Why our benefits products need to be like the Sainsbury supermarket

Introduction

 I was undertaking my weekly grocery shop in my local Sainsbury supermarket in Brentwood. It struck me that I have always shopped in this place. Why, and what lessons could be learnt that applies to reward offerings? The store gives me good value, consistency and familiarity.

 

Too often employee benefits are seen as an afterthought, an “add-on” to our pay strategy.  We can do much better than that. Employee benefits are an important signpost of both our employee proposition and organisational culture.  

 

Strong branding, good value, consistency

Sainsbury prides itself on its strong branding, good value and consistency. This is exactly what our benefit portfolio should be offering.  Our benefit products should reflect our organisational culture and values. Employee benefits provide a signpost to both employees and external stakeholders of organisational values. 

 

Good value is important. By which I mean the benefit is valued by the employee and drives business value creation by supporting and sustaining value added employee behaviours.  Consistency is also important. Employees should know what they are getting and benefit costs should be stable unless a revision is taking place. Consistency should also extend to culture. Giving a consistent and congruent benefits message supports and sustains our cultural memes

 

Product placement – the irritation factor

If you do your shopping like me, you are on autopilot; you know what you want and where everything is. However, every so often a ball comes from left field. The supermarket has moved my favourite jam. Panic.  I have to look around to see where it has been placed.  More often than not I see something I have not noticed before. Perhaps I will try a new jam? This again is like our benefit offering. On occasion we should move things around, shake them up a bit.  Just to get our employees thinking a little differently. 

 

Big data is watching you

I use a loyalty card at the supermarket. It is not the most streamlined of user processes.  However, it does mean that Sainsbury knows what I buy and when.  It can offer me money off coupons on things I normally buy and even tempt me with offers of things which I have not tried.  It knows what brand of aspirin I take, the stages of my children growing up and even how often I entertain It does not, in my view, make full use of this data; but it will.  These are valuable approaches we can use to enhance the personalisation of our benefit offering.  We can use the HR data to offer changes in risk benefit levels or pension and financial advice. All this is available and value added. It moves employee benefits from something staid to a dynamic, interesting, value added process supporting business success.

 

Visualisation and info graphics

I do not know, but if I was a betting man, I would place a wager that Sainsbury marketers and merchandisers use visualisation and info graphics to better understand the very large amount of data they hold. I can see in my imagination maps of the UK stores, showing geographic variations in value added products, perhaps with an Acorn (consumer classification) overlay. 

 

This is exactly what we should be doing with our employee benefits. Collect the data. Turn it in to meaningful visualisations.  Use these both as analytical tools but also to inspire creativity and innovation.  After all, why not?

 

Conclusion

There is a great deal to be said to applying the Sainsbury or Wal-Mart approach to our benefit products, services, and communication. The supermarkets make a great deal of money out of offering their customers what they want when they want it, at the optimum price. No doubt we can apply some of these lessons to benefits to build value, add credibility and build business success. 

Advertisements

Pay reviews; the Compa ratio magic. Strong Analytics V

Compa1

Introduction

We are, in most organisations, in pay and bonus round season. I have been involved in running pay and bonus rounds for over fifteen years.  One of the most helpful ratios and presentation tools is the compa ratio.  It is an incredibly powerful analytical tool.  At its most simple the compa ratio is the role is the position salary divided by the market salary.  This gives a ratio.  The magic is the amount of information contained in that number.  A compa ratio of 1 indicates that the position is paid at the market rate.  A ratio of less than one show the position is paid at less than the market rate and by what percentage and a ratio of more than one shows the position is over paid against the market and by what percentage.

By building graphs and visualisations of the compa ratios you have a powerful tool to assist management in making decisions on where to spend the limited salary increase resource.  Compa ratios can also be derived from total cash or even total compensation figures; although please see the methodological warning below.

What is it?

Most of us have salary data information from salary surveys.  We use this data to see how various positions sit in our labour market.   If I work in an insurance company I may have the excellent Mercer survey on insurance pay; if I work in banking I may very well use the methodologically sound McLagan survey.  Provided the jobs or roles have been correctly matched we will have a mass of market data on most of the roles in our organisation.  We will also have the average salaries for the same roles in our own organisation.

Here are some examples of comp ratio calculation:

Position salary Market salary Comp ratio
100,000 100,000 1 (Salary at the market position)
100,000  90,000 1.1 (Salary 10% above the market)
100,000 110,000 0.9 (Salary 10% below the market)

By using the simple compa ratio we will be able to see how our roles fit to the market.  Here is an example from a data set:

Role Average of Current Base Salary Average of Salary Compa
Actuary

$370,000

1.03

Management   Team

$370,000

1.03

Analytics analyst

$36,000

1.03

Analytics

$36,000

1.03

Analytics Manager

$100,000

1.01

Analytics

$100,000

1.01

Asst Trader

$47,648

0.94

Commodities

$41,603

0.95

EM

$29,347

0.96

OTC

$57,696

0.93

Special   Sits

$44,335

1.02

Treasury

$31,333

0.63

Here we have roles categorised by department with the compa ratio.  We can immediate see that there is an issue with the Assistant Trader role in Treasury.  At 0.63 we are clearly paying well below the market.  At best this warrants further investigation; at worse we have an immediate problem that should be prioritised in the pay increase distribution.   The concept becomes more powerful when we convert the data in to a graph

Image

In this example I have produced a graph showing both compa ratio and the attrition rate.  There is a strong negative correlation between compa ratio and attrition rate.

Getting clever

Using compa ratios it is possible to compare departments against one another as well as roles within a department.

Image

This shows the compa ratio by department; again illustrating where our pay round fire power should be concentrated.

The analysis can be extended to looking at sex discrimination, for example.  In this graph we look at the differences between males and females by compa ratio.

Compa3

This chart again gives an indication of areas that will require to be considered when carrying out the pay review.

Making connections

Another very useful application of compa ratios is to compare department compa ratios against a range of business analytics.  So, in the table below I have compared compa ratio with return on risk capital.  The concept is to focus our pay increases on to those areas that give the best return for the business.

Department Average of Salary Compa Average of RORC
Political Risk

0.93

32.00%

M&A Advise

1.00

28.00%

Treasury

0.89

18.00%

EM Debt

0.99

14.00%

Special Sits

1.00

14.00%

Derivatives

0.97

12.20%

Swaps

0.95

8.20%

OTC

0.95

7.40%

FX

0.95

7.23%

EM

0.96

5.50%

Vanilla

0.94

3.50%

Grand Total

0.95

11.60%

This approach shows a low correlation between market position and return on capital of 32%.  Depending on our reward strategy we may wish to focus our pay budget on, for example, Political Risk which has the top return on capital but has a compa ratio below one, showing we are paying, on average, below the rate for the market.

Thinking bigger

A similar approach can be taken when using a compa ratio for “total cash” – that is salary plus annual cash bonus.

A word of warning

I will talk of some of the methodological issues later in the article; but of particular note is that great care must be taken when looking at total cash market survey results.  Survey organisations use different methodologies so be sure you are comparing like with like in terms of cash bonus definition and the timing of the payment of the bonus.

Combining data

One of the most powerful ways to use total cash compa is to compare base salary compa, total cash compa and, for example performance ranking or even better, a business KPI to ensure alignment of bonus payments with outcomes.

A common reward strategy is to place salary at the median of the market place but to pay bonuses at the upper quartile, or better, for upper quartile performance.

He is an example of a table of salary compa ratio, total cash ratio and return of risk capital.

Compa4

This is a very powerful analytic graphic.  It shows that there is a major mismatch between the areas achieving the best return on risk capital and the market position for both salary and total cash.  It further shows that two areas with very similar RoRC have different compa ratios for both salary and total cash.

We can carry on with this type of analysis with almost any business metric and any mixture of KPI’s and compa ratios.  It is a really powerful way to think about pay and bonus analysis.

Methodological warning

A major consideration when thinking about this type of analysis is that salary survey data relates to positions, not individuals.  Further, accurate job matching is essential to ensure a good “fit” to the data.  Salary surveys are best viewed as not absolute numbers but as indicating relativities in the marketplace.  It is more important to look at the relative position of a role than the absolute salary level.  This is because roles are different between organisations as are the people who fill them.

To use the compa ratio approach well requires a good understanding of the statistical methodology underlying the raw numbers, it advantages and its limitations.  We need to understand both the size of the data population and its stability.  Even quite large populations used for data can cause issues if that population changes year on year.  This applies both to the organisations taking part in the survey as well as the roles and the individuals within the roles.    Survey data is averages of samples; good statistical approaches can ensure that the samples closely resemble the total population; but in many cases there are no more or less than a sub-set.

This applies still further when looking at total cash survey data.  The definition of total cash and the age of the data are essential consideration when manipulating the analytical outputs.

Conclusion

When we are analysing data in preparation for the pay round the compa ratio is a very powerful analytical tool.  Used effectively it can give a great deal of data in a simplified format that is amenable to graphs, diagrams and info graphics.

Used in conjunction with business data it can create meaningful business insights that will shape and direct the nature of the pay and bonus round in your organisation.

If you would like to understand more about data analytics and the pay round please contact me at idavidson@rewardresources.net