… and I have only recently understood the implications of this statement – I somehow happen to be more regular on my photoblog

But jokes apart, when I started the photoblog I wanted to follow the theme where I have a picture and I write an essay on it. However when I tried, I land up totally at loss of words. It felt that the picture tells its own story and it is better left for self-interpretation. So I’ve decided that I’ll have all my pictures on my photoblog and all my essays on this blog

One good thing that my newfound hobby has done is made me rise early on weekends for my experiments. I woke up at 5 am (really early for me) on a Sunday morning to be prepared for the following shot, taken with the rising sun.

The following shot was taken in full daylight.

I recently realized that I have gone without a real hobby for quite sometime now. I have always been interested in photography. So decided that this might be a good time to try my hand at the same.

The advent of digital photo equipments has made photography a lucrative hobby. It now has instant gratification, is inexpensive and great for experimentation, especially for a novice like me.

All I needed was a good digital camera, preferably an entry level dSLR (digital single lens reflex). An SLR because here what you see is what you get – WYSIWYG and it gives much better controls and hence a totally satisfying photography experience. Of course for a pro it means much more than just the “satisfying experience”. Anyway, after a little research I finally zeroed down on the Canon EOS 400D.

The following are some of my initial experiments with my dSLR. These are photos from my recent vacation at Dubai. These are just my baby steps into photography as a hobby, so the pro-eyes please excuse all the errors!

… some more photos

I hope I can stick to this new hobby of mine. Words of inspiration are most welcome!!

Let me re-start my blogging journey with a topic of utmost importance but which is very easy to overlook during the startup days. It is about building a legal shield for your company and your products. Any startup that is accelerating its way up the popularity charts is very likely to find various forms of legal speed breakers on its way up. If enough precautions have not been taken beforehand, the speed breakers can actually become major deterrents to the progress of the venture.

Never take anything for granted and be rationally paranoid !!

If you follow this simple philosophy you would be safe. This ensures that you are neither over-confident nor over-paranoid.

Before you go live with your startup a very minimum evaluation of everything from a legal perspective is essential to ensure that you are immunized from all potential legal attacks.

Some very basic steps would be as follows:

1. Seek advice from a good (experienced) legal counsel. This is very important, unless you are a qualified legal professional yourself. There are various aspects, holes and vulnerabilities that only an experienced legal eye would notice. But even more important, your counsel should have expertise and experience in the specific domain. Just like you wouldn’t go to an eye-surgeon for medical advice on a bad toothache, you wouldn’t go to a criminal lawyer for advice on trademark and copyright.
2. First protect your assets – get trademarks, copyrights and patents, whichever is applicable, for your product, technology, design, logo, names etc. Consult a good IP lawyer / law firm to get them done on day one itself.
3. Register an appropriate domain name. If you are an internet based startup most likely this is one of the your first steps. Even otherwise this is essential because the Internet is a way of life and you would want a web presence for your business at some point of time. This would also prevent others from cyber-squatting. It is better to register / buy a name exactly matching your product / company name and for at least the common and your country specific Top Level Domains (TLDs) – .com, .net, .org, .in, .co.in etc. You should host your site in the primary TLD (.com is what most users identify better) and redirect traffic coming to the others TLDs to your primary (unless you have clearly identified different things that you want to host at different TLDs).
4. Prepare the “terms and conditions”, “term of use”, “privacy policy” and other legal documents for your product and services. Ensure that your users / customers explicitly read and agree to them beforehand. The documents should be detailed enough and cover all areas of concern.
5. If you are doing monetary or other forms of transactions maintain detailed records of orders, transactions, receipts etc. Ensure that the terms, conditions and clauses are clearly specified and agreed to by all parties involved in a transaction.
6. Have an easy to find and comprehensive “About us” or “Contact details” section, where you provide correct information on how anyone can get in touch with you for their queries and concerns. And you or anyone from the respective departments of your company should be easily and immediately reachable at the contact medium published, especially for the legal contact.
7. Take the necessary permissions and agreements with the rightful owners if you are using content, software – pretty much anything that you don’t have the rights for use. If they are indispensable but require money which you can’t afford, then either negotiate a rev-share deal or raise money to afford them. If nothing works out, it is better to refrain from using them.
8. Ensure that you are shielded against personal liabilities. The very basic step here is to register your company as private limited.
9. Be completely clean, regular and careful with your company books and accounts. Be 100% compliant with all Government rules and regulations. Consult a good chartered accountant and a good company secretary as and when required.
10. Be extra careful with the statements that you make in press / public. You have to pick the right and unambiguous words. If required hire a PR agency to guide you.
11. Never indulge in anything that is unethical or which looks legally incorrect.
12. Avoid declaring or getting into any open war without adequate preparation (skill, strength and strategy).

Bottomline is – be legally correct. A legal hassle is the last thing that you want when you are starting up. It is time wasting and energy / resource draining. And if you don’t have the appropriate shields, it can be the biggest weapon that your opponents can use against you in business wars.

Contrary to the common belief, the first step should not be to invest hours .. days .. weeks in trying to write a concrete business plan. I don’t intend to devalue a bplan. It sure is necessary to help you identify the growth path, stay focused as you grow your business and to set achievable milestones to that desired goal. Needless to say, that a bplan is a must if you are going for raising any type of external funds. But what I mean to say is that the first step on startup, especially in the Internet space, shouldn’t be to run after funding.

It is really difficult to sell an idea! – Mainly because an idea in itself doesn’t quite have any value. It is all about the execution that matters. A good execution can do magic to even the most ordinary of ideas (e.g. gmail which totally changed the webmail ball game). So rather than proving that we have a brilliant idea what is more important is proving that we have a brilliant execution strategy.

The most common mistake that we all make is in believing that our idea is a kohinoor diamond – unique and exotic. When the biggest irony about unique ideas is that dozens of people have the same idea at the same time. Just that very few get past the thought process and even fewer embark on the implementation path.

The common fear is that when we discuss our idea with the outside world people might steal and implement it. When the reality is that ideas don’t always strike the same level of interest chord with everyone – half the people don’t care about it and the other half that do would not really bother to implement it.

Common fears about exposing one’s ideas:

1. Newbies fear big companies might take and implement it. In reality big companies are the ones to fear the least, mostly because they are too big to notice something till it starts making big enough ripples in the market. They have their own set agenda to worry about more. Besides the growth of a company is inversely proportional to its ability to innovate.
2. Newbies fear other startups might go after their idea. It is somewhat possible but unlikely. Startups that have got funding would be running full speed with their own idea because their investors will make them to. Startups that don’t have funding would be running after the potential investors with their own idea.
3. Newbies fear other wanna-be entrepreneurs might steal their idea. This is very much possible. Just that most of the wanna-bes won’t really get out from the wanna-be state. And even if someone does implement it and does it better than the original idea generator then it is he who deserves the idea more.

So discuss the idea with as many trustworthy people as possible. Such discussions are always educative and often are eye-openers on various fronts.

Unless you have good contacts with the-people-with-money who would be willing to fund your idea in good faith or unless you have proven track record in building a business or have strong IP rights (patents ideally) on a powerful and world changing idea, running after funding from day one is really a waste of time and energy.

Most important first step, hence, should be to achieve the proof of concept. This needn’t and shouldn’t be a complete implementation of the product. First identify what is the most important factor central to the success of your business and what are your strengths. Then build the p-o-c using your strengths around the central factor. If technology is the core, then building a minimalist demo of the product would be a p-o-c. If the idea is a research concept, a published white paper or a patent is a strong p-o-c. Sometimes design wire-frames / slides could be compelling enough as a p-o-c. If clients or business tie-ups is the success determiner, then having at least one willing client / partner could be a strong p-o-c demonstrator of salesmanship.

There are several advantages in conducting the p-o-c exercise:

1. It will make others see what you have visualized.
2. It will give you a good idea about the feasibility of the idea itself.
3. It will generate new thoughts on better ways of doing things.
4. It may even generate a totally new and probably better idea.
5. It will help raise funds, comparatively easily and assist at the negotiation table.

But ensure that p-o-c exercise is quick and the output is compelling enough before you pursue the idea to build a business.

Corporate world is the best example of the game of incentives. Compare the work culture in government offices in India vs that of the software industry. The productivity graphs of them are way different – rather productivity in the former is negligible compared to that of the latter.

In government offices – people come in late, leave early, take more frequent breaks, take longer breaks, work strictly in the self-interpreted office timing only and get work of about 1/4th man day done in a single day per person.

In software companies – well people do come in late (thanks to flexi-timing) but leave much later and work or think about work almost 80-90% of their awake time. An individual gets work of about 2 man days done in a single day.

So are people working in government offices any less capable of working hard than those in the software companies? Why is this behavioral difference?

The answer is in incentives. People working in government offices don’t have any clear incentive to work hard. They have job security – which is the biggest anti-incentive! Their promotion is not tied to their performance but age.

Lack of incentives is a productivity killer!

Now let’s come to the core topic of discussion.

Ambitious companies can’t avoid growth. Growth, not just in team sizes but also in product-line. Especially in Internet companies, where to get into the big league, you really have to diversify i.e. reach out to a larger audience with a wide spectrum of products of interest. Growth and diversity come packaged with the problem of synergies. The problem is with managing multiple teams and product groups to work together in harmony, exchange ideas, help each other and at the strategic level integrate products and cross sell.

So how can this problem be solved?

There are two solutions:

1. Superficial one: Directive from the top level management to the teams to synergize. This rarely works! It is more like the naughty students behaving themselves when the teacher is watching. Let the teacher turn away and hell break loose!
2. Deep rooted: This requires an understanding of why the problem occurs in the first place and fix at the source.

Why does it always take an extra effort to build synergies amongst teams in a big or growing organization?

Because there is no clear incentive for people and teams to synergize otherwise.

Let’s face it – No one is in a company for the love for it! They are all there with a reason which is almost always anything but love for the company. People work because they have a clear personal incentive to. They work hard .. harder .. to meet their own goals!

The individuals work hard to out perform their KRAs. The manager works hard to make his team over deliver. The business head works hard to make his business unit the top revenue generator for the company.

In this rat race no one has time or incentive to think for the performance of another person, team or business unit other than their own, unless this thought has been imposed on them a.k.a. a top management directive to integrate efforts.

What do I get by helping another? What do I get by taking initiative? What do I get by driving traffic to another product? What do I get by doing a job that isn’t mine? – are the questions that play somewhere in the subconscious mind when one is asked for an effort that lacks definite incentive.

So the root problem that acts as the deterrent to automatic and effortless synergy building is the lack of incentive.

The fix is simple – create incentives. This most easily can be achieved via concrete attribution schemes. Credit points for not just your own performance but also for that of others.

How often is there revenue sharing between different teams / business units in a company? How often in a person’s appraisal is she given a bonus based on the appraisal of a person whom she helped to perform better?

I feel, a company that can fix the incentive schemes and come up with concrete attribution mechanisms would automatically and effortlessly be able to build inter-personal and inter-group synergies.

I don’t want to get into a war front with a defend-my-company gimmick here. What I present is my perspective on this discussion from as neutral a viewpoint as possible. Comments and course correction most welcome!

First, I would like to recommend everyone to read Free Prize Inside by Seth Godin, an awesome book on the new age soft product innovations and marketing. In fact most of Godin’s books are must reads.

As Indian, we have always been trained to look for a free prize in everything. As such things like “Buy two get one free”, “Free X with Y”, “50% off” capture and hold our attention much more than statements about the quality of a product like “100% fat free”. It is usually with the high ticket items that we really give a serious thought about quality and other aspects. How often have you asked (or at least tempted to) “what’s free with this” or something to that effect?

Free prize marketing has always worked in India. Every company does that at some point of time. Every company also exploits that at some point of time. Consumers fall for it every time. The gimmicks like “25% extra on a soap bar at just Rs 5 more” do work.

Consumers run after the free prize even when it is elusive i.e. there is no guarantee of getting one. Lotteries work everywhere, be it in form of the sms answer to a dumb question in a game show or the actual money gamble. Consumers would rather buy a detergent where they have a remote chance of winning a free gold coin. They would be willing to pay a premium for that detergent and even buy it multiple times.

Essentially free prize generates attention and engagement on a product.

Now let’s focus our attention to the internet products. There is a classic rule – most web products have 90% users as consumers and only 10% users as contributors. This ratio is probably worse for a new product. Products with high user contributions are engaging products. Most new products, with user-generated contents, face the chicken and egg problem – existing contents influence contribution from new users and initial contents cannot be generated without users’ contribution.

The various options for user acquisition, irrespective of the budget size are:

• Media ads (TV, Print, Radio, Internet banners) – The most expensive and usually the least effective.
• SEM – Moderately expensive and moderately effective.
• SEO – Least expensive and moderate – highly effective, depending on the keywords optimized for, but is a slow process.
• Word of Mouth – Most effective and can be zero cost to highly expensive, depending on the channels.

The effectiveness mentioned above is the ratio of the money spent to the number of active users acquired.

The next and bigger problems in a web product after acquisition are retention and engagement. Retention and engagement are even more difficult when the users aren’t accustomed to the product or the product idea. Blogging is still a new concept to most Indian netizens (I am talking about the average netizens and not the savviest of the lot). And even quite a lot of savvy internet users aren’t bloggers – because not everyone enjoys writing. Also original content generation is difficult without incentives. Let’s not forget the pains of typing for someone not skilled in the same.

Under such circumstances, a blogging platform targeted at the mass Indian internet users (and not just the savvy bloggers or journalists) requires very clear and immediate retention and engagement mechanisms. What better incentive than ability to make money in lieu of the effort invested to understand and engage with the product?

Then why not reward only the best and quality content generators? Now the incentive mechanism defeats the purpose if it is tuned to favor only the best few. The mass would shy away if the incentive scheme is too complex or requires too much of effort for the return or is designed only for the best men to win. Some of the best monetary incentives on the web (Google Adsense and Amazon Affiliates) are very simple and low effort mechanisms of making money. Even most of the current contests in the mass media world are designed simple, intentionally, to achieve maximum participation and not quite to identify the talented one.

What about spam? Whenever there is money involved there would be spams. Even adsense and affiliate schemes are subjected to spam (dummy sites). Besides, blogs originally started as web diaries, is a medium of self expression and a record of things of one’s interest. So whatever an individual collects, be it for whichever reason, cannot really be called spam! I would rather call the outburst of contents on ibibo.com as content explosion due to high levels of user engagement. That’s where the various search mechanisms become useful to find the right content of one’s interest from the ocean.

When Ibibo launched in the Indian market, it had the mammoth task to first appeal to the mass and mostly non-savvy Indian net users and engage them on a product, which they are not really familiar with, and convert them into contributors than consumers. For that to happen there had to be a clear and easy incentive for the users. I feel GIBH achieved that for Ibibo, easily and in a short time.

GIBH, I feel, is the free prize on Ibibo!

Last but not the least – once someone earns a free prize (especially money) from a product she becomes a bit more loyal and engages more with it. Once a person engages with a product for a long time, it becomes a part of life!

Here are my thoughts on the factors that are prompting web startups in India. These are in no particular order.

1. Web enthusiasts working in some technology area, either in India or abroad (mostly the latter) have seen how web products have eased the life of people (and also the founders) in countries where the Internet penetration is high. They are charged up to replicate the same success in India. This also gives these technologists a lucrative enough reason to return back to India.
2. PC / Laptop and broadband prices heading south (but increasing bandwidths) and more providers offering them. With these, the internet penetration is increasing rapidly in this country where the scope for growth is immense.
3. Most offices (even govt. ones) are computerized and quite a few of them have ready access to the Net. As such it isn’t just the youngsters who are cyber savvy but the middle aged and the older generations are increasingly getting hooked to the Internet.
4. Most service providers now have useful and functional websites. As such the general awareness about the Internet is higher. Even newspapers are available online now.
5. The big boost to the Indian Web came from the travel sector. It started with the Indiatimes auction of air tickets, boosted by irctc offering railway ticket booking online and by air deccan’s move of having their booking available primarily via their website. Now we have every – airline – offering – ticket – booking – online and the - myriads – of – travel – portals that have cropped up are adding fuel to the fire. All these have provided stronger reasons to a lay user to learn using the computer, get familiar with web browsing, get an internet connection, get a credit card and be confident about spending online.
6. Internet banking and its convenience offered by almost all banks. Ability to do quick account lookups and even transactions like money transfer, bill payments and stock market trading – all sitting at home rather than sweating in the bank queues is definitely compelling enough for most people to get Net-familiar.
7. Start of formal forums like IAMAI which promotes the Internet and Mobile sectors and TiE which promotes entrepreneurship.
8. Ample seemingly available Venture Capital money with both global VC firms who have started their India presence with dedicated India specific funds and also Indian VCs and seed funds who have raised money locally.
9. Cheaper cost of startup. Hosting charges are cheap usually when hosted outside India. (Good options to start with – Dreamhost shared hosting, GoDaddy virtual dedicated servers)
10. Google Adsense. Immediate revenue stream usually good enough to recover, at least, the hosting cost. Btw the sky is the limit, if Adsense clicks.
11. Easy and low cost marketing channels. Good word of mouth marketing is done by the various – tech – and - business – bloggers who closely follow the Web startups. SEO and SEM strategies with firms to do that for you, who work on result oriented payment models. Also there are firms to help with buzz marketing, again working on result oriented compensation.
12. Easy and rapid web development frameworks like Rails, Django, Symfony, Catalyst etc. which make launching an ace web-app a breeze.
13. Me too! – Indian mentality. Launch a web product because everyone else is doing it.

Any other? I’m sure there are!!

**Update: [via Contentsutra] According to the recent stats published by comscore.com the Indian internet audience grew 33% between Jan ’06 and Jan ’07. This is the highest growth rate recorded worldwide. According to this report, India had more than 21 million internet users above the age of 15 years by Jan ’07.

Whatever be the reason, the growth of Internet startups and the buzz around it is a good thing. It is helping in product arbitrage and innovation for the Indian market. The competition makes people do things better. The consumer benefits from the multiple useful and usable products that create greater levels of convenience. Needless to say it helps the economy too.

But data needs to be viewed with the right lens or it can be very misleading too.

Data analysis is a research approach and not a mechanical approach. So we will refer to it as Data Research here. People commonly confuse the web analytics tools and report generation process as data research. These actually are tools that can aid the process, but they themselves don’t constitute data research. Often data research is also confused with data mining. Data research requires using data mining tools and techniques to slice and dice data.

What is Data Research?

Data research is about finding a practical solution to a problem or finding plausible explanation to an unexplained phenomenon by digging insights from the available data using various analytical tools, algorithms and techniques combined with domain knowledge.

Data Research Process

Data Research is research. As such all the rules of a research process apply here, which effectively means that there is no single and absolute way to do data research. The process pretty much varies from person to person. But there are few things that remain constant in any form of research that you do.

1. Domain knowledge is a must. You have to know your area of research inside-out to be able to conduct a fruitful research.
2. Experiment a lot. Research is about experimenting. You should be prepared to try out several different things quickly.
3. Fail early. You should have the intuition to know when to call an experiment quits and try in a different direction. Otherwise it is very easy to get lost and far.
4. Positive attitude. Failures are integral part of research. 9/10 experiments fail. And the 1 that succeeds makes a person famous. However you need to have the right attitude to absorb failures the positive way, learn from them and quickly jump start in a different direction.

Dive into Data Research

From my experience with data research, the following is an approach that I found to be most useful.

The steps towards an effective data research are:

i) Start with a clearly defined problem area.

Have a clear problem statement before you start any research work. This gives some kind of a direction to your research. Otherwise it is very easy to go astray. You cannot really start by saying “Let’s evaluate Google search and see what we can find in there”, because you wouldn’t know where to start and what to look for. Have the problem statement a little more concretely defined like – “Are users finding Google search useful?”

Of course the problem statement need not be rigid and it can change mid way based on your findings from data. You should be flexible enough to quickly change gears when a more lucrative research direction crops up. But at any given point, pick only the low hanging ripe fruit for research i.e. it should be an important and urgent problem from the business perspective with the prospects of coming up with insights that can make big and positive impact either on the user base – AREMs: Acquisition, Retention, Engagement and Monetization – or on the business: increase in revenue or decrease in cost or optimizing operations.

ii) List all possible questions in the problem area.

Once you have identified the problem area list all possibles questions that come to your mind in that area. Push your imagination. The more questions that you have, the more fulfilling will your research be. Get the questions from as many people as possible – domain experts and novices equally. You never know where a useful suggestion may come from.

When you have a comprehensive list of question, group the related questions together. The grouping can be of questions that will require a common data source to answer or which are sequential i.e. the answer of one will help advance the quest for the answer to another.

Prioritize the questions based on their importance for the end objective, availability of data, complexity of analysis and clarity. The first steps are always important and sensitive. So ensure that you pick an easy question to answer first.

But question generation isn’t a one time process. You cannot probably collect all possible questions at day one. So keep adding to the list and repeat the grouping and prioritization every now and then. In fact you would have a lot more questions as you proceed with your research.

Examples of a few possible questions for the problem area – Are users finding Google search useful?

• How many result clicks happen per search?
• How many searches on Google go without a click?
• What are the user activities in a session where there is a search but no click?
• What is the distribution of the length (in time) of a search session?
• What percentage of users revisit search in – a day, the next day, a week, a month?
  ….. etc.

iii) Identify data sources and meta data

Now that you have the specific list of questions, the next big step is to map the questions to the data requirements and identify the sources of data. The data sources can be web server logs, market research data, transaction records, data from point of purchase etc.

You have to do a feasibility study of whether the data is a) available b) accessible c) usable

Define the meta data that you need to track to find answers to the questions. The meta data can be basic ones like timestamp, pageviews, clicks etc and composite ones like CTR ( clicks / views) etc, which can be computed from the available data sources.

Always start by analyzing the simple and easy to understand meta data first.

iv) Get comfortable with the data

Before you start your analysis, the most important thing is to get a feel of the data. The best way to achieve this is to eye-ball the data randomly. This will give an initial idea of the data quality and requisites for the following:

• Cleanups
• Filtering
• Transformation
• Sampling

Give yourself sometime to browse through and visually play with the data. You may even get some new ideas on how to analyze them.

v) Analyze

Extraction of information from data is what data analysis is all about. This is when the fun really begins! For dataholics there can be no bigger kick than unraveling the mysteries stored in the ocean of data.

The broad phases of data analysis are:

1. Cleanup, Filter and Transform: Remove the incomplete and spurious elements from the data set. Also perform necessary transform required before proceeding with your research. e.g. transformation could be date / time normalization.
2. Sampling: Running analysis over the entire set of data can be very expensive operation, especially when the data volumes run into tera-bytes (usually the case in server logs of a popular web application). Hence analysis should be done on a smaller yet representative sample of the data. One of the keys to a successful analysis is to be able to come with the best sampling function – which does more or less equal representation from all possible sets in data and generates a sample size which is statistically significant.
   e.g. suppose we are analyzing survey data to determine the color preference of people. Here gender can be a bias (usually men prefer blue and women prefer pink). So we have to ensure that we have equal representation of responses from both the gender groups.
3. Slice Dice Aggregate Segregate: This is the computation phase where you use different analytics and data mining tools, models and techniques or even write your own scripts (I prefer doing this) to generate meaningful numbers from data. A few recommendations here:
   • Don’t only look at averages. They can often be misleading. Look at the entire distribution. e.g. you can have 50 people spending Rs 10 each and 5 people spending Rs 1000 each. Average spend of the user base will give you Rs 100 which isn’t quite the true picture.
   • Remove outliers. Outliers are the exceptions in the data. It is always recommended to remove the exceptions from your analysis unless you have a strong requirement for having them (like calculating totals etc). In the above example data, the 5 people spending Rs 1000 each are exceptions.
   • Segment aggressively. Look at as many logical fine-grained groups as possible. The view is much different from 10 ft than from 30000 ft. e.g. In the above ‘spend’ example, you would want to do separate analysis for the group which spends Rs 10 and the one which spends Rs 1000. The comparison might have some clues on how to convert Rs 10 group to Rs 1000 group. Even within the Rs 10 group you may want to segment them based on their spending patterns.

4. Validate and Triangulate: When you are analyzing large volumes of data at different levels and different directions, it is easier to make errors in assumptions and calculations. A single wrong calculation can render the entire analysis useless. Hence it is very important to validate and triangulate your findings – if there are two possible ways of calculating a metric, you should calculate both ways and both should yield the same result.
5. Extract Insights: Any piece of data has something interesting to show. You should be able to conclude something after every analysis. The conclusion need not always be an answer but can also be more questions to explore. If you find nothing - look hard! Look at the data multiple times. Combine multiple analysis and domain expertise to draw appropriate conclusions from the various data points. Repeat your experiment, if required, by slicing dicing aggregating or segregating differently. If you still don’t see anything either you are doing something wrong or what the piece of data is showing is not significant for the problem area to capture your attention. In either case prune your analysis path and try with something different. You can revisit later if required.
6. Next Steps: Irrespective of whether you are able to extract quality data or not you always have to be ready with the next steps, in your research, till you arrive at your desired goal.

** Always remember to document your questions, action steps and findings in data analysis. You can never hope to keep everything in your head. Besides, something that is not immediately obvious in the first look can become visible when you look at it later at leisure.

vi) Present your findings

Your research is meaningful only when you are able to convey it, aptly, to others. This is done usually through presentations. But presentations shouldn’t always happen only at the conclusion of the entire research effort. You should try to make intermediate presentations in small groups. Often others can see or question things that may have slipped your mind. Such presentations will help you fine-tune your analysis better.

Presentation in itself is a skill. I don’t consider myself an expert in this arena. I’m still learning. But following are a few things that I have learnt over time:

• Establish the context: Only you, who have lived with the data for a long time, will know it inside out. So when you are presenting, it is very important that you understand the aptitude of the audience in the domain and establish the right context with as much of background information as required.
• Focus on insights than numbers: Since numbers are what have led us to the insights, it is very tempting as a researcher to fill the slides with numbers and to explain the micro-steps that led to the conclusion. However too many numbers can get very confusing to the audience who weren’t a part of the research. Just restrict your slides to highlight the primary findings with only the bare necessary numbers to support your arguments.
• Use fewer and shorter statements: In your slides use short statements. Try to restrict the bullet points to 3-4 per slide. These make it easy for someone to parse them during the presentation. Too much of information distract the audience.

These are some quick notes on an approach for Data Research. As I’ve emphasized earlier, there is no one defined way of doing research. You should adopt whatever strategy makes you feel comfortable and motivated. As these are two key traits of a successful researcher. But constantly evaluate your progress on whether it is headed in the right direction.

Experiment! Experiment! Experiment!

One thing is for sure – data is addictive and data research is an addiction! .. so just dive into it!

In the software world and more so in the internet world there are two primary and more or less dis-joint groups of people -
a) product people (experienced and knowledgeable hi-tech group building a product)
b) average-joe users (the novice mass who use the product)

The former is probably the top 5% of the population and latter is bottom 95% of the population. There are antipodal differences in thoughts and perception between the two groups, when it comes to understanding the product. The problem, when a product fails to deliver the desired results, is almost always because of the disconnect between the perceptions of the developers and the users. Users don’t always see what the product people can visualize.

For a person who has seen products and technology from inside-out, it is very difficult to sedate the grey cells and turn one’s back to the multiple super-cool geeky ideas. It is really difficult to think like someone one is not. Things which seem obvious to a product person may not be obvious at all to an average-joe. To give an example – in the earlier version of our job search engine product, Bixee, we had the usage examples below the search box as hyperlinks. The idea was that, the users can click on a link and instantly see the example of a search.

However we found that users would specify Job Keywords like “software engineer” and then click on the example link “Java”. They were using the example link, as a potential refiner for the keywords they specified in the search box. They would then get confused when the results weren’t as expected.

So we had to remove the hyperlinks from the examples.

So the bigger questions now are:

How does one build products whose functionality is obvious to even a novice user? How can one know what a user thinks?

The answer, of course, is data!

A common practice of data collection is to conduct user surveys and closed room usability tests. I wouldn’t deny the effectiveness of such tests. In fact we discovered the problem with Bixee usability, as mentioned above, from a similar experiment. But the problem, largely, with such surveys is that most people tend to behave differently when they are being observed. People are always alert in surveys. As such survey results don’t always reflect the user’s real online behavior. Example – between the options of visiting a news site and a “fun” site, the probability that the user chooses the former is much higher under observation than otherwise.

You can get the true picture of usability only from the real usage data of users interacting with the product in their natural environment. Essentially, in the Internet context, this means looking at the web server logs of a production machine.

However before a product is launched you cannot have real user data. In such scenarios, the way to go about data-driven product development then is to design the product with limited interactions defined, based on market research, study of related products and intuition. Then release, as alpha, to a limited group of randomly picked real users on invite only basis. Get on your invitee list a couple of web critics too. Based on their usage patterns, fix the holes and if required redesign the whole product. Since these group of users were invited they would feel privileged and evaluate your product sincerely. You would get some really useful suggestions too. But ensure that you highlight and maintain user anonymity and let them evaluate at ease. Otherwise you end up with the same problem as that with surveys.
We had launched our photo bookmarking site, Pixrat.com, by the same approach. Just that instead of a limited group of users we had launched it for all users (which is not a recommended thing to do). You can see what magic analysis of usage data and user feedback can do to a product in the following:

Pixrat at initial launch:

Pixrat after data analysis and user feedback:

The important factors in data analysis are correctness and completeness of the data views. A wrong/incomplete view of data can often lead to a big fiasco. Example – During the days of early browser wars, Netscape was happy because data indicated that its market share was consistent when in reality IE was gaining grounds in the browser market. What Netscape failed to look into were the data points that indicated that the same user was using both IE and Netscape and in fact using IE in more browse sessions.

Bottom line is that data is the most powerful tool when it comes to building products that appeal to users. Data is the best story teller about how users are interacting with the product. Using data the right way is an important key to building a successful product.

I would be sharing certain tips for conducting user centric data research, based on my experience with data research, in a later post.

“If your father is poor, that is by fate. But if your father-in-law is poor that is your stupidity”.

“Shopping is more fun than marriage. If you don’t like what you have got you can exchange it for something better”.

“See No Evil! Hear No Evil! Speak No Evil!. Thank God that some monkey didn’t say Do No Evil! So let’s enjoy life fully”.

“A woman shouldn’t marry the perfect man. She should make her man perfect after marriage.”

“My advice to girls is that keep your eyes fully open before marriage and half shut after marriage”.

They are quotes of someone called Vamshi. I casually asked the person at the cash counter about Vamshi. But he appeared completely clueless. So I didn’t pursue further.