Imagine you have just sampled a truly marvellous cake and want to know exactly what went into that cake and in what quantity. How would you go about working this out if you didn’t have access to the recipe?
Enter econometrics: a statistical, correlation based methodology that allows you to do just that, except this time sales are your ‘cake’ and factors such as store location, in-store marketing, sales visits, pricing etc are your ingredients.
So, just how does econometrics allow you to reverse engineer and get the ‘recipe’ for your sales?
The answer lies in the way in which econometrics correlates sales with other variables thought to influence sales. Let’s take a simple example: assume you have a single store and you want to determine if sales within that store are influenced by merchandising activity. In this case, an econometric model will allow you to compare the movements in sales over time for that store with the amount of visits and level of compliance achieved when the sales representative was in store. Should sales go up when compliance goes up, then the model will identify a positive correlation between the two and determine the rate at which sales increase as visits or compliance increases.
Fairly obviously, such a simple model is fraught with problems, the most obvious of which is one of how we are sure the sales increase is down to the merchandising activity and not something else. In this case, this is where econometrics comes into its own, as it provides a framework to not only correlate a single factor with sales at any given time, but compare multiple factors with sales at any given time so we can be sure that any increase in sales assigned to a particular factor is down to that factor and not something else.
At Retail Alchemy, we take the standard approach to econometrics listed above and enhance it in a number of ways.
Firstly, rather than model each store in a dataset on an individual basis, we are able to model large groups, or indeed all of your stores simultaneously by employing a technique known as ‘pooling’. Aside from the obvious time and efficiency advantages this generates, pooling also allows us to understand why average store sales in one location differ from those at another location meaning, in addition to explaining fluctuations in store sales over time, we can also understand what makes that store perform in the way that it does relative to its peers.
Such a technique therefore allows us to understand the impact that a whole host of locational and in-store factors make on store performance, enabling you, ultimately to isolate the impact your in-store marketing activity has on sales.
Secondly, through our years of experience in the industry, we have become well aware of the
fact that not all marketing activity has an immediate impact on sales: rather, and especially in the case of more ‘brand led’ marketing, the effect of such activity may not be felt for some time until after the activity has taken place. Our approach here is very much one of using the ‘right tool’ to do the job: why model sales if sales don’t accurately reflect the impact of the activity? Far more effective would be to model the effect of the activity against so called ‘upper funnel KPIs’ such as Awareness, Consideration or Persuasion as these will likely reveal the true impact the activity has had.
That’s not, however, to say we stop purely at just modelling the brand KPI, rather, we always ultimately tie these movements in a brand KPI back to movements in sales so you can get the complete picture of how your marketing is performing, both in the long and short term.
Quantification of sales drivers
Through our modelling, we can tell you exactly how your sales are broken down in terms of their component parts so that you can easily identify both the controllable and non-controllable leavers that affect your business.
A key output from our models, we can help you understand the true ROI your in-store marketing investment generates so you can identify which activities work for you.
Optimal in-store placement
Whilst having the right marketing assets in-store is important, equally important is having them in the right place. We can help you identify the placements that work best so that you can maximise the potential of our marketing assets.
Optimal number of assets
How many assets are too many assets? Quite often less, rather than more, is important in terms of gauging the number of fixed marketing assets to have in store. Through our modelling, we can tell you exactly how many assets you should have in store so you can again ensure your assets perform to their potential.
Once we know what drives performance in store, we can use this knowledge to understand how your stores should be performing. Often called ‘Headroom’, we can use this to identify which stores to target for improvement to yield the highest rates of initial return.
A very common problem facing agencies that specialise in activity such as merchandising and training visits, is how often should I visit and how much is too much? Our models are able to not only identify exactly what impact your visit has but also how long the effect of a visit lasts so that frequency of visit can be assessed.
Closely related to all of the above outputs, we can use our models to assess exactly how much should be budgeted to generate the required rate of improvement, and because our models tie activity to ROI, we can tell you exactly what rate of return to expect from a given in-store budget.