CLV & Customer Intelligence
Channel performance can distort CLV if you read it too quickly
Retailers often compare customer value across channels and assume the answer is obvious.
That question matters even more now that *Gartner says digital channels account for 61.1% of total marketing spend.* As more investment flows into digital, the risk of overreading digital customer value also increases. More spend means more attribution, more visibility, and more confidence in numbers that may partly be artifacts of how the channel is measured rather than how the customer is behaving.
I have watched one channel get celebrated in a strategy review because the spend per customer looked higher. Then returns, discounts, and retention quality were layered in, and the picture changed enough that the winning channel was no longer obviously winning. That was not a data error. It was a framing error — the initial view rewarded what was easiest to measure.
This is the trap. Channels differ in how cleanly they surface customer data. The channels with the cleanest data are usually the ones that look strongest in CLV rankings. That is not the same as having the strongest customers. Sometimes it is just the strongest measurement.
A few patterns show up repeatedly:
That is why CLV should not be used as a vanity ranking tool. It should be used as a decision tool:
The retailers who get better at this tend to do one thing differently. They treat channel comparisons with mild skepticism by default. They assume the model is understating or overstating something until proven otherwise. That habit alone prevents a lot of expensive overreactions.
I go deeper on this in the full article, especially why channel rankings get shaky once real economics are layered in.
- E-commerce customers look stronger.
- Store customers look less engaged.
- Marketplace customers look different again.
- Different acquisition costs.
- Different promotional pressure.
- Different product mixes.
- Different return patterns.
- Different levels of customer identification.
- Different margin structures.
That question matters even more now that *Gartner says digital channels account for 61.1% of total marketing spend.* As more investment flows into digital, the risk of overreading digital customer value also increases. More spend means more attribution, more visibility, and more confidence in numbers that may partly be artifacts of how the channel is measured rather than how the customer is behaving.
I have watched one channel get celebrated in a strategy review because the spend per customer looked higher. Then returns, discounts, and retention quality were layered in, and the picture changed enough that the winning channel was no longer obviously winning. That was not a data error. It was a framing error — the initial view rewarded what was easiest to measure.
This is the trap. Channels differ in how cleanly they surface customer data. The channels with the cleanest data are usually the ones that look strongest in CLV rankings. That is not the same as having the strongest customers. Sometimes it is just the strongest measurement.
A few patterns show up repeatedly:
- Digital channels often look better than they are because identification and attribution are tighter.
- Store channels often look worse than they are because identity capture is thinner and some of the best customers are partially invisible.
- Marketplace channels often look efficient in one view and weaker in another, depending on how fees and repeat potential are handled.
That is why CLV should not be used as a vanity ranking tool. It should be used as a decision tool:
- Which channel really delivers the most durable customer value once full economics are included?
- Where is the business most likely to be overinvesting based on flattering measurement?
- Where is it most likely to be underinvesting because visibility is poor?
The retailers who get better at this tend to do one thing differently. They treat channel comparisons with mild skepticism by default. They assume the model is understating or overstating something until proven otherwise. That habit alone prevents a lot of expensive overreactions.
I go deeper on this in the full article, especially why channel rankings get shaky once real economics are layered in.