Study Cites 40% Of Conversion Rates Miscounted With Antiquated Attribution

A cross-device measurement study released Tuesday from Criteo suggests that consumers convert more often than previously believed, and that marketers risk miscalculating key metrics and return on investments for one in three transactions when applying old attribution models.

It turns out that measuring conversions by device, rather than user, can be misleading. Typically these conversions are undercounted by 40%, according to Criteo's State of Cross-Device Commerce Report.

Consumers use multiple devices in the buying journey, from searching and browsing to buying. The focus should remain on the consumer, not the devices. In an attribution model that focuses on devices rather than people, traditional analytics tools provide a partial and distorted view of a multi-device journey.

By focusing on the consumer, the attribution model analyzes a cross-device measurement approach that can accurately follow the buyer across devices and capture the behavior and the intent of the consumer rather than the devices.

As this report shows, it is highly informative for marketers to understand the full view of their customer's journey via cross-device measurement," said Miriam Newton, VP of product marketing at Criteo. "This is key to enabling multi-touch measurement and subsequent attribution. Because attribution is so complex, we have seen an increase in customers who use more than one attribution model, including last click, to inform their marketing investment decisions."

The analysis in the report, which outlines results from the second half of 2016, suggests that correctly applying cross-device measurement reveals a buyer's journey can be 41% longer, for example.

While Criteo's prior reports focused on mobile commerce, this one reminds advertisers -- mainly retailers -- of the need for a deeper understanding of cross-device consumer behavior and the importance of optimizing the user experience at every step in the consumer journey to purchase. It will increase the likelihood of conversions on any one of the devices being used.

Some 31% of all transactions involve two of more devices from the Web site to the point of purchase, according to the study.

The research also shows that when applying cross-device measurement tactics, conversion rates rose up to 40% higher than when measured with a per-device view method, rather than user-focused method.

Cross-device measurement enabled retailers to more accurately track conversion and optimize marketing spend. Those using the strategy observed conversion rates at 1.4 times higher in the U.S. than seen with a one-device approach when using cross-device measurement.

Interestingly, this attribution method confirms that shoppers now spend $80 on smartphones for every $100 spent on desktop per transaction -- up 7% year-over-year.

The report also suggests 25% of all cross-device transactions completed on a desktop started on a smartphone, and 35% of all online purchases in the U.S. were completed on mobile, up 20% year-over-year. Order values on both smartphones and tablets rose. Smartphone spend increased 7% year-over-year. Mobile apps captured 55% of transactions, compared with 45% for mobile browsers.

2 comments about "Study Cites 40% Of Conversion Rates Miscounted With Antiquated Attribution".
Check to receive email when comments are posted.
  1. Robin Caller from LOLA GROVE, February 28, 2017 at 4:36 p.m.

    Very timely article. 
    Sure, we all know click based attribution is broken. 
    But too many people have to protect the company line and the entrenched view. 

    But Laurie,  what if we showed you exactly how attribution using user-focussed methods could also miss 35% of the sales? Come and visit the best kept secret and best story in London,  and we'll show you. 

    Let me know.

  2. John Grono from GAP Research, February 28, 2017 at 5:59 p.m.

    What is really antiquated is still doing attribution models just on digital interactions.   There is a BIG world out there, that just may have an influence too.

Next story loading loading..