5 things that could scuttle your Analytics initiative
As a retailer who’s getting into setting up an analytics practicewhat you don’t want is to end up with a substandard practice that delivers precious little, or one that takes years to get its act together. Here are a few things we’ve seen that scuttle efforts at setting up an analytics practice.
Perfection Is a Step Away
This is where you do nothing till everything is perfect. Till you have data on all your customers, till all data sources are in place, till that shiny new Big Data PoC is complete, till your new team is hired, till…
There is only one good time to get more Analytics driven, and that is now. The number of steps to go through and the amount of learning to get through is larger than you can imagine, so 3 months of delay is just that – 3 months of delay.
- A new database is no reason to wait – analytics usually works on different systems, not off production systems anyway.
- Lack of customer data capture is no reason either – there’s plenty to be done on billing/ product/ store/ pricing/ promo data till customer data flows in.
- Waiting for your team is a real bad reason – unless analytics is in the DNA of the organization you will need to outsource – not for the arms and legs but for perspective and experience.
Don’t wait for the perfect day. Get started!
Let’s go shopping for a Drill
This is where you want to get into analytics, to understand customers, product, store etc. and decide to go shopping for software. Unless you have the perspective, expertise, and bandwidth to guide at a senior level, all the tools in the world (those DIY ones included) cannot get you on top of your analytics needs.
Sell Through Analysis
Look at it this way. If the Question you want answered is “How do I improve cohort repeat?” the answer is not going to be “Version 2.4 of this tool”. Things can be different in a mature industry but in retail where chances are you’re just about foraying into the practice, you do not need to focus overmuch on tools but on the team and culture.
If you bet on the tools and can’t get value out of them, you end up weakening the faith in the entire system that analytics works for you. Bet on teams instead, and go with their advice on what tools they need to do their job.
Some of us are more equal
Once you do get started, move the culture across departments sooner rather than later. Marketing, Merchandising, Retail Operations, Buying… all can and should benefit from the analytics practice you’re building up. The real way to look at it is that everyone who has a question to ask (that data or analytics can help answer) should be part of the initiative.
Going by the team’s current exposure is no good too. If the teams are exposed to insight and lift from analytics led action repeatedly, month after month, they will pick up the language, they will find the kinds of questions they should ask, they will develop a sense of that a good analytics is.
One thing to really watch out for is elitist thinking – “I’m an engineer so I get this” or “I worked on predictive models in my first job 15 years ago” kind of stuff. Everybody can get it, everybody can benefit – the difference may be in how the story from the data is told. The onus is on the story-teller to make it work.
Some organizations pride themselves on their clear headed action and scorn anything that even closely resembles “gyan”. In the analytics realm that quickly translates into all the analytics being campaign oriented, while strictly insight stuff that cannot stand up (immediately) in the face of “so what do I do with this” is shot down.
I am all for action oriented analytics, but I feel people sometimes miss beautiful work when they close their minds to exploratory stuff. You just can’t see how it will all play out right away,especially if the practice is new. A great piece of insight may be useful at a different time, so listen, assimilate, and keep it away for later – most importantly, don’t discourage it. Don’t destroy exploratory analysis just because there’s no immediate use for it.
A good way to do this is to separate out Insight and Action thinking. Action thinking can tell you which customers to target for a promo. Insight thinking could tell you that there’s a missed opportunity in stocking up extra of large sizes in some stores. Do both.
I get this, so why try anything new?
The methods are many. The error you’ll often make is building up so much belief around one approach that you stop exploring further. A classic case is RFM. Retailers have so much faith in RFM that when it comes to any kind of campaign or customer intervention, RFM pops up as the only method being used.
It has it place, yes it does, but if your objective is to get campaign response, there could be a whole set of other variables you must consider – like the inter-purchase cycle, or past campaign responsiveness, or discount behavior, or range of products bought. In some cases these are far more important than one or more of R/ F or M.
Quite often the fitting one framework will extend to things like using Logistic models for everything, or top down RCAs for all diagnostics, or using a Fishbone analysis for all decisions. Yes we all have our favourites, but keep your mind open to it being a far larger and more varied world out there.
Bonus 6th Point:
A Sample of ONE
This is where there’s too much of a single voice running down everything the analytics team does. Data tends to validate real life (kinda obvious), but some people like to jump with “I knew this already, your analysis is useless” or “tell me something I don’t know” or “this is not my experience as a consumer, your analysis is BS”.
Quell those voices, focus on what you’re getting that you didn’t have before, and give it time! It’s a practice and way of life you’re building, the culture cannot be quick to judge, and it must always respect Statistical thinking where your individual voice is a hypothesis, not an epiphany.