My Five Stages Of The Analytics Process

(EDIT 12/03/2018: I’ve given this some thought and am sticking a fifth and very necessary stage onto the front of my process here, not that I discounted it at first but I do think it’s a vitally important step for overall success and needs to be in there.)

I’ve covered a bit of the widespread debate on Analysis vs. Analytics in a previous post but to really demonstrate the relationship between the two I’m going to look at the four main stages in my analytics process.

DISCLAIMER #1: This is how I group some of the main terms up and how I understand and apply them in my day to day work and in discussions in the wider community. YMMV and if it didn’t I wouldn’t have seen so much confusion in forums and blog comments.

DISCLAIMER #2: I’m not trying to enforce my method on anyone else, there are a number of sub-levels and complexity to this, don’t write to me and tell me I’m doing it wrong. Paraphrasing Swiss Toni, Data Analytics is very much like making love to a beautiful (wo)man. If you both get to a mutually acceptable end result, does it really matter what the exact order of the steps in between where?

ALAN’S FIVE STAGES ON THE ANALYTICS JOURNEY

1) Requirements Gathering

Yes, I know I missed this one out in the first version of this post but it’s so important to overall success that I’m adding it in now. Learning how to speak to business users (especially those without a technical background) is one of the most important skills an analyst will learn. It will stand by you much longer than any coding or database technical knowledge and will ultimately dictate a large part of your career trajectory.

Listen, pick out what the real pain they are feeling is and use your knowledge and their experience to find the best solution. Often it’s not what they originally ask for. Learn (the hard way if necessary) that the customer is not always right and your job is to help guide them and you’ll have a lot more success in every area of your analytical work .

2) Preparation

Invariably the bulk of your time will be spent at this stage. Gathering, cleaning, prepping, cleaning again, joining datasets, extracting, transforming, loading, unloading, cleaning again. You get the picture.

No point going through with the rest of the steps if your data is in a terrible condition, is full of holes or doesn’t actually contain the really important parts you need for the rest of the journey. Cut corners here and you might as well forget about the rest of it, run a random number generator and go to the pub.

3) Reporting

Covers the bulk of Management Information (MI) and Business Intelligence, this is looking at WHAT has happened. Got yourself an Excel spreadsheet with the weekly sales figures for Widget #568, split by date, region and sales location? That’s Reporting.

Fancy a nice pie chart, line graph or interactive heat map by geographic region for your presentation? Yep, still Reporting (even it does look damn pretty).

4) Analysis

WHY did that happen. Looking at your weekly sales figures and see a massive spike in week 24? Now you’re analysing your reporting. Dig into what happened in the business in the weeks leading up to that point and find a certain location was running a local promotion and they cleaned up when the price dropped for that week – that would be your Analysis.

This is where the Inquisitive Nature of the data analyst comes in. We are all wannabe detectives or else we wouldn’t be in this line of work in the first place. See something strange? Step into your inner Poirot alter ego and start digging. Without understanding the business or being willing to learn, you will never level up at this nor the next stages.

5) Predictive Analytics

Working out WHAT WILL HAPPEN NEXT? The next level up from analysing what has happened to see why that happened is predicting where things will go from here into the future. Algorithms and trend analysis are two tools for building the crystal ball that shows you were you can expect to go in terms of sales or customer sign-ups (or losses).

Get through all five of those stages and you’ll have a lovely clean dataset, a fine spreadsheet or data viz of reportable numbers, a (succinct) story to tell senior management what happened and why and a set of numbers predicting where it will go for the next few weeks/months/years.

Revise. Rinse. Repeat. Retire.

If after all of that you still find yourself worrying whether it was Analysis or Analytics you were doing I suggest asking your manager for more work to do.

Worry about cleansing your data. Worry about not spotting the reason behind a shift in your sign-ups or drop in turnover. Worry about missing a coming trend that will blow your predictions out of the water.

Anything else is just noise and not worth your time or energy.