Data-driven marketing decisions you can make today

data-driven_decisionsFor crucial life decisions, I consult my head, my heart, and my gut. At least two of the three need to be in agreement for me to act. But marketing decisions are not life decisions, and marketers in general make most of their decisions based on their experience and their instincts.

Of late, however, as marketers realize the power of data to help them in their job, they want to start making data-driven decisions rather than relying on instinct or old habits. They often don’t know where to begin, so we offer here a guide.

We need to make an obvious assumption, that you do have data. We assume it exists in one or more of the following forms:

  • A marketing database
  • Transaction data from your POS, ERP, or order entry systems
  • A Longbow system (Longbow is customer analytics software for direct marketing; it includes a marketing database and a browser interface to access the database)

If you’ve got the data, here are some important decisions you can make based on it.

Decide when to consider a customer ‘lost’

Some marketers call this the ‘inactivity period’ — how long a customer can go without making a purchase before you classify the customer as ‘lost’ or ‘defected’ or ‘attrited’.

Why make this decision: Customers in the ‘active’ file usually get a different marketing mix than inactive customers, with different offers, different messages, and different frequencies. You want to put customers in the right group to maximize revenue.

How this decision is usually made: Companies pick a chronologically convenient period, typically one or two years.

How to make a data-driven decision: For customers who have made more than one purchase, look at all transactions and find the time after which 90% of those customers have made their next purchase. Call that the activity period. In other words, the active file includes customers who made their last purchase within the time span that it takes 90% of customers to make their next purchase. Yes, there will be some customers who purchase outside this time span. And yes, they will be marked as inactive and then put back in the active file when they return to purchasing. Setting the activity period this way ensures that almost all customers who repurchase will stay active.

Decide how to divide your marketing budget between acquisition and existing customers

Why make this decision: Your marketing expenditures should be based largely on your sources of revenue. Most stable companies get most of their revenue from their existing customers. Too many e-commerce companies depend on first-time buyers for most of their sales, what we call being ‘addicted to acquisition’. If the bulk of your revenue comes from existing customers, it’s folly to under-market to them while allocating most of your budget to acquisition.

However you can’t simply look at how your revenue is distributed and say that if 60% of my revenue comes from existing customers, I’ll give 60% of my marketing budget to that group, with the rest going for new customer acquisition. Acquisition is expensive.  It costs less to make a sale to an existing customer than to acquire a new one. Acquisition needs a disproportionate share, but not so much that will restrict the lifeblood of sales from your current customers.

How this decision is usually made: Growing revenue from existing customers means cross-sell, getting the customer to purchase from new categories of products. Most companies don’t know how to do cross-sell, and typical cross-sell response rates for direct marketing are well under 1%. Faced with this failure, companies fall back on what they know how to do, acquire new customers.  Marketing automation systems primarily support acquisition, so by default most of the budget goes to this effort. Marketing to existing customers happens, but it is under-funded.

How to make a data-driven decision: Our assumption here is that revenue from a group is roughly proportional to the marketing budget allocated to them.  Generally this is true, but of course there are exceptions. With this assumption, here’s how to adjust your revenue distribution.

First determine your penalty, how much more it costs to acquire a new customer compared to making a sale to an existing customer. The commonly accepted number for this penalty is from 5x to 8x — it costs from five to eight times more to acquire a new customer than to make an equivalent sale to an existing customer. If you don’t know the number for your business, pick one in this range. Call this number P.

Next, calculate the ratio of revenue that comes from new customers compared to existing customers. Call this ratio R. If you are satisfied with this ratio, keep it.  If not, set R to where you want it to be.  R=3 means that three times as much revenue comes from existing customers, a healthy ratio. Then a simple calculation will show you that the share of your marketing budget that should go to existing customers is:


For example, for R=3 and P=5, 3/(3+5) = .385, or 38.5% of your budget should go to existing customers for them to contribute 75% to your total revenue.


While some writers advocate SWAT teams to implement analytics — see the recent Forbes article here — making data-driven decisions is easier than you may think. The first step is to collect and organize the data you have in a marketing database. Next, get it analyzed on a regular basis. Identify the actionable metrics that best suit your organization and match them against the marketing problems you are trying to solve. This post is a starting point to begin leveraging the trove of data in your corporate systems. My next blog post will have three more decisions you can make based on the data you have.

Tags :