Comparing the ROI of different customer experience strategies


When calculating the ROI of CX programmes, it's tempting to aggregate both costs and benefits across all channels - but this would be a mistake, because different approaches have different costs and returns.

8th Jun 2020

When we try to calculate a Return on Customer Experience (RoCX), it is very tempting to aggregate both costs and benefits across all channels, but this would be a mistake – different approaches have different costs and returns.

There have been a number of articles recently about calculating the Return on Investment of CX, including my own. All of them agree on the growing importance of writing a robust business case based on reliable data and predictably quantifiable results – especially in the ‘new strange’ world we now find ourselves in. However, few get into the nitty-gritty of the comparative RoI of CX of a multichannel environment.

In this article, I will explore one concept: that the RoCX formula is different for different channels or approaches. I will do this by using two exemplar models; one I will call 'High Touch' (which may be human-centric) and one 'Low Touch' (which may be wholly digital). I know that it is rarely this clear-cut, but bear with me.

Return on CX is rarely linear

In researching Return on CX, I quickly discovered that the relationship between the return and the costs of CX is not linear – it typically gets more expensive to deliver equal incremental improvements in satisfaction (or however we measure the performance of our CX programme).

Using this as a basis, let’s compare two examples:

The ‘High Touch’ example

Return on CX

In the diagram above, we can see a number of features about the high touch model:

  • Even delivering a very poor level of customer satisfaction (point 1) involves some effort (or cost).
  • Improving customer satisfaction to point 2, involves some modest additional cost, from 2 to 3 costs comparably more, and going from 3 to 4 is much more expensive still.
  • So, moving between points on the curve above delivers the same level of improvement in absolute terms (all the vertical steps are equal), but varies markedly in the effort required to deliver.
  • Nonetheless, if we are prepared to spend the money on improving the customer experience we deliver, we can theoretically follow this curve to the maximum possible level of satisfaction.

NB: It is because of this phenomenon of diminishing returns that organisations with already high levels of customer satisfaction should think long-and-hard before committing to stakeholders that they are going to improve the absolute level of satisfaction still further: improving our satisfaction score by five points is a lot harder at point 3 than at point 1.

The ‘Low Touch’ example

Return on CX

  • Just shifting to a low touch (e.g. digital) channel may well reduce the cost (point A) to deliver the same, or better, absolute level of satisfaction as point 1 in the high touch example – it is this that makes digital transformation so appealing to many organisations (and why not? if the customer remains as satisfied).
  • For the same cost as point 1, we can potentially deliver a significantly better experience, resulting in higher customer satisfaction (point B).
  • However, the low touch approach may well have a ceiling level of satisfaction (partly because it lacks the added experiential value of a human touch) and at point C the two RoCX curves intersect.
  • Note that by point 4, the low touch approach is delivering a lower level of satisfaction for the same level of effort as the high touch one (by now, we may be using incentives, gamification and gimmicks to increase engagement and satisfaction).
  • By point D, the low touch experience has plateaued - even spending more (and there may be a limit to how much can be spent (point E)) has no measurable effect on satisfaction.

NB: In a different modelling exercise, I noticed something else that is relevant here – in some cases, a higher level of customer satisfaction did not translate into noticeably improved business performance at the customer level. For example, highly satisfied customers did not exhibit significantly more loyalty than those that were just satisfied.

Other considerations

  • As I said at the outset, it's rarely this simple - different customer segments expect, and may well accept, differing levels of satisfaction - if you have ever used an Interactive Voice Assistant (IVA), or an online search tool, you will know that we have to learn about their capabilities and limitations - once we do, we can accept their limitations.
  • Having a clear strategy about the purpose, the advantages and limitations of each channel, and when to switch between them pays dividends; in my experience, those organisations that offer hybrid experiences are those that can balance cost and reward most effectively. They also tend to exhibit more agility.


  • The relationship between cost and return for customer experience is rarely linear – achieving incrementally higher levels of satisfaction often come at exponentially increasing costs.
  • The RoCX for different approaches is different, with costs to deliver the same net level of satisfaction varying significantly.
  • 'High Touch' isn't necessarily better, 'Low Touch' isn't necessarily cheaper - it all depends upon where you are on the curves.
  • Organisations close to point C may find, perversely, that adopting a low touch approach ends up costing more to deliver a similar level of customer satisfaction, or they may even end up losing ground. For some of these organisations, 'digital transformation' may be deemed a ‘failure’ – often because expectations were set too high, or limitations ignored.
  • If you try to calculate a Return on CX using just one 'average' curve, you are missing out on a lot of important detail - there is no such thing as an 'average customer', neither is there an average multichannel experience.
  • If you are at the top of the curve, spend even more time thinking about what your next steps should be - you are in expensive territory if you get it wrong, and none of us can afford that right now.

Final thoughts

Although I have used ‘customer satisfaction’ in my illustration, as I explained in an earlier article, using measures of satisfaction alone is not how I would write a compelling business case – indeed, in selling the value of CX to the Board, there are a whole host of other things we need to consider.

Nonetheless, there is a good correlation between high satisfaction and improved business results overall. But I would recommend swapping out 'customer satisfaction' for something more meaningful to the business - the curves will be different, but the approach still holds.

If you would like know more about making a strong business case for your CX program, please feel free to contact me directly, or leave a comment below.

Replies (3)

Please login or register to join the discussion.

By LinkedIn Group Member
08th Jun 2020 13:08

This comment originally posted in the MyCustomer LinkedIn group by member Christophe Charrier:

Thanks for sharing. On the final thoughts, it also depends if the correlation between Customer satisfaction and Sales revenue has been established within the Business. And if Customer-centricity is really a value and a pillar of the Corporate strategy, CX and Customer Satisfaction will both matter !

Thanks (0)
Michelle Spaul Customer Experience Consultant
By Michelle Spaul
11th Jun 2020 09:49

Great post, love how you draw out the differences between high and low touch.

Thanks (0)
Steven Walden
By Steven Walden
11th Jun 2020 14:16

From my side there are a few challenges:

1. CX is not the same as CSAT, NPS, CES. The return on the experience the customer has is akin to asking the question what is the return on innovation. It depends on what you do and how you do it. So generalised measures are wrong.

2. Linear relationships are incorrect (ref CAS). Sure they may work ok at lower ends not well beyond that: (see my book)
(1) the line itself can move, generating different values - 'what any number means in terms of quality' is not a stable function - an 8 out of 10 can in Aldi can mean something different to an 8 out 10 in waitrose - same number different relationship to value.
(2) emotion theory shows us that emotions are for learning, once learnt things tend to fall to a norm for the good or service in question (ref: Baumeister, Lazarus).

3. Scale psychology. For instance, gifting and gaming comes in to play with scaled data

Thanks (1)