My vision on Interview Street
Update: I put the below blog, just to get some criticism and write all that I had on my head. Excuse me for it's tone, or any pretension of knowledge here forth.
Interview Street (IS) will hope to partner with institutions on a grass-root level. The approach here is to partner batch by batch, at least in the initial stages.
Filtering the list, by removing those who put Google Ads on their page, that leaves to about 10 players. [R3] There is no body who attract high quality crowds -- For instance, there's no one whom Microsoft or Google places an Ad on. [R4] Further the number of sites that specifically attract freshers, is less than 5. Even the moderate ones, seem to have a high number of audience, which shows the need for better solutions in this space.
It's time to go
1. Don't be shabby:
There are basics in web-site design optimization, that are not met in the the big player's websites. For example, Naukri clutters it's home page with all job listings. A great page, is simple and the parts that take the click are highlighted. Filled with carefully photo-shopped media, it briefs the same content redundantly, at differrent levels of shortness. This is no small deal. Monster does a better job, in their UI, than most people. The content clutter they dont seem to be able to avoid it as well.
2. Match - Dont Search:
Everyone either shamelessly classify and list all their openings for click navigation, or spend a decent effort to provide a search option. The list is either too large or too small. It's either too unclear or not so useful. It's ok if the list is too large, but the top results have to solve your problem. Again relevance is so poor, since it expects users to type everything down in a search box.
There's is nothing to search about jobs. Most options used for search are static, and must belong to a person's profile. Also job searching has very little choices to make. In most cases there exists a well defined ordering of job preference that holds good across a critical mass.
InterviewStreet will build custom web-apps, that are: [R5]
A. Recommendation engines that match user profiles to jobs, with high clarity of their initial report. Better targeting will improve precision, and form the theme of the game.
B. You apply to about 5 jobs and track, their status -- not a blank [applied] sign, something more user involving than that. He makes job specific notes, and book-marks, with tracking web-apps specifically designed for this purpose. This can be extended socially, with he being able to share it to people who study in the same college, class, etc.
C. Job targeted training or training recommendations will be offerred -- More on this later.
3. User generated content:
The web is about a constant user presence. Stickiness is important. People spend time in two ways on your page.
Firstly, and most commonly, they are in between something and open your site, just the way they open e-mail. To engage them, a good update rate on your quality content is necessary. User engagement rate is measured in number of articles published per week. Necessarily, the older ones get read lesser and lesser. [R6]
Secondly, people come when they are in need. They have an interview tomorrow, and want to look for historically relevant content. A well linked web of pages will do a great deal here. There are certain articles that cover the "head" of the domain specific problems. They will get heavily linked into. Content will be reached by links and equally often by classified navigations. The actual collection that the user first reaches will essentially be user generated. The tail is so large, and scaling will happen with user generated content.
There are no websites that do both of this. People who work like publishers today, have unworthy content with low update rates. [R7] Those who generate content from users, have an incredible lack of "structure". Everything is plain text, posted on forums.
There is need for structure and better apps. Eventually specific algorithms will 1. cluster, 2. classify and 3. and search this data pool. For a concrete example, when a user posts a new interview question that he came accross, it'd get matched with the existing pool, and the relevant questions will appear. He then can edit, with a simple interface and see if he can merge the two to mean, a complete question, adding value to the problem. No history is ever lost. In addition rich linkage structures across question papers are set, on top of which user generated apps can mine this data. For instance, it's now simple to determine, how many questions overlap across time in CTS interviews.
The vision is to make it so simple that, a student who just got an interview done, will be sent requests for a paper by his class mates, and instead of explaining it to all of them, he hosts it on your page (or) his own page, as add-on widgets. The motivation to host it online, is he get's to see some cool stats, as to how long people (a lot of people), took to solve the problem he solved. Added ease of distribution, with a feel of content ownership.
In the early days, for the first type of content, the founders can post upto one article a week, and get quality posters to write once a month. The second type of content needs a lot of importing todo. Semi-automated scripts, will import rich content by crawling other sites, and impose structure. Quality here is more important than quantity, since this sets the best case examples for later content added.
4. Scaling:
Any one who chooses the domain of a jobs and a training site, makes a trade-off between how much 1:1 human engagement you can commit, and how large a rate you want to grow at. There's nothing new in understanding that, human engagement doesnt scale. People like NIIT represent the left end and others like, Monster India chose the no-training path. But training and jobs go together. They cant be seperated. Either of them nurture the other. It's like making buyers and sellers meet, with online shopping. [R8]
The challenge is in providing scalable training. I was recently impressed by AppJet's efforts to create engaging tutorials. Web-seminars and recorded content goes a great way as well. Apps built around a rich repository of questions, that can auto recommend puzzles each time a user logs in, is a great way to engage. Auto evaluation systems that can score user generated problem-sets will emerge. A domain specific example, for most programming problems you can write engines to evaluate by running the program. The ease and simplicity with which ordinary users can create evaluators will determine how many sets are scorable.
The online interview meetings that they host, can be recorded and archived. The context information about the interview can be built around user profiles - again something recommendation engines can leverage.
Institute specific work shops will serve as starting points. (ofcourse that doesn't scale.) Facebook grew by dominating specific universites and opened up gradually.
5. Marketing:
Marketing trends in the web are changing and most players havent adopted yet. They still do news paper advertising as thier main reach. For web-sites, apart from the biggest form of advertesting, which is creating sticky users, social networks are big tools for growth. A new web-app model is necessary which can easily build apps into facebook, orkut, nokia, and run on web-sites as add-on widgets. Youtube and Slide grew immensely with that.
Google contact list is a great place to viral. Lets say, A new user who wants to register, if he lets your app to mail all his gmail contacts, and has greater than 20 contacts, gets a 10% off. [R9]
Opportunity
This space is fast growing and more players will eventually jump in. More mature and clean sites will have an edge, long term.
Most university placements happen through dedicated departments, that puts restriction on the number of firms that a student can apply for. That will eventually dis-appear, and a lot more companies will have visibility to better students at the expense of a low joining rate.
InterviewStreet, with about 5% odds [R10], can shake up the way recruitment happens in India, and forge a fortune, and the change may well be, irreversible.
Footnotes
- R1: The closest analogy is Topcoder, who is a premium programmer attractor.
They don't teach the masses how to program. There are 3K+ reasonably
active low quality Indian participants, while the quality participants
are close to 20.
- R2 Thirdly, the original structure of interviewing etc, will still remain.
This will serve as a source of marketing, in two ways. Firstly premium
institutes will be engaged more than, the rest, since that's the
interviewer's choice skew. This engagement will create a constant
presence, in places with highly employable mass, which in turn are
extremely monetize-able. Secondly, The more important implication is,
It will create a brand for the firm, as others like to look up to it.
- R3 Wait a minute -- Even naukri does google ads!
- R4 There's no specific quality barrier. We know the numbers are larger towards lower quality candidates than at the top, and as any site expands to capture crowd, it would continually lose its bar. Rating users will go a big way. Branding that the earlier growth creates, if it's oriented towards quality, growth rate will be directly meaningful. Audience are more valuable.
- R5 I don't want to say, a "new player will build" instead of saying "interview street will build", as it gets too verbose and distracting.
- R6 As a direct result, unfortunately for both the users and the publisher, the same thing get's re-written. For example, Paul Graham's essays, talk repeatedly about a surprisingly small number of his strong views. There are a set of common principles that the author beleives in, that arenot very obvious and can create a spark to the first time a reader comes accross it. (Example, Creation of Weath, Market Value). The genius of republishing content is essentially telling the same thing at differrent angles and at differrent depths. TechCrunch has another work around for this problem. They just link to their old content, in the newer ones.
- R7 Quoting "Interviewers have short memories. A thank you note is your final chance
to stand apart from all of the others who want the same position."
- R8 One could say that buyer and sellers are here, trainers and trainees. Scalable training is a blue market, and arguably a hard problem. It might make sense to leave it alone and focus on one of the two domains. The truth is, they are heavily co-related and it can be a product differentiator.
- R9 RemindMeSam does that. It uses google contact list when you say Invite Friends on it's home page. He doesnt have to give you his password. He just has to say Grant Access, when google asks for permission.
- R10 5% are impressive odds for growing at this scale. The odds go much higher, if prominent (Sridhar Vembu; Rajesh Jain.) angels who sit for ideas but don't push you in directions are on your board. Google started off that way. The less obvious purpose of a board is that you report status to someone and forces you to think on a longer term. Also, saying what you did this week and what you plan to do next week, is a useful data to track - and pushes harder. Men like Rajesh Jain circumvented it, by blogging regularly.
A bunch of my class mates are venturing into: Interview Street. This article is my view of what I see of the opportunities in this space.
OK, so in short, Interview Street connects the interviewer and the interviewee for a virtual interview. Interview Street targets the Indian university students, as interviewees. Eligible seniors with less than 2 years of experience are targeted as interviewers.Disclaimer: I do not represent the team, and I have nothing to do with their direction. I'm no expert to write on this subject. I write this, to regroup for myself and sketch - the opinions about businesses I formed, mainly over reading online or meeting people. If you like the story below, -smile-.
Statement
Statement
A portal for people to interact with professionals from their industry in the form of virtual interviews so that they can,
- Get candidates prepared for the real interview.
- Share their experiences about their interview process.
- Provide guidance to people to hone their technical skills.
- Give pointers on what skills the industry expects out of a candidate.
- Bridge the gap between what is taught in colleges and what is expected by the companies.
Numbers
If Interview Street prices their services at Rs. 300 - Rs. 3000 per student for 10 interviews., at the higher end, it comes to Rs. 300 per interview.
If IS, takes at least a 20% cut on revenue, it passes on Rs. 280 to the interviewer per interview.
The wage rate is about Rs. 280 per hour.
Indian wage rates for a worker making Rs. 30,000 / month = 30K / (4 weeks * 40 hrs/ week) = Rs. 180 per hour.
Quality workers who are willing to work at this rate will be outliers, who want to outsource a very small amount of their time. Outliers will exist and will get crowded with interview requests. The quality and hence the number of interview request going to interviewers, will fall in this distribution, with accumulation peak on it's cream.
After a year, there will be about 20 (quality) interviewers taking 50% of the interviews, and that number will grow to 100 after the second year. [R1]
After a year, there will be about 20 (quality) interviewers taking 50% of the interviews, and that number will grow to 100 after the second year. [R1]
The most active interviewers will average 5 interviews a month.
The bottle neck will be on the interviewers.
Revenue:
Year 1: 20 * 2 * 20% * 5 * 300 = Rs. 12K / month
Year 2: 60K/month.
Growth
The revenues on interviewing are low. 1:1 engagement doesn't scale. Even at the high end of the charges, there's not enough trainers. But wait, users still have the problem you are trying to solve -- To get a good picture of the recruitment process in a firm. The more specific you can get, you add clarity.
Much of the differentiation is going to come, over the content. The biggest asset is not the money in user's pockets, but the users themselves. Firstly, to add value, to engage, and to scale, IS has to generate a lot of content. Much of this will be user generated. Secondly, much of the revenues are going to come from actually placing people. Advertisements will be a small addition to the revenue.
Before digging into the above two in detail, let's pause a little bit and bite the bullet. This means meeting the big players. [R2]
Players
I'm trying to put together a list of top players to get an idea of the market's existing state. In 2007, Naukri and TimesJobs were the leaders, with 2.3M and 1.3M users respectively. Growth rate is about 15-20K users/day. Which means the pie is growing at a 3X every year. Most fast growing markets end up accommodating multiple players, because there are so many non-customers to convert.
Filtering the list, by removing those who put Google Ads on their page, that leaves to about 10 players. [R3] There is no body who attract high quality crowds -- For instance, there's no one whom Microsoft or Google places an Ad on. [R4] Further the number of sites that specifically attract freshers, is less than 5. Even the moderate ones, seem to have a high number of audience, which shows the need for better solutions in this space.
It's time to go
1. Don't be shabby:
There are basics in web-site design optimization, that are not met in the the big player's websites. For example, Naukri clutters it's home page with all job listings. A great page, is simple and the parts that take the click are highlighted. Filled with carefully photo-shopped media, it briefs the same content redundantly, at differrent levels of shortness. This is no small deal. Monster does a better job, in their UI, than most people. The content clutter they dont seem to be able to avoid it as well.
2. Match - Dont Search:
Everyone either shamelessly classify and list all their openings for click navigation, or spend a decent effort to provide a search option. The list is either too large or too small. It's either too unclear or not so useful. It's ok if the list is too large, but the top results have to solve your problem. Again relevance is so poor, since it expects users to type everything down in a search box.
There's is nothing to search about jobs. Most options used for search are static, and must belong to a person's profile. Also job searching has very little choices to make. In most cases there exists a well defined ordering of job preference that holds good across a critical mass.
InterviewStreet will build custom web-apps, that are: [R5]
A. Recommendation engines that match user profiles to jobs, with high clarity of their initial report. Better targeting will improve precision, and form the theme of the game.
B. You apply to about 5 jobs and track, their status -- not a blank [applied] sign, something more user involving than that. He makes job specific notes, and book-marks, with tracking web-apps specifically designed for this purpose. This can be extended socially, with he being able to share it to people who study in the same college, class, etc.
C. Job targeted training or training recommendations will be offerred -- More on this later.
3. User generated content:
The web is about a constant user presence. Stickiness is important. People spend time in two ways on your page.
Firstly, and most commonly, they are in between something and open your site, just the way they open e-mail. To engage them, a good update rate on your quality content is necessary. User engagement rate is measured in number of articles published per week. Necessarily, the older ones get read lesser and lesser. [R6]
Secondly, people come when they are in need. They have an interview tomorrow, and want to look for historically relevant content. A well linked web of pages will do a great deal here. There are certain articles that cover the "head" of the domain specific problems. They will get heavily linked into. Content will be reached by links and equally often by classified navigations. The actual collection that the user first reaches will essentially be user generated. The tail is so large, and scaling will happen with user generated content.
There are no websites that do both of this. People who work like publishers today, have unworthy content with low update rates. [R7] Those who generate content from users, have an incredible lack of "structure". Everything is plain text, posted on forums.
There is need for structure and better apps. Eventually specific algorithms will 1. cluster, 2. classify and 3. and search this data pool. For a concrete example, when a user posts a new interview question that he came accross, it'd get matched with the existing pool, and the relevant questions will appear. He then can edit, with a simple interface and see if he can merge the two to mean, a complete question, adding value to the problem. No history is ever lost. In addition rich linkage structures across question papers are set, on top of which user generated apps can mine this data. For instance, it's now simple to determine, how many questions overlap across time in CTS interviews.
The vision is to make it so simple that, a student who just got an interview done, will be sent requests for a paper by his class mates, and instead of explaining it to all of them, he hosts it on your page (or) his own page, as add-on widgets. The motivation to host it online, is he get's to see some cool stats, as to how long people (a lot of people), took to solve the problem he solved. Added ease of distribution, with a feel of content ownership.
In the early days, for the first type of content, the founders can post upto one article a week, and get quality posters to write once a month. The second type of content needs a lot of importing todo. Semi-automated scripts, will import rich content by crawling other sites, and impose structure. Quality here is more important than quantity, since this sets the best case examples for later content added.
4. Scaling:
Any one who chooses the domain of a jobs and a training site, makes a trade-off between how much 1:1 human engagement you can commit, and how large a rate you want to grow at. There's nothing new in understanding that, human engagement doesnt scale. People like NIIT represent the left end and others like, Monster India chose the no-training path. But training and jobs go together. They cant be seperated. Either of them nurture the other. It's like making buyers and sellers meet, with online shopping. [R8]
The challenge is in providing scalable training. I was recently impressed by AppJet's efforts to create engaging tutorials. Web-seminars and recorded content goes a great way as well. Apps built around a rich repository of questions, that can auto recommend puzzles each time a user logs in, is a great way to engage. Auto evaluation systems that can score user generated problem-sets will emerge. A domain specific example, for most programming problems you can write engines to evaluate by running the program. The ease and simplicity with which ordinary users can create evaluators will determine how many sets are scorable.
The online interview meetings that they host, can be recorded and archived. The context information about the interview can be built around user profiles - again something recommendation engines can leverage.
Institute specific work shops will serve as starting points. (ofcourse that doesn't scale.) Facebook grew by dominating specific universites and opened up gradually.
5. Marketing:
Marketing trends in the web are changing and most players havent adopted yet. They still do news paper advertising as thier main reach. For web-sites, apart from the biggest form of advertesting, which is creating sticky users, social networks are big tools for growth. A new web-app model is necessary which can easily build apps into facebook, orkut, nokia, and run on web-sites as add-on widgets. Youtube and Slide grew immensely with that.
Google contact list is a great place to viral. Lets say, A new user who wants to register, if he lets your app to mail all his gmail contacts, and has greater than 20 contacts, gets a 10% off. [R9]
Opportunity
This space is fast growing and more players will eventually jump in. More mature and clean sites will have an edge, long term.
Most university placements happen through dedicated departments, that puts restriction on the number of firms that a student can apply for. That will eventually dis-appear, and a lot more companies will have visibility to better students at the expense of a low joining rate.
InterviewStreet, with about 5% odds [R10], can shake up the way recruitment happens in India, and forge a fortune, and the change may well be, irreversible.
Footnotes
- R1: The closest analogy is Topcoder, who is a premium programmer attractor.
They don't teach the masses how to program. There are 3K+ reasonably
active low quality Indian participants, while the quality participants
are close to 20.
- R2 Thirdly, the original structure of interviewing etc, will still remain.
This will serve as a source of marketing, in two ways. Firstly premium
institutes will be engaged more than, the rest, since that's the
interviewer's choice skew. This engagement will create a constant
presence, in places with highly employable mass, which in turn are
extremely monetize-able. Secondly, The more important implication is,
It will create a brand for the firm, as others like to look up to it.
- R3 Wait a minute -- Even naukri does google ads!
- R4 There's no specific quality barrier. We know the numbers are larger towards lower quality candidates than at the top, and as any site expands to capture crowd, it would continually lose its bar. Rating users will go a big way. Branding that the earlier growth creates, if it's oriented towards quality, growth rate will be directly meaningful. Audience are more valuable.
- R5 I don't want to say, a "new player will build" instead of saying "interview street will build", as it gets too verbose and distracting.
- R6 As a direct result, unfortunately for both the users and the publisher, the same thing get's re-written. For example, Paul Graham's essays, talk repeatedly about a surprisingly small number of his strong views. There are a set of common principles that the author beleives in, that arenot very obvious and can create a spark to the first time a reader comes accross it. (Example, Creation of Weath, Market Value). The genius of republishing content is essentially telling the same thing at differrent angles and at differrent depths. TechCrunch has another work around for this problem. They just link to their old content, in the newer ones.
- R7 Quoting "Interviewers have short memories. A thank you note is your final chance
to stand apart from all of the others who want the same position."
- R8 One could say that buyer and sellers are here, trainers and trainees. Scalable training is a blue market, and arguably a hard problem. It might make sense to leave it alone and focus on one of the two domains. The truth is, they are heavily co-related and it can be a product differentiator.
- R9 RemindMeSam does that. It uses google contact list when you say Invite Friends on it's home page. He doesnt have to give you his password. He just has to say Grant Access, when google asks for permission.
- R10 5% are impressive odds for growing at this scale. The odds go much higher, if prominent (Sridhar Vembu; Rajesh Jain.) angels who sit for ideas but don't push you in directions are on your board. Google started off that way. The less obvious purpose of a board is that you report status to someone and forces you to think on a longer term. Also, saying what you did this week and what you plan to do next week, is a useful data to track - and pushes harder. Men like Rajesh Jain circumvented it, by blogging regularly.
Comments
Post a Comment