Data science for marketers (part 2): Descriptive v diagnostic analytics

In this series, we previously talked about the essential steps you should take before starting your big data analytics programme –  see part 1: decide on your end game and start the data consolidation process.

Data consolidation in data science process

Now it’s time to move on to the more thrilling part – learning how to ask the right questions (using data science) and receive the best answers (using data analytics).

Your big data, once ready to use, enables you to understand:

  • What has happened? (descriptive analytics)
  • Why did this happen? (diagnostic analytics)
  • What will happen if…? (predictive analytics)
  • What should we do about this? (prescriptive analytics)

In part two of our series, we’ll focus specifically on descriptive and diagnostic analytics.  Both of them serve as a pillar to your big data value chain and are essential to developing more advanced algorithms supporting predictive and prescriptive analytics.

What is descriptive analytics?

Descriptive analytics mines historical data to identify common patterns and correlations between certain outcomes. It is the best way to distill large volumes of data into succinct easy-to-understand insight.

Google Analytics is a prime example of descriptive analytics. You don’t need to go through a variety of numbers and apply formulas to see how your keyword positions have changed in the last week. However, you are not provided with the exact reason of why that happened – that’s up to you or a diagnostic algorithm to decipher.

Descriptive algorithms also help establish different relationships wired in your data. For example, they can be used to classify different prospects into groups. So a predictive algorithm attempts to forecast the likely behaviour of a consumer group. A descriptive model helps that algorithm to estimate the relationships between different consumers and different products, so it can later ‘join the dots’.

Use cases of descriptive analytics in marketing

Estimate intent

Social media monitoring tools and sentiment analysis tools help you identify who your potential customers are, what matters most to them and how they tend to behave on different social platforms.

Group leads

Curate your incoming leads automatically based on the available demographics information and additional data. Set different parameters that will keep your lists well organised for the sales teams. By adding predictive analytics to the mix, you can also estimate how likely it is that a certain prospect will buy from you.

Deploy advanced audience segmentation

Locate returning customers and those who will buy from you with little-to-no nudging or those who won’t budge unless prompted with an additional sales deal. You can also segment your audience to learn what kinds of content and messages resonate with them the most; whether they are responsive to certain types of advertising and so on.

Ultimately, descriptive marketing analytics tools help you receive more information about your audience, your leads and your past ad campaign performance. There are a lot of metrics worth estimating this way to get your daily fix of numbers.

What is diagnostic analytics?

The goal of diagnostic analytics is to understand why something happened. Why did you have 10% fewer sales in October, despite investing more in LinkedIn marketing? – asks your CEO.

A valid question you don’t have the answer to at the moment. In the Event Actions/Goals, you notice that some users from LinkedIn added the product to the cart, but did not check out. The data shows that Exit Rates are through the roof when customers were asked to provide their shipping address and payment details. A lot of things could have gone wrong:

  • The form didn’t load correctly
  • The shipping fee was too high
  • The form was too long and non-mobile friendly
  • Not enough payment options available

The goal of the diagnostic analytics is to help you locate the root cause of the problem. To do so, the algorithms use owned proprietary data, and leverage outside information (e.g. reports from LinkedIn or Google) to understand what exactly happened and help you find a quick fix.

Use cases of diagnostic analytics in marketing

Spot and respond to anomalies; some questions cannot be answered by simply looking at existing, albeit distilled, data. For example, what caused a sudden drop in website search traffic without any obvious reasons? A diagnostics tool can tell you that you have an imbalanced link distribution among inner pages and that caused a plunge.

Prioritise your insights and action. By knowing the root cause of certain problems, you can prioritise your actions to avoid them in the future. Know the areas where you need to pay special attention to achieve positive results, or avoid setbacks.

Refine your marketing messages and sales offers. Diagnostics helps you determine causal relationships between different data points. Your data can tell you that a free shipping promo deal leads to a 15% increase in average order size, whereas without the deal, you’ll get 25% higher exit rate. Big data consists of a lot of moving parts and diagnostic analytics helps you determine how exactly your actions will shake those up.

Uncover new data stories. You may be missing out on a bunch of hidden relationships within your stash of data. For instance, you can compare consumer response to the same advertising campaigns in different areas and discover that female travellers aged 18-25 from Manchester are more likely to fall for the “sunny and luxurious resorts in Spain” rather than “affordable and boho homestays in Greece”. You can diagnose how different tech health aspects of your website contribute to your rankings and so on. Furthermore, you can then turn those data stories into marketing storytelling.

Continue our ‘data science for marketers’ series

Your data has now gained new context and can be further cooked up for the final step of your journey – understanding “what will happen if…” and what is the right course of action to take. That is exactly what we will be discussing in the third and final part of our series – the transition to marketing foresight and unlocking accesses to actionable insights.

Part 3: Predictive v prescriptive analytics

George Karapalidis profile picture
George Karapalidis

George is an SEO Specialist and Data Scientist in the Portsmouth office. He has worked in E-Commerce and Digital Marketing across many industries, for and with companies all over Europe. Before joining Vertical Leap, George worked as Marketing Director for his own company, for which he managed to expand the company’s activities to 5 European countries. George has been creating websites for more than 10 years and he has in depth experience in designing and bringing optimised E-Commerce websites to market.

More articles by George
Related articles
Data cloud image to illustrate Data science for marketers (part 3): Predictive vs prescriptive analytics

Data science for marketers (part 3): Predictive vs prescriptive analytics

By George Karapalidis
Traffic lights image illustrating the article Data science for marketers (part 1): Getting your data ready

Data science for marketers (part 1): Getting your data ready

By George Karapalidis
AI in travel: 3 ways to build customer loyalty

AI in travel: 3 ways to build customer loyalty

By Chris Pitt