There is no secret that any e-commerce tracks its sales data through Google Analytics. But how correct and reliable are the sales figures?
We tried to compare the tracking of the most used free analysis tool in the world with the precision segmentation of Rfmcube – let’s find out the differences!
The Method
For our research we have selected a sample of 3 e-commerce stores of different sizes using Rfmcube.
For each, 3 segments have been created on Google Analytics:
- New Customers (transactions = 1)
- Recurring Customers (transactions per user> 1)
- All other clients
Then we analyzed the buying behavior of the 3 segments in 2 different periods:
- last month
- the same month of the previous year
Finally, we compared the Average Receipt (AOV), Order Number and Revenue metrics with the Rfmcube charts to see if and to what extent the numbers differ.
As a result of doing this we have selected an e-commerce with Analytics accounts as “healthy” as possible to exclude gross tracking problems.
Data comparison Rfmcube / GA shop # 1
The first E-commerce examined sells hi-tech consumer products and has a volume of about 700 orders / month.
COMPARISON 1. February 2020 data
| No. of Orders | Revenue | AOV | |
| New customers (Rfmcube) | 471 | 110.794 € | 235 € |
| New customers (GA) | 674 | 131.770 € | 195 € |
| Applicants (Rfmcube) | 283 | 37.923 € | 134 € |
| Applicants (GA) | 48 | 13.235 € | 275 € |
| Aggregated data (Rfmcube) | 754 | 242.564 € | 197 € |
| Aggregated data (GA) | 722 | 145.005 € | 200 € |
As you can see, the numbers shown by Analytics deviate significantly from the reality shown in Rfmcube.
What stands out is the greater attribution of sales to new customers in GA, followed by a negligible number of recurring customers: only 48 are tracked compared to 283 by Rfmcube, with a turnover of only 13.235 euros instead of 37.923 euros as it is supposed to be
Finally, the most surprising data is: these 48 recurring customers (instead of 283), who spent 13k (instead of 37k), had an average receipt of 275 euros, while the average receipt of recurring customers accurately tracked by Rfmcube is only 134 euros.
Having read the Analytics data it may seem that there are very few recurring customers who buy with an average receipt of 80 euros higher than new customers.
Instead the reality that clearly emerges in Rfmcube is that:
- sales of recurring customer are much higher: 283 instead of 48
- consequently, the turnover generated by the applicants is highertoo: 37k instead of 13k
- the average receipt of applicants is much lower: 134 euros instead of 275
- on the contrary, the average receipt for new customers is much higher than what is reported in GA: 235 euros instead of 195 euros
- GA tracked a total of 722 orders instead of 754: an error margin of “only” 32 orders is lower than the only recurring segment
GA data always have a certain margin of error, which tends to deteriorate in the Recurring Customers segment.
The most accurate figure of GA is the aggregate figure of the average receipt – 200 euros instead of 197 euros.
We continue the analysis to see if our suspicions are confirmed.
COMPARISON 2. data February 2019
| No. of Orders | Revenue | AOV | |
| New customers (Rfmcube) | 471 | 106.759 € | 235 € |
| New customers (GA) | 576 | 123.474 € | 214 € |
| Applicants (Rfmcube) | 283 | 41.128 € | 184 € |
| Applicants (GA) | 54 | 13.626 € | 252 € |
| Aggregated data (Rfmcube) | 754 | 147.887 € | 218 € |
| Aggregated data (GA) | 630 | 137.100 € | 217 € |
The data from February 2019 shows a similar trend with far fewer recurring customers tracked by GA than in reality (54 instead of 283).
Compared to 2020 data, we observe:
- a better match in overall turnover volume: 137k instead of 147k, i.e. only 10k difference
- a greater margin of error regarding the number of orders: 630 instead of 754, i.e. 124 orders of difference!
Shop # 2 data comparison
The second e-commerce analyzed sells costume jewelery products is smaller than the previous one with a sales volume of about 200 orders / month.
COMPARISON 1. February 2020 data
| No. of Orders | Revenue | AOV | |
| New customers (Rfmcube) | 156 | 6.240 € | 40 € |
| New customers (GA) | 140 | 6.029,24 € | 43 € |
| Applicants (Rfmcube) | 140 | 7.361 € | 52 € |
| Applicants (GA) | 29 | 1.558 € | 53 € |
| Aggregated data (Rfmcube) | 296 | 13.601 € | 45 € |
| Aggregated data (GA) | 169 | 7.588 € | 44 € |
We find more or less the same discrepancy in the data, which increases severely in the case of Recurring Customers:
- 29 orders instead of 140
- 1.5k euros of turnover generated instead of 7.3k
In relation to the lower numbers, the discrepancy in the aggregate data is even greater than in the previous case:
- 169 orders instead of 296
- 7,588 euros instead of 13.601
Instead we see a good match in the turnover of new customers (6k instead of 6.2k), which is probably because many sales attributed to them by Google analytics are actually attributable to recurring customers.
Once again, the most accurate metric remains the average receipt.
COMPARISON 2. data February 2019
| No. of Orders | Revenue | AOV | |
| New customers (Rfmcube) | 99 | 3.581 € | 36 € |
| New customers (GA) | 92 | 4.394 € | 47 € |
| Applicants (Rfmcube) | 69 | 3.155 € | 45 € |
| Applicants (GA) | 17 | 509,72 € | 29 € |
| Aggregated data (Rfmcube) | 168 | 40 € | |
| Aggregated data (GA) | 109 | 4.904 € | 44 € |
The data of February 2019 confirm the discrepancies of 2020. What is more, there is a significant error regarding the Average Receipt of the Applicants: 29 euros instead of 45.
Comparison 3
COMPARISON 1. February 2020 data
| No. of Orders | Revenue | AOV | |
| New customers (Rfmcube) | 1.297 | 145.918 € | 123 € |
| New customers (GA) | 2.003 | 230.894 € | 115 € |
| Applicants (Rfmcube) | 1.030 | 87.237 € | 103 € |
| Applicants (GA) | 154 | 18.038 € | 117 € |
| Aggregated data (Rfmcube) | 2.327 | 233.155 € | 114 € |
| Aggregated data (GA) | 2.157 | 248.932 € | 115 € |
Here we have the third and final E-commerce of our analysis. Sales volumes are much higher than the previous ones, namely over 2k orders/month on avarage.
The same errors are unavoidable, and we can already begin to consider a constant:
- a significant but not serious difference in the number of total orders (2.157 instead of 2.327)
- an overestimated number of new customers (2.003 instead of 1,297)
- a small number of applicants compared to reality (154 instead of 1.030)
Also in this case, the average receipt is the best comparable value.
COMPARISON 2. data February 2019
| No. of Orders | Revenue | AOV | |
| New customers (Rfmcube) | 1.377 | 145.918 € | 105 € |
| New customers (GA) | 1.832 | 182.311 € | 99 € |
| Applicants (Rfmcube) | 958 | 87.237 € | 91 € |
| Applicants (GA) | 247 | 22.685 € | 91 € |
| Aggregated data (Rfmcube) | 2.335 | 233.155 € | 99 € |
| Aggregated data (GA) | 2.079 | 204.996 € | 98 € |
The situation in the February 2019 report is the same.
Conclusions
GA’s data exhibits discrepancies at various levels, but these are critical when dealing with recurring customers.
The most likely explanation for this discrepancy is quite simple: Google Analytics struggles to determine when a customer is recurring.
This is because the recognition of users in GA takes place through cookies and the tools that the browser makes available (local storage), so this cannot guarantee the user’s identity over time.
Having a hard time recognizing a user as a recurring user (because they deleted their cookies, changed their device, etc.) they will tend to attribute much more sales to new customers than required in reality.
Concerning the orders as aggregate data that Analytics was unable to count, they can depend on several factors such as:
- the user closed the thank-you-page before the Analytics code was activated
- the customer subsequently canceled the order
- by mistake some marketing employee has viewed the thank-you-page because he had to do some work, but has not set an IP filter on GA and is counted as a customer
- backend orders have been generated, for example in the case of closed sales with the help of telephone support
How do I solve these problems? Should I throw Google Analytics in the trash can?
Absolutely not! Google Analytyics remains the best free tracking tool ever, and a powerful ally for analyzing the performance of your E-commerce
Particularly when you have to analyze data such as browsing behavior between pages, or traffic acquisition sources, there’s nothing better than Google that can give you the data you are looking for.
The problems begin when it comes to tracking less “fluttering” entities such as orders and customers.
In this case, as we have seen, common tracking via javascript turns out to be approximate at best, misleading at worst (as in the case of recurring customers).
It is possible to partially fix this if you have a programmer willing to take the time.
GA in fact provides APIs for developers and allows you to significantly improve the quality of tracking through technologies such as User ID and Server-side tracking (go to the Google Analytics developer guide).
But keep in mind that in the free version of GA, even if you have the time and resources to set it up, the tracking problems will never completely disappear.
Although cross-browser and cross-device user recognition is possible (at least in theory), certain limits seem objectively insurmountable, for example:
- the problem of tracking a return or order cancellation remains
- you can not measure backend closed orders (e.g. telephone orders)
- during platform switching periods, it is easy to make mistakes and run into “holes” of orders in the migration period
However, the most insurmountable mistake of all remains one:
- the inability to show retroactive data (and I would also like to see, it’s free!)
This means that even now if you decide to start a more serious tracking on GA, it will always and only from NOW onwards.
As an alternative you can continue to use Google Analytics for the countless valuable Reports it provides, but rely on Rfmcube when you need to accurately measure order and customer data.
Starting from the free trial, you will be able to have an instant and retroactive overview of all your customer history, from the first to the last day of your Store’s activity with 100% accuracy.
Do you think your Google Analytics is better and doesn’t have all those discrepancies seen above?
Challenge accepted! Compare the data of your GA with the data you will find on Rfmcube: if the data is 100% aligned (2020 and 2019 data) you will receive 1 full year of free subscription as a gift!
What are you waiting for? Find out the real story of your customers now ⬇️⬇️
It might also like:
