Changing the expansion landscape by understanding consumer’s behaviour

Understanding consumers, competitors and attractors is key in any retail strategy. Retail Intelligence gathers together many different techniques to drive decisions from insights obtained by data. By making informed decisions from real and reliable data, the risk is reduced significantly. Quoting our Head of Data Analytics “At the end of the day, it is all about making more money or losing less”.

When we talk about characterizing some retail point of sale, the term catchment area is often used, which represents the area where the point of sale has some influence, i.e., whose consumers are potential customers.

In previous posts we have already discussed how these catchment areas can be built, based on isochrones, mobility data and consumption data. In this post we will show some hints to learn some insights from these pieces of data.

For example, let’s say that we own a point of sale and we are planning some expansion initiative with a new opening in the same city. Let’s take the example of figure 1. In this case we can see isochrones of 10 and 15 minutes on foot, respectively for the existing store (green) and the new candidate (orange).

Figure 1: Isochrones of 10 and 15’ on foot.

One of the top 3 fears for any expansion manager is called cannibalization. This is, the situation when an action like opening a new store draws customers away from an existing location in your network, instead of capturing new market share in this area.

This is something key in the decision process and a highly demanded feature by our clients.
If we take our example, we see that if the catchment area of our business is ~10 minutes on foot, then we are safe from cannibalization, but if our catchment area is larger than 15’, then we see there is some overlap between the areas, and the overlapping surface represents consumers impacted by the presence of the two stores.

Figure 2: Overlap between two isochrones

So, first problem already arises, when the size of the catchment area is not clear. In fact, most of our clients couldn’t guess at first the most realistic size of their catchment areas. And, under this scenario, missing that information may have a significant impact on cannibalization calculation, as we just saw.

The picture changes if catchment areas are considered in a different way, for example using mobility data. Taking the procedure described in our previous post, we can use mobility data to understand the real movements of consumers. In figure 3 we took the example of the store in the left, and we can see how different catchment areas are when computed in different ways.

Figure 3: catchment area obtained from mobility data

The advantage of this methodology is that we can repeat the process for the new store and study the areas of overlap, and therefore the potential cannibalization. Figure 4 shows in blue the areas of potential cannibalization. The darker the blue tone, the more people living in that area visit both locations of study. Comparison with figure 2 is obvious. New mobility patterns are visible now, like a bias north-south in mobility patterns which could be due to the city design and/or metro lines design. The main point is that now we can quantify the cannibalization potential from the real mobility patterns of the consumers, and take much more accurate measurements.

Figure 4: areas whose residents visit both locations

Does this mean that all the blue area is cannibalized area? Of course not. As galicians always say…. it depends. In figure 4 it looks like there are a significant amount of cannibalized consumers in the city center, at the bottom of the picture. But the blue cells only show mobility patterns!!! This means that if your business is convenience, most likely few people will take a 20 min underground to buy at your business. Probably, these people living in the center of Madrid were never your customers in the first place.

However, if your business is different (fashion, leisure, restaurant….) then you may very well impact people from more far away, and in that case considering these people may make a difference in your predictions.

The key point in any informed decision starts with having good and complete reliable data, but if you don’t have good analysts to extract value from it, it is useless. As you may not want to buy the most expensive and exquisite ingredients and then ask the most terrible chef to cook them.

Thanks to Geoblink’s exclusive partnerships, we can get an holistic view of the consumers behaviour. Again, going back to our example, we can use spending data from the purchases paid with card on retail, and we see that most consumers spending in groceries around our candidate location live in the same zipcode. But, in another type of business, the story is different, as we have already explained here.

Finally, another key aspect to look at when designing your expansion plan is competition. Any analyst needs an accurate database of all attractors and competitors in the area. Fortunately, Geoblink has an unique database built upon many different sources and proprietary algorithms to enrich and dedupe our POIs.

The problem is that, in a typical case, a lot of business knowledge would be needed to account for the fact that, indeed, not all competitors are as dangerous. Scaling that to a region/city where there are unknown players, might be dangerous in the decision making.
Fortunately, thanks to our spending datasets, we can also see the areas where the consumption is larger and therefore identify the strongest players in the area.

Figure 5: Competitors (black dots) and spending for 1 specific business category (red rectangles)

Figure 5 shows the main competitors if we chose to play in the supermarkets business, while the rectangles show the consumption in food in those areas. Some areas do not show any spending data because there are some legal restrictions to preserve anonymity. Other squares with spending data and without any black dot have to do that, for this specific example, we are considering only supermarkets and not other convenience food businesses.

Now, from an analysis of these situations, interesting studies may be derived of which are the more harmful competitors, and which are the winners and losers in this game.

All this information is very related to the concept of penetration rate. From both mobility patterns and consumption patterns, penetration rate is no longer a matter of distance to the point of sale.
Gravity models have tried to understand and describe consumers patterns from concepts like distance (to the point of sale and to the competitors) and some sort of importance of each point of sale. Although they can be very useful, limitations are known as they may over simplify the reality of the consumers patterns, as well as to need a lot of a priori knowledge and fine tuning.

With our exclusive datasets and algorithms, we see what is going on in the retail sector with new fresh eyes and different perspective. This allows us to come up with different strategies to solve better the same question expansion always had, as well as tackling new business questions.

Challenging and exciting times are ahead of us, for those with passion for solving problems through technology and data.
If you feel like one of them, don’t hesitate to take a look to our open positions!!