on the <the-website-paying-for-his/her-work>.com.
If this is happening. And unless, someone has a better explanation, this is clearly black hat in our opinion. And a very sophisticated one at that. This is evil!
And Google (Attention: Matt Cutts!) may do well do to get to the bottom of all such activity. Perhaps they are already aware of such activity and their search engine algorithm takes care of this. So it will help, if somebody, points to some article/post which explains how this is handled.
We have heard that Google looks at hundreds of signals, to rank a website/page. Hope this case is also easy to handle. We hope that it does the following:
Ignore all increase in bounce rate, and decrease in avg-time-on-site, when accompanied with a proportionate increase in overall visits and a decrease in percent new visits.
If its already handled, it would not hurt to get reassured,by a post from somebody at Google (official or unofficial). If somebody has an alternate explanation for this, we would like to listen.
It must also be said here that our direct and referral traffic has been very consistent during this time. Normal long weekend dip. With flat avg-time-on-site and bounce rates (Displaying two relevant ones for direct users below for reference):
a) Direct visits graph:
Expected early dip for the long weekend.
2) Average time on spent by users who come directly:
By and large remains flat. Rather the usual upward hump, in the weekends due to visitors in perhaps a relaxed weekend mood and not in the usual hurry.
So, would want some action to be taken to curb such activity. And also, if possible, would be great to get reassured, that Google takes care of this, if indeed this is a sophisticated black hat at work, as we have deduced that it may just be.
Why bother about all this? Why not just focus on your own work?
Because, if this is indeed happening, then it directly affects us negatively. If somebody is doing conventional SEO of link building, in whatever shade of grey, it is usually none of our business. But what allegedly happened in this case is not fair, its evil. We will do our bit to bring it to notice.
And on a totally unrelated note, but nevertheless important. Not to forget, Wishing our Indian users a very Happy Independence Day Weekend and safe traveling if they are!
June 27th, 2011 by khushnood
The inspiration for this post is, some of the feedback, we got on our facebook page recently regarding savings some travellers, were able to make while using us to find the best flight/train and booking directly (at Airline or Railways site).
One user was able to save Rs. 2018/-. Where as another found a flight for 3.5K on our site, where as some of the big Online Travel Agents (OTA) were showing it for much-much higher!
We also did a post explaining our travel search model some time back. Here is a fresh take on it:
- Internet is about removing redundancies, and also superficial intermediaries. And that is primarily by virtue of making the information flow, easily possible, between entities (peoples & systems).
- In an ideal scenario, on the Internet, when user wants to book a flight, information from all the Airlines should just flow into the user’s tool (computer/any thing), without any intermediary. There is a seller there is a buyer. The two kinds are sufficient for the trade to happen. And both need each other. Every thing else, can be dispensed with, if not helping the trade, by adding value.
- Let’s say for the moment, that no intermediary exists, whether its the OTAs, or travel search engines like 90di. This removes the intermediaries, but then what happens is: User has has to open 10 tabs in the browser, one for each Airline, and one perhaps for IRCTC. So it will be kind of painful for the user.
- So basically, some kind of tool is required to make the travel booking process easier. To assist and assimilate the flow of travel information.
- But the question is where should the tool stop – meaning what all features it should not have. If it goes on to do booking on behalf of the user, and by that it adds to the cost of travel – hefty commissions; and the double cancellation charges (one to the Airline and one to the OTA). And thereby starting to drain the travel booking eco-system.
- More over, Airlines can’t pass on some of the advantages to consumer as part of the booking – For example purchasing food at the time of booking. This particular aspect is a lose-lose proposition for both the Airlines and the traveller. As user is constrained by the limitations of the “generic-for-all-Airlines” features provided by the OTA. And the Airline can’t ‘demonstrate’ more value to the consumer upfront.
- So it seems obvious to us, and hopefully to many of you reading this, that travel search engines are the kind of tools you need for travel planning, and nothing more. Anything more – like the booking – and the tool starts to come in your way.
We are glad to be of value to our users. And although, there are some difficulties(+), we try and continue, in the belief, that we are instrumental in the travel search model getting established in India. Which, we believe, is in the best interests of both the suppliers (like Indian Railways and Airlines) and, equally if not more important, for the consumer. We didn’t just say it when we said ‘We are on the traveller’s side’!
+ Difficulties like not getting paid by all the Airlines yet — it’s taking some time. And not making anything on the trains’ search – Indian Railways does not yet support the travel search model (although we are hopeful, and have been patiently following up with them). But there are silver lines. Big Thanks to the Airlines that pay us, we are able to keep ourselves motivated and, which reinforces our belief in this model.
June 24th, 2011 by khushnood
Have been feeling the need for Internet 2.0 (not web 2.0!), with social protocols, to support applications like Facebook, Twitter wired in. Thereby allowing people to choose between multiple ‘social’ providers.
Just like mails (via SMTP etc.) have been there from early days of Internet, with multiple email providers including Desktop clients.
Else, we will have the problem of a private body (e.g. Facebook) controlling way too much. Like what Google does by being the ‘start page’ of the Internet, and now people are talking about search being part of Internet (there were also a couple NY Times articles some time back on the same).
A paranoid user, should be able to have everything on her Desktop, or her own storage in the cloud. But communicate to others via a ‘social’ protocol (akin to SMTP).
So just like one has gmail/yahoo and other mail, one should be able to choose his/her ‘social’ account(s) and still be able to collaborate with his/her friends using the ‘social’ protocol.
To be more clear in such an environment it doesn’t matter if I am on Orkut or Facebook or on both. Or simply have hacked up a desktop tool, which can talk ‘social’.
I think the Open social protocol has not really taken off in that direction. What I have in mind is just an application level protocol like HTTP/SMTP, which is directly over TCP. Hence it should be more of an Internet standard (owned by IETF types) rather than a consortium of corporates (which I think Open Social is).
NOTE: This blog has opinions and thoughts, which are personal of Khushnood Naqvi, and do not necessarily reflect the opinion of 90di.
PS: I have not done deep research to write this note. And this was written, some time back. But I still believe, most of the things are still valid – as in Facebook continues to be the dominant social network. Would like to be educated on any efforts, which are in this direction, or counter view points.
February 10th, 2011 by khushnood
Don’t worry, its not a post on psychology. Just a whacky title.
Not just systems, but humans (and by extension data) too needs backward compatibility. Case in point the Train numbers have changed from well over a month. But lots of users are still searching for the old numbers:
And also, Got one feedback yesterday, which said:
“i asked for rail fare, it gave result for flight…”
search query pertaining to above feedback was:
“rail fare of 2433, rajdhani express”
What was happening in this case was, since it got a flight with no. 2433. It unwittingly showed the results of the same.
So hey you anonymous friend, who got us to act, thanks! And some good news for you, and also to other users, who must have been suffering in silence. We made a change so you can search like this also, for as long as you like. And when you do get used to the new numbers, that are of course supported as well.
January 31st, 2011 by khushnood
What would a future search engine be like?
The age of brute force search engines may be about to be over. Or atleast the nature of search engines, are due for a drastic change. By brute force I mean, an approach to crank out meaning, information, and relevance, to the search query, by just looking in tera bytes and peta bytes of data on the Internet. And of course, things like number of back links to that page.
To those, following the areas of Search Engines, its pretty common knowledge, that the conventional Google-like Search Engines are no good for “high quality relevant discovery”. In fact they have been no good, for such information for a good 4 years or so, in my opinion. What we have now, is a situation of “relevant but random discovery”.
Let me explain, it with an example.
Try searching for “video cameras in India” and you get some results. And they all are relevant. But there’s no guarantee, that they are sorted in terms of a criteria, which you may want, e.g. sites having quality information about a camera. So I call this as a “relevant but random discovery” rather than a “high quality relevant discovery”.
As the information they get is still relevant, in some way, is the reason, why most folks are still happy to live with it. And people are used to just getting by, by tweeking their queries, and getting some satisfactory enough result. The problem is that people don’t even know what they may be missing. Remember no body actually wanted a real nice search engine, when Google silently arrived.
Another thing, why no body sees a problem is that, when using the search engine just for navigation. For e.g. finding out “Java 5 TimerTask”. Its brilliant. You get precisely what you want in the first two results (Aside: they still have 32,100 results for that. That’s why I perhaps used the phrase ‘brute force’)
Was the situation, always like this and people didn’t know?
Certainly not. After the Google arrived in 1998, for some good 5-7 years or so, its PageRank algorithm worked very well. The algorithm works very well, if people don’t game the system. But, we have full blown SEO industry around the Search Engine. And despite Google’s efforts to keep the SEO clean. It’s simply not possible to do it. In today’s Internet world, it will be increasingly difficult to identify for a clean link to a site, to a paid (SEM’d) link. The links from social media, only further add to this noise.
So what’s the solution?
On your machine, how do you find a file. You often search for it, using the file system search. But you know, what file you are looking for. Think, the conventional brute-force search engine, will move more towards navigation, in which a person knows what he wants, and tries to find it.
So something similar will happen on the Internet. Ofcourse the person, doesn’t know which website to look for often, so he will ask his friends for advise. Wait a minute, did you say, ask friends, come on who has the time to ask, and then wait for a reply?
Well the answer is that the asking and replying will happen asynchronously. On all the social media, people are continously telling what they like and what they don’t and sharing various experiences. “This rose smells nice”, “This coffee is not so great” and “this travel site is so great” Or “this site doesn’t rock”. So some body has to just assimilate all these statements of expressions and churn them as replies when a question is being asked.
Several folks are already at play in this area e.g. quora.com – the arguments against it are that its a closed service, but that’s a different story. But there is far-far from a clear winner. Its possible, but unlikely, that the likes of Google itself, will totally transform themselves into a totally different kind of a search engine. Unlikely I say, is because, its not always possible to kill a cash cow. The conventional brute-force search engine is perhaps at the peak of money making curve, by the various Ad programs – Ad words and Ad Sense. So in such a scenario, will somebody be able to shake of the inertia and start from a clean slate. Possible but unlikely.
There were some efforts, which I was very excited about like blekko.com. But they appeared to be working on a conventional ‘brute-force’ search engine (of course on a better version of it), atleast until about a couple of years back. This was around the time when cuil (remember?) failed spectacularly. It will be interesting to see how they launch, in a post twitter/facebook world.
It will be interesting to see, who get’s there first, and importantly who gets it right.
NOTE: This blog has opinions and thoughts, which are personal of Khushnood Naqvi, and do not necessarily reflect the opinion of 90di.
PS: Had written this piece a few months back, when blekko was yet to be launched. But shied away from blogging it here. Thanks to Vijay, Patrick, Shaurabh, Kiran and Abhinit for reading it. Since then the noise around search quality has only increased as seen in these blogs: Codding Horror; Blekko founder; Google.
January 12th, 2011 by khushnood
‘Green’ is the word in vogue, along with other phrases like ‘Carbon footprint’, and rightly so, whenever there is talk about eco-friendliness or lack of it.
So here is one simple thought to ponder. Travel Search is more ‘green’ …!
To complete the phrase:
‘Travel Search’ model is more ‘green’ than the Online Travel Agent (OTA) model.
First lets define what ‘green’ means in this context. Green means being eco-friendly, which means doing things in a way, which is light on the environment. That is, in simple terms we can define it as: ‘Consume less and produce more’.
For the uninitiated: Travel Search Engines allow the users to search on their sites and for booking take them to the Airline site (or Railways/IRCTC). Whereas OTA sites give a search option to the user, and also make the users book with them.
Now lets go on and explore why Travel Search is more ‘green’.
People who want to travel need to compare flight (or train) options before buying – i.e. search the cheapest and the best flight (or train). So some kind of search on top of the Airlines sites and Railway sites is required. Else the user will have to do the comparison on various sites themselves. Both the Travel Search model and the OTA model fulfill that need.
One thing which OTA also additionally provide to the user is booking. I contest, it is this aspect which makes it less eco-friendly to the entire flight booking eco-system. It creates an extra (larglely redundant) middle layer of providing the booking itself. Which the Airlines themselves do, very nicely.
In doing so, it adds a thick layer of actions, which are largely duplicated and can easily be performed by the Airline itself (Or by the Railways/IRCTC).
To meet this extra cost, it has to charge heavy commissions. So it adds significant cost to Air travel, by charging commissions often as high as 7-8%.
In addition, the OTAs charge their own cancellation fee (or a change fee), in case of cancellation or a change in itinerary. It also has to run call centers, since it does booking. Also it can come in the way of loyalty programs between the customer and the Airline (Some of them make it workable, but its not straight forward).
Now compare this with what ‘Travel Search’ does. It serves an essential function, of allowing travellers to compare flight options (or train options or both) and choose the best flight (or train or a bus). And doesn’t duplicate the function of booking, which the Airlines do equally well.
It costs less to the eco-system. The referrals paid to travel search players are typically just about 1/10th of the hefty commissions which the Airlines pay to the OTAs. (For this reason its less attractive to VCs, but that’s another story)
It costs less to build. Just a small team of may be around 10/20 people, who sit in a garage (and largely survive on Maggie noodles!) to build a nice and strong search engine. Which allows the users to search the best flight and book directly at the Airline.
Compare this with typically hundreds of employees employed by an OTA, largely to fulfill the additional function of booking.
But don’t the OTAs offer lots of deals and discounts? Yes they do. But in simple terms, it adds further to your average travel cost, as the discount and deals on may be less than 5% of all the bookings/routes, increase the cost of overall bookings. As they have to earn money, and somebody has got to pay for it. The deals and discounts, serve mainly a marketing purpose.
So Travel Search, costs less to build. And gives more value to the user – in terms of providing just what the user wants and then getting out of the user’s way. Less is indeed More – and Travel Search model is yet another example of that!
This is our honest opinion, and would love to debate this, and learn more about it. If you agree to this, then please spread the word about us and this model, to your family and friends (we don’t have the budgets for a TV advertising you see!)
1. Travel Agents: We believe all the field travel agents (i.e. the ones having a physical presence in form a store, you can walk-in to), serve a very important need, and the above argument does not apply there. Majority of the folks in India, still purchase their travel offline.
2. OTA evolution: It will be unfair to not even mention a bit on the history of this Industry. In the pre-internet days, when travel agents in US had to do bookings for their customers, they relied on Global distribution systems (GDS) – like Sabre, Amadeus etc. to do the booking. So in a way, the GDSes, used to manage the entire Airline seat inventory. The first generation of online booking sites, didn’t sort of look at the system, from a clean slate view, and just moved everything online –> Online Travel Agent (OTA). Travel Search is a clean-slate perspective, of looking at things overall. It has established itself in US, firmly, with the likes of Kayak. Now, in India, its in the process of establishing itself.
3. Disclosure: We allow users of our site, to compare flight results from Airlines with OTAs. Also we get a commission from the OTA if a booking is done there. So in a sense, you may detect a shade of hypocrisy, in what I say, above.
But I have an explanation for that. We do see that the majority of the bookings happen through the Airlines itself. Something like more than 95% of the bookings, we see, happen on the Airlines’ site. We give an option to the user, to catch any deals, which may happen at that point in time, which the OTAs do as part of their marketing.
So this blog post is about our world view, but we do accept the reality (of OTAs commanding a large mind share) and we try to make the best use of it. But we totally stand by what we say above, and believe that, it will serve the people better, when ‘Travel Search’ is the dominant model, rather that the ‘OTA’. Since Internet is also about removing redundancies, we believe ‘Travel Search’ is the model, which is going to survive and do well in the long run. Only time will tell…
Acknowledgement: Thanks to Patrick, Sanjeev, Abhinit, Rockey, Kalyan, Barry, Vijay, Birla, Gopi, Kiran and Iram for reading and reviewing drafts of this post.
December 15th, 2010 by khushnood
You must have heard about it. Indian Railways is going in for a numbering scheme change. Effective 20th Dec, we will have the scheme change from 4 digit numbers to 5 digit numbers. Largely, it should be smooth, as the bulk of the trains – all Mails/Expresses/Rajdhanis/Shatabdis, will just have a prefix of 1 to the existing number. For example:
2833 (Howrah Express) becomes 12833
2493 (Rajdhani Express) becomes 12493
The special trains are supposed to have a prefix 0. But a check into the page set up by Indian Railways, doesn’t show this to be the case consistently. For example:
0233 (Hmh Jp Special) becomes 09733
0465(Dli Pnp Special) remains 0465
You can check for a train number change, on this page set up by Railways.
Update (on 21/Dec):
We moved to the new train numbers: Here is a full list of changes.