The 80/20 Rule of Analytics every CMO should know


Jain, Piyanka

May 27, 2013

This blog is part of Aryng’s analytics tips series for executives: CMO, Chief Product Officer, and CEO

With all the talk about Big Data and Predictive Analytics – both of which involve complex, advanced skills and tools, driving millions of dollars in marketing – it is hard to believe in the power of simple analytics.

The truth, however, is that only 20-30% of the decisions really need the use of advanced techniques like predictive analytics.   Seventy to eighty percent of marketing decisions can be judiciously addressed with simple analytics techniques, which can be learned by any marketer and executed on an Excel spreadsheet.

Consider a breadth of industries: financial services, consumer goods, eCommerce, automobile, technology, media, and so on – a CMO broadly expects 3 key outcomes for his business:

  1. Bring more “future” customers to the door in the most cost-effective manner.
  2. Convert more of those who come to the door into customers.
  3. Keep the current customers “buying.”

In essence, he seeks a wide and targeted top of the funnel, and higher conversion at every stage to achieve maximum revenue at optimal ROI.  Data can support an optimized funnel through questions like: who and where to market; how much to spend on each channel; what drives response and conversion; who likes what message, what offer and what product; and what drives churn.  While this seems like a compelling case for Predictive Analytics, let me in contrast lay out a framework with better ROI using simple  analytics techniques to arrive at insightful and informed decisions for this CMO.

  1. Bring more “future” customers to the door in the most cost-effective manner by:
    • Increasing marketable universe by identifying new channels based on existing customer profile. (Aggregate Analysis, Sizing/Estimation)
    • Targeting messages and offers based on past marketing campaign to increase response.  (A/B testing, Correlation Analysis)
    • Optimizing channels to Increase ROI and decrease cost of customer acquisition (Correlation Analysis)
  2. Convert more of those who come to the door into customers by:
    • Identifying Conversion Drivers: Does certain fulfillment options, user experience, reviews options, cart options, payment options, offers and promotions drive incremental conversion? (A/B testing, Correlation Analysis)
  3. Keep the current customers “buying” by:
    • Segmenting the base to drive engagement (simple segmentation based on past product usage or RFM or similar).
    • Launching engagement campaign, customized by segments to drive “buying”.
      1. Understand Engagement drivers (like certain offers, discounts, bundling, loyalty memberships etc) for each of the customer segments (Correlation Analysis)
      2. Campaign analysis – what resonates with customers and what doesn’t (A/B testing, Aggregate and Correlation Analysis)
      3. Understanding drivers of Churn – factors that make customers leave your business (Correlation Analysis)

Note that all of the above analytics techniques I would use initially are simple techniques that can be done in Excel and can be learned by any marketer. The success and efficacy of  these techniques would be powered by hypothesis driven planning, using a “Data to Decisions”™ framework like BADIR™. As the insights from the business mature, simple techniques may then in some cases point to a need or an opportunity to leverage advance techniques like predictive analytics.

Let’s say, an Ecommerce marketer uses the framework above and increases the cart conversion to 60% by identifying two major detractors to conversion, a redundant extra page in the flow and a bug in the cart using correlation analysis. That is a big win! But the cart conversion is still below that industry’s benchmark of 65%. From the analysis earlier, she finds that many additional independent attributes (like Google Checkout as primary payment option, page load time etc.) have an effect on conversion, but they are all individually insignificant and not worth the ROI for making the changes in the flow. Having established that, the marketer could then take the next step to engage with the Data Science/Analytics team to build an advanced conversion driver model that incorporates the effect of all of these independent attributes. This model can help identify the biggest factors that drive conversion and explain the 5% delta. Equipped with this information, the marketer can then work with the site engineering team to make the smallest/easiest/cheapest changes to the flow to get the biggest conversion impact. This is the perfect case for the use of advanced techniques. But this is typically only 20% of all use-cases, usually as a build up from simpler analysis!

Predictive Analytics are resource and time intensive – to the tune of 10-20x of simple analytics.  They need advanced skills and tools, historical data, operationalization, live validation, and constant maintenance and that is the reason for not using advanced skills to solve every business problem. A marketer, a product manager or an operations manager equipped with the right “Data to Decisions” framework and easy access to data can optimize 80% of their day-to-day workflow on their own, without having to rely on scarce and expensive analytics resources. For the 20% of decisions, where the potential ROI justifies the use of advanced techniques, they can work with their analytics counterpart. This is the picture of a well-functioning organization competing on analytics. In contrast, when organizations and their leaders are misled by Big Data and Predictive Analytics hype, they end up investing lopsidedly on advanced data analytics tools and resources, often resulting in poor ROI.

In summary, a smart CMO knows that a marketing team equipped with a “Data to Decisions” framework and easy access to data without the support of a data science team would fare much better than a marketing team with no data skills supported by a large data science team. Marketers have the advantage of having the deepest knowledge and experience about their product and their customers. That context, when married to data, is phenomenal. Without that context, even the best models would fail.

To learn more about BADIR (Aryng’s “Data to Decisions” framework) and the 5 Myths of Predictive Analytics, download the respective whitepapers here. If you are CMO, CEO/GM or CPO, ready to get your team equipped with the “Data to Decisions” framework, contact us for a free consultation on what your team may need. If you are ready to equip yourself with the “Data to Decisions” framework, start by taking our “Data to Decisions” intro analytics course online today! Once you have taken the level-1 course, you would get access to level-2, “Hands on Analytics” course and Level-3, “Hands-on A/B Testing” course online to complete your “Data to Decisions” skill upgrade.

Follow @AnalyticsQueen | Sign up for our Newsletter | | Contact Aryng

Great ! Thanks for your subscription !

You will soon receive the first Content Loop Newsletter