5 Ways to Ensure Your Analytics Provide the Best Results
9th April 2018 | Mike Grigsby
Are your analytics providing you with valuable insights? (Hint: they should be!)
The role of an executive is to guide an organization by making decisions, but how do they know whether these decisions are based on facts? How do they know which lever to pull to alter key metrics?
The purpose of analytics is to provide insights, and good decisions are based on good insights. Yet all too often, insights are being used incorrectly or not often enough.
Information Week (August, 2005) reported that most retail managers who are responsible for price setting prefer using their intuition to make decisions, rather than analytics. In fact, only five percent reported using a decision support system; Rich, MaCarthy and Harris (Accenture Report, 2009) found that around 40 percent of major decisions are not based on facts or insights, but on the manager's intuition. Similarly, Boston Consulting Group (M.K.Egan 2009) found that less than 25 percent of executives managing at least a $1.58B company thought their analytic functions provided competitive advantage or positive ROI.
Here are some quotes from senior executives about their insights:
"What they give me is not even relevant to my business" - CEO, $4B retailer
"The results are usually too little, too late." - COO, $12B manufacturer
"I get mostly tactics, not strategy." - VP Strategy, $2B insurance provider
"When I question the results, the next revision has drastic changes." - CMO, $22B hotel chain
"Gives no actions, just describes problems." - CEO, $33B entertainment conglomerate
"Most of the time I only see obvious, trivial output." - CMO, $2B casual dining
If the above are representative, then generally speaking, executives do not value very highly what are purported to be ‘insights’. It seems to suggest that senior executive do not trust what they receive from the analytic functions or see it as very meaningful.
How can this be, when predictive analytics is entirely about ‘this causes that’? Perhaps the analytics produced is not of use to executives, because they do not provide exactly what they need to uncover?
Below are some other statements I've heard, typically from consumer insights or advanced analytics groups, usually presented to senior marketing leaders:
"92% of our customers wear some kind of jeans when they come to our store."
"We have the highest rated product in our industry."
"Net revenue decreased over 3.5% YOY in same-store sales."
"Our customers trust our brand more than all other brands."
"Both market research and independent field tests show treatment X outperforms treatment Y."
"This trend is going up."
This last 'insight' is particularly disturbing. This should be considered an observation, not an insight, as an insight is defined by a selection of interesting factors. There are a number of elements that define the insights you should be using. For example, all insights must...
1. Contain new information
The information should be new, relevant and non-trivial. To be an insight, and not a mere observation, it must be something more than “this trend is going up.” It may even be counter-intuitive, which often make the most interesting findings.
For example, more than once, in doing a model which has marketing communications as independent variables, email has been found to be negative. That is, as more emails are sent out, it actually decreases the amount of the dependent variable, say units or revenue. Email fatigue is a common answer to this.
2. Focus on understanding consumer behaviour
In marketing, the consumer is king; customer-centricity is a guiding concept. The whole point of marketing is to understand the intent and change of consumer behaviour to the benefit of both the consumer and the business. This makes tracking the consumer’s decision-making process extremely important and valuable.
3. Quantify causality
An insight must be about cause and effect. A marketer needs a lever to pull, something to do, which will affect a change. An insight needs to measure how a change in one variable impacts a change in another variable. For example, if a marketer was given the below model, what could they do with it?
Units = f (seasonality, consumer confidence, corporate tax rates, industry growth and inflation)
What insights can we garner? What actions does it provide? Does the marketer have a lever to pull? The answer is no - this model provides nothing actionable for the marketer and is therefore of limited use.
4. Provide a competitive advantage
An insight must be a piece of intelligence that an organization’s competitors do not have. All intelligence is based on having awareness of specific and relevant information, which provides the intelligence for an organization to win the competitive advantage.
5. Generate financial implications
An insight should be measurable. Whether it’s ROI, contribution margin, or risk assessment, any insight should have some financial implications. If there is not a measure increase in revenue or satisfaction, or a measurable decrease in expenses, then the validity of the analyses should be questioned.
Essentially, all of the above drills down to one thing: actionability. If an insight provides actionability, within the dimensions mentioned, it is providing marketers and managers with the information they need to make better decisions. The hypothesis is that if these decisions are based on data, then the chance of making the right decision increases.
About the Author
Mike Grigsby, author of Marketing Analytics, has over 25 years of experience in customer insights and analytics. The second edition of Marketing Analytics enables marketers and business analysts to leverage predictive techniques to measure and improve performance. By exploring real-world marketing challenges, it provides clear, jargon-free explanations on how to apply different analytical models for each purpose.