Message to all retailers: Don't waste time with analytics
In layman terms, analytics is about making sense of data to aid in decision-making processes.
Retail analytics is a contextualised application of using data that comes from your retail operations to make decisions that impact profitability. The goal of this can be to influence short-term sales or to set-up the decision-making infrastructure for long-term success.
You don't need analytics, you need a job to be done.
There are many domains within retail analytics that you can explore but we will expound on them in a different post. This post will be a summarised version of how you should be thinking about retail analytics.
You don’t need analytics…
…you need a problem solved or you need to work towards a goal.
Before going out and hiring a bunch of data engineers and data scientist to wrangle with your transactional data, focus on the problem set that you have. What goal do you want to achieve?
Then, classify them.
This matrix is simple but it is powerful. If you haven’t decided on what you want to achieve with retail analytics, you will be wasting a lot of time and resources blindly chasing a vague goal.
Let’s use this as an example (you should have some hard numbers in your version):
Working backwards from these goals, you need to get two things ready- data on how your shoppers are behaving in your store and how they are interacting with your promotions.
You have to be realistic…
…about what you can do with your data.
The media has hyped up the promise of data and artificial intelligence but in reality, 70% of the job of the individuals in the data team involves cleaning up data before they can be used for their machine learning applications.
Even if you are not planning on building a neural network platform, you still need to ensure that your data is in a state that can be processed.
For your use case, think about your data sets like this…
If your goal is to improve the conversion rates of the various marketing channels, you are going to need to know the outreach of your ads, how many individuals interacted with the ads and how many of them actually converted.
If you only have data of how many people purchased from your store, you don’t have enough relevant data to do any analysis.
This may sound ridiculous to some but some retailers don’t actually have access to their own data. Either because of a bad vendor choice or poor software implementation, they might not be able to access their data.
If you cannot get your hands on the raw data that you need to do your analysis, this has to be someone’s full-time job to fix it. NOW. Every month that drags on leads to a data debt. You don’t want to be 12 months into a business and realise that you can’t even do a comparison to last year’s product performance because you have no access to the data.
Imagine a clean lake. Clear as the sky with water that you can drink from.
Now throw a plastic bottle in. Maybe a cigarette butt. Soon you will see a lake filled with garbage. Dead birds and fishes floating by with flies scooting around their corpses. If you don’t put in the effort to maintain a clean lake, you are going to be drinking garbage.
Your data is kinda like that. If it is no one’s responsibility to clean the data and store the raw files somewhere, more and more individuals are going to start polluting the files with random ad-hoc analysis and the raw files might be lost forever.
I blame data nerds like us for overpromising and overhyping the entire field of artificial intelligence. Now every meeting about analytics involves some discussion around machine learning or deep learning techniques. As a frame of reference, unless you generate millions of transactions per week, it might be too early for you to be thinking about these topics.
The reverse is also true if you do only about a dozen sales per month, it might be too early to invest in a full analytics platform. (Excel should do just fine).
With the data in place, you are still not ready…
…until you have a great data team (or person).
This portion is quite obvious but you need someone who is good at number crunching and can communicate effectively with the rest of the team.
The biggest mistake here is not that you will hire a person who doesn’t have the skill. The biggest mistake is likely that you didn’t hire anyone at all. If someone doesn’t have the skill, it doesn’t take more than a few months to get up to speed with your business and the relevant data science techniques.
You will likely feel as a small and mid-sized retailer that you can manage with your current headcount and you are going to delegate this to someone. This is going to fail because that person was hired for another job and that job is going to be the priority.
The only hack around this is that you give this person some extra budget to go pick up some new skills and procure some analytics tools to make them more efficient at what you are asking them to do. Great employees can do it, but you are literally asking them to learn to swim and giving them scuba gear after you threw them into the deep end.
Give them the tools and training first.
If you have any questions or feedback, we would love to hear from you.