In a world where information access and customer interaction happen anywhere and anytime, marketing is becoming a connected web of communications and calls to action (CTAs) designed to immerse the customer with a 360-degree experience and a guided journey.

In such a world, how can you attribute customer actions to a specific marketing activity that stimulated that action? No wonder all the discussion and new products focus on attribution.

Unfortunately, it's easy to invest a ton without getting clear results. In fact, it's probably not that relevant anymore, or at least we need a new way of looking at it.

To illustrate the problem, let's say you mail coupons or a catalog to customers. You also email them regularly with offers, you run ongoing search advertising campaigns, and are active on social networks. Then, one day a customer places a $100 order on your e-commerce site.

What was the contribution of any one thing? Should you apply rules based on your understanding of how customers think and behave, or look into the latest data science or machine learning platforms? If either reliability or cost give you pause when it comes to those things (and they should), you could simply attribute the full $100 to the last communication sent or the last action taken by the customer just before the purchase.

But what if the last email triggered the customer to look at the catalog from weeks ago, which prompted a search for competitors where he saw your ad, but was ultimately swayed by an online review? Some credit ought to go to the email that started this journey, but the customer never clicked through the CTA in that email.

Face it, your customer's engagement with your company will be increasingly blurred across channels, campaigns, messages, content and networks of all types, such that it's hardly relevant or meaningful to get too specific about attribution.

Except for one little problem: Marketing budgets have to be allocated on some basis of ROI, which requires attribution.

Maybe a different approach is needed. Instead of trying to accurately attribute value to each campaign source or channel — which are all a blur — maybe it's better to focus on how your marketing mix is changing customer value. This idea starts with the premise that there is no more important measure of your marketing success than getting people to purchase more of your products and services.

So instead of getting too hung up on the response rate (e.g. click-through) of a particular communication, measure changes in customer purchasing against the total "experience" you are throwing at them.

In terms of attribution, this means not thinking about one source or channel in isolation. That's hardly meaningful anymore. You have created a mix of integrated communications that are becoming "business-as-usual," the new constant.

So if you want to find out the impact of a particular campaign or channel, target certain customers for it (the target group) and withhold it from others (the control group), but subject both groups to all other elements of your marketing mix in a business-as-usual manner. Then, see who actually purchased more — the target or control group.

For example, if you wanted to know the impact of personalized offers to each customer, send that communication to the target group but otherwise everyone is exposed to the regular mix of communications planned or ongoing at that time. It would not be appropriate to measure this campaign purely on click-throughs of the communication itself. As discussed, the marketing mix and customer interactions today are far too complicated for that.

A customer seeing the offer by email could easily, and might perhaps be more likely, to go to your e-commerce site without ever clicking on the CTA. And once on your e-commerce site, they might not purchase exactly what you were offering based on their research or browsing.

The only question in the end is, did the target group purchase more than the control group? Simple and factual.

This breaks the problem down to simple and familiar A/B testing. A/B testing doesn't scale if you look at everything as a variable. But you can't control all the variables involved in customer engagement. Instead, consider the dynamism of everything you and your customers are doing as a constant, except the one thing you are testing.

The trade-off for this simplicity is the underlying fact that so many variables are in play and affecting any given customer's actual engagement with your company, so you cannot be sure that some other factor did not skew the results. For this reason, it is important to do more then one test.

If you do 4-5 similar tests over different periods that in aggregate show a meaningful lift in purchasing by the target groups, it does not mean that other factors did not affect results somewhat, but you can reasonably conclude a positive impact. And that's mainly what you want to know.

You can't be sure the ROI is exact, but you'll have an approximation that is as reasonable as any — at far less complexity and cost.