How we found the best companies to work for

Why do we work for the company we work for? Is it because the company gives us the highest salary we can command? And a career profile that best suits our abilities? And a career prospect that best fits our aspirations? And training and mentorship to enhance our skills and confidence?…the list can go on and on. In reality, getting all of these in one company would mean getting a dream job. That may be an unrealistic target to set for ourselves.

But we can surely aim for a good job. For the past eight years Business Today, along with survey and knowledge partners, have been bringing you an annual listing of India's Best Companies To Work For. The exercise goes much beyond just the listings. In the process of arriving at our rankings we have also tried to highlight and analyse the best HR practices and how they have evolved in India. This year's survey takes a huge leap in this direction—it not only captures a much larger number of companies, but also reveals several facets of HR as it is practised by India Inc.

The Old and The New Earlier Company participation was voluntary—they could drop out if they wished. The rankings were based on weights across factors like HR processes, views of stakeholders and employees. Company universe not large enough to arrive at more than 10 Best Companies To Work For. Now No direct company participation sought. Only employee participation obtained. Rankings are based on what employees think about employers—present, past and future. Universe big enough to arrive at 25 best companies and some sector-specific best companies.

The superior outcome of this year's survey wasn't exactly in our minds when we first thought of drastically overhauling the methodology sometime in the middle of 2009. That we should be reviewing some of our processes had occurred to us while concluding our last year's survey that appeared in BT dated January 25, 2009. While reporting on companies we came across certain discrepancies in the data that companies had provided during the course of the survey and what they were sharing with us after the survey results were ready.

It wasn't that companies were misleading us through gold plating of data, but due to regional and inter-temporal variations in ratios like attrition rate, promotion ratio and training budget, the data recorded in the survey and that in the reporting were beginning to differ. While we could have found ways to fix this, the occurrence of variations made us wonder if there is a totally different way to arriving at The Best Companies To Work For.

A new way that not only helps address the discrepancies, but also settles some other outstanding issues like:

• Enlarging the scope of the study to include more companies.
• Identifying sector-specific best employers, in addition to the overall best employers.
• Getting out of the participation route, which exposes the study to a situation where a company can simply opt out.
• Getting more "public" data—in the past companies would share lots of information only if we promise not to publish.

To address all these, and some more issues, the consensus was to make the study a "by the employee, of the employee" kind. That is, instead of reaching out to companies, we reach out to employees directly, asking them about the best employers based on a structured questionnaire. We could then also ask them about what they think constitutes a good HR practice, what would make them quit their existing job (both the pull and the push factors), which company according to them is India's best employer (not only in their sector, but also overall). This way, among other things, we would be able to test the true HR equity of a company across the entire employee landscape. And, after all, a company's HR is only as good as its current, past and future (potential) employees think it to be.

But setting on this course would require committing to certain big tasks. Like reaching out to as large a set of employee base as is possible (we achieved close to 9,000); designing a questionnaire that captures all the nuances of HR, and yet is short and simple enough for maximum number of people to easily respond to; using a technology platform that allows us to reach out to people across regions, age, professions, functions; monitoring the survey process to ensure that dummy and motivated responses are not filled in that would skew results.All this meant choosing a survey partner that has expertise and demonstrated comfort in conducting such surveys. We chose Indicus Analytics, a partner in several studies with our group's flagship publication India Today, as also with BT. For the design of the questionnaire and to help us interpret the results, we partnered with PeopleStrong, an HR consultancy.

The results of this year's study match most of the expectations we had. We have more companies, we have some of India's biggest employers and we have made a beginning towards sector-specific best employers. While details of the results and philosophy of the study have been dealt with elsewhere in this package of cover stories, a few more words on our approach to the study are in order.

The Approach: Internet was chosen as the media for the survey. This was because online research is known to get more honest opinions, as the employee reveals his views without the company "watching over his/her shoulder". The employee reveals preferences about the work environment in his own company, and also of other companies that he knows about, either from past experience, or from the experiences of his colleagues. In addition, tie-ups with rediff.com, advertisements with Google.com, as well as other online entities such as Facebook, LinkedIn and Twitter were utilised for ensuring a wide coverage.

To ensure data authenticity, a threepronged check was done. One, at the questionnaire level itself several inter-connected questions were incorporated, and contradictions were identified. Such responses were rejected. Two, a verification back-check of about 30 per cent of the accepted responses was carried out by calling up respondents and re-administering part of the questionnaire.

Additionally, on analysing the data, cases of collusion (respondents of a single organisation attempting to get their organisations up on rankings) were identified based on comparable responses by persons working in the same industry and rejected if highly divergent opinions were received. One case was found where employees at a maufacturer colluded to give higher rankings. This company was excluded from this year's survey. The overall response rejection was under 5 per cent. Only those dimensions are being reported for which sample sizes were significant (at least 150).

If you like statistics Use of standard multiple regression to arrive at weights: The idea behind standard multiple regression is the same as simple linear regression (Y=a+bx), except that you have several independent variables (typically X1 to Xn instead of just X) predicting the dependent variable (Y). In our context we wanted to predict dependent variable "overall satisfaction" (Y) from independent variables like "career and personal growth"(x1), "Prestige /Company reputation"(x2), "Training/ Coaching/ Mentoring"(x3), "Financial compensation and benefits"(x4), "Good job content (efficient and less stressful)" (x5) and "Merit-based performance evaluation"(x6). The equation is built by using all the rating scores given by all the respondents for each of the above variables. Our equation is Y=a+bx1+cx2+dx3+ex5+fx6. How much of the variance of "Overall Satisfaction" was accounted for by the joint predictive power of knowing a person's satisfaction with all the 6 independent areas of employee satisfaction? This value is denoted by R Square, which was 0.678. Best model fits are in the same range of 0.7 to 0.8. Does the model allow you to predict "overall satisfaction" at a rate better than chance? This is denoted by the significance level of the overall F of the model, in our case 0.005. If the significance is .05 (or less), then the model is considered significant. In other words, there is only a 5 in a 100 chance (or less) that there really is not a relationship between "overall satisfaction" and other independent variables.

Data Analysis: The survey questionnaire had two parts, the main survey and the demographic data. Using the two in conjunction, various slices of specific segments were constructed. Standard accepted statistical analysis tools were used for analysis. The key analysis (that of importance of 6 parameters), which involved converting importance rating to relative weights is described in the box If You Like Statistics. In step two of the analysis (rating of own company), scores were converted into linear scales ranging from very low to very high.

In case of ranking of companies, a single company was to be nominated by a respondent and percentage nominations within a segment was the basis for ranking. In case of factors that attract and factors that cause dissatisfaction, respondents were asked to rank the top 3. The overall ranking was computed by assigning higher weights to higher ranks. The analysis was sliced for various segments of respondents—age, experience, geography, gender, functional area, size of company worked for, type of company worked for, industry sectors, salary levels etc.—a total of nearly 80 segments with significant samples.

The data this year's survey generated was so much that we aren't sure if we have done full justice to it. But this is only our ninth year (BT'S Best Employer survey is India's oldest). Sure that this new methodology is the way forward, next year we will return with an even bigger and more comprehensive survey, which will reach even blue-collared employees. If you still have doubts and questions on our methodology we will be happy to address them. Just write to us on bt@intoday.com Also, do log on to www.businesstoday.in for further details on the new methodology and a look at previous years' methodology.

WHAT WE ASKED Respondents were asked the following distinct sets of questions: Rate the company you work for on six key factors and 18 supplementary factors (The Six Key Factors). Career and personal growth Prestige/ Company reputation Training/ Coaching/ Mentoring Financial compensation and benefits Good job content Merit-based performance evaluation 18 supplementary factors that further elaborated on the above six factors. Factors that attract employees What are the key factors in a new job offer that would make you change your current job? There were 9 factors and each respondent selected the number 1, 2 and 3. All other factors were unranked. Factors that cause discontent What are the key factors in the current job which can cause you to search for new jobs? There were 9 factors again and each respondent selected the number 1, 2 and 3. All other factors were unranked. Would you recommend your company as a good place to work The statement "I would recommend our company as a place to work in to people who trust me for advice" was posed to the respondents. They were then asked to respond— as Very Strongly, Strongly, Yes, To some people, No. Rank the Best Company to work for—Overall Respondents were asked to nominate a single company or organisation across industries as the best place to work in. Ranking the Best Company to work for—within your industry Respondents were asked to nominate a single company/organisation within their industries as the best place to work in. WHO WE ASKED A total of 8,742 respondents participated in the survey. Demographic composition: 800 Urban Centres 86 per cent Male 72 per cent Chief Wage Earners of Households 28 per cent >33 Years, 48 per cent 19 per cent PGD, MBA, PhD, Professional 12 per cent Non-Graduate, 69 per cent Graduate 40 per cent with Children, 43 per cent Never Married 15 per cent Mumbai, Mumbai suburbs and Pune 14 per cent NCR, 24 per cent from Pune, Chennai and Hyderabad 16 per cent Senior Management, 32 per cent  Middle Management 32 per cent >10 Years’ Experience, 19 per cent 4.4 per cent Salary > 12 L, 63 per cent 27 per cent from MNC, 40 per cent  Indian Private Sector 14 per cent PSU, 7.5  per cent  Govt Over 1,000 organisations Sector composition: 10.2 per cent — BFSI 14.2 per cent — Softeare 6.7 per cent — Education, training and consulting 9.3 per cen — ITES 10.3 per cent — Service 17.3 per cent — Manufacturing 5.7 per cent — Core sector 4.4 per cent — Telecom 4.2 per cent — Health 17.1 per cent — Others HOW WE ASKED 80 per cent from online panel of over 100,000 regularly tested and authenticated respondents 20 per cent from open promotions 30 per cent back checks and verification PROMOTIONS E-mails to companies Social Media: Twitter, Facebook, LinkedIn, others BT magazine and web site SMS alerts Regular panel mobilisation Incentives — The survey was open to participation from September 15 to November 30, 2009