Three Robust Big Data Metrics that Serve Up Satisfying Reports

An executive sits at a conference table, reviewing a newly released digital metrics report. She asks a few questions trying to relate business value to the numbers on the page. She finds, and perhaps you do too, that relating digital data to business value is like eating ice cream for dinner. It might look good, but it simply does not satisfy. Digital reports are often full of lightweight metrics.

Lightweight Metrics

Lightweight metrics usually imply something positive has occurred, but do not drive action alone. They do not directly relate to revenue or reduce costs. They will not pass the "so what" test. Here are some common lightweight metrics:

Facebook likes

Tweets

Uncut visit or visitor traffic

App downloads (free)

I see these a lot in reporting tools/vendors not run by analysts. If you ask an analyst to attach business value to these lightweight metrics, it will be a stretch. It is the equivalent of counting the calories of energy from the sugar in your ice cream. Eating a lot of ice cream will not make you a faster runner any more than gathering hundreds of likes alone will drive up your business revenue.

Creating Robust Big Data Metrics

Instead of trying to superimpose value on lightweight metrics, use a more robust metric designed to handle complexity. These are amazing combinations of metrics that satisfy because they stand on quantified research already attached to business value. These metrics are actually formulas that describe the digital business and its trends.

We sometimes leave these formulas to quant jockeys to cook up but there are several standards that anyone can use. Here are three rich and robust metrics to use in the digital world.

#1 Engagement Rate

When put together well, this index can be a great leading indicator. Think of the S&P 500 index which only contains a portion of total stocks yet generally matches the trend of the market. Engaged customers may play videos, read key pages, use tools or otherwise find value in your site or mobile application. They are willing to exchange their time for the value you offer and perhaps solve a problem with the products you sell.

Engagement rate = (each valuable actions) / total actions

However, like any index, there is an art to adding enough ingredients to this rate to give it meaning without over-stuffing which makes it flat and meaningless. To keep Engagement Rate tight, I like to segment it. More on using Engagement Rate in a minute.

#2 Customer Lifetime Value

Customer Lifetime Value is simply the future profit from customers. There are actually many ways to calculate this - and you may find your CFO is already calculating this to derive your marketing budget - but I think the formula must contain two things. First, a discount on future cash. Second, profit (which factors out marketing spend). I would personally use this to open a conversation with the financial team and further refine and align the model.

Using your existing data by looking at historic trends, find out:

1) How much does a customer (or class of customers) is spend with you now? Some companies break customers apart by how much they spend, some by location, some by product purchases. However you group customers, make it mutually exclusive so they do not overlap. For example, if a customer spent a total of $150 they cannot also have spent $500 for the same time period.

2) How much more or less will they spend with you in the future? This is an expected probability modeled on data you already have from similar customers. It makes sense to pick out low, medium and high probabilities from which you can multiply the third piece, time.

3) How long you expect each person to remain a customer? This could be months or years.

The math (oh no!) is not that bad. You can skip ahead to the next paragraph if you like.

Let's say a group of customers "Small Spenders" spends $100 a year online with you today. In future years, they could spend high ($1000), spend medium ($500), spend low ($100) or lapse ($0). According to my historic customer data which is based on other customers that look like this one, the probability of spending high is 20%, medium is 30%, low is 25% and lapse is 25%. Note the total is 100%. Calculate each one: High ($1000 * .20) = $200. Medium ($500 * .30) = $150. Low is ($100 *.25) = $25 and lapse is (-$100 * .25%) = -$25. Lapse and low cancel out. Add High and Medium together to get $350.

So now I know my Small Spenders spend $100 with me today and have the potential to spend as much as $350 next year. Multiply out by the number of years you expect to keep them as a customer and you have lifetime value. The per year calculation can get more complicated if you adjust probabilities each year, but let's keep it simple. If we expect to keep them for 5 years our customer lifetime value is $350 x 5 = $1750. Keep going to reporting to see how this can be integrated.

#3 Customer Satisfaction Score

There is a direct correlation between customers who are satisfied after the purchase and their future spending. Traditionally, marketers spend little time and attention here and this is a BIG mistake, especially in the digital world where reviews and recommendations have immediate impact. Several studies have found strong correlations between satisfaction and profit. Here is one study that found a 1% increase in satisfaction can increase profits by 11%. Digital consumer satisfaction is usually scored from online surveys. This is not the same as a user interface survey such as the OpinionLab plus sign which looks at navigation. This is a quantified pop-up-when-the-visitor-meets-specific-conditions survey such as the kind Foresee and iPerceptions produce. These companies also have industry benchmarks which you absolutely need for proper context.

Back to Reporting

Now your digital reports can talk revenue which is very satisfying. They can do this by reporting conversions by customer group ("Small Spenders") and how much above or below the expectation each group is (We had $100, and so far in year two we have $200 of an expected $350). However, we need to expand our view upstream and downstream.

To expand upstream, match the behavior illustrated by each class of customer BEFORE they became customers. This is the Engagement Rate which is now a leading measure. For example, after reviewing their history, you find Small Spenders engage infrequently, consume little key content and no video. To encourage Small Spenders to spend more, you might look at the products and messaging they are missing but other higher spending groups are receiving.

To expand downstream, use the same customer groups to chart customer satisfaction scores. Benchmark your scores against the industry. If you are trending high, keep it up! If you are trending low, you can unpack the problem with refined survey questions or by reviewing call center transcripts.

By combining all three metrics across a segmented customer base, you now have a compelling story to tell, logical actions to test, and a more meaningful report. Hooray!

Conclusion Using richer, yet simpler metrics is common in other industries. For example, investment banks used the Value at Risk (VAR) metric to determine if their investment risk was appropriate each day. And yes, this does carry one big caveat which is to continue to use common sense and not rely ONLY on this metric. It is time for digital analytics to mature into better indexes and richer equations that really satisfy when it comes to describing customers and their digital impact on business value.

Previous
Previous

How to Turn Your Big Data Firehose into Insights in 3 Easy Steps

Next
Next

Book Review: Naked Statistics . . . Aka How Not to Kill People with Statistics