The Author: Anne Golombek
Anne Golombek is COO and Marketing Lead at minubo, the Commerce Intelligence Company. As an expert in data-driven commerce, she is one of the initiators of the Commerce Reporting Standard project.
It has been a really successful proof of concept: The first CRS workshop, having taken place on Thursday, December 7th. Right in the heart of Hamburg’s academic focal point when it comes to digital commerce, the University of Applied Sciences Wedel, 12 commerce reporting experts from companies like Project A, the Otto Group and REWE vertical ZooRoyal, spent a day full of productive discussions about the very core of commerce reporting: transaction metrics. With that, we do now have a solid basis for further discussions on metrics and KPIs for single stakeholders and departments. Before we’ll process and publish the results of those discussions on a detailed level (probably in January), we’ll now share a high-level summary of some key discussion topics here in our blog.
Differentiation of Cancellations: System-wise vs. Customer-wise
As there can be huge differences in handling the one and the other, one point of discussion has been the question of whether to split the cancellations column in the transaction metrics matrix up into two levels: system-wise cancellations and customer-wise cancellations. Especially for companies that build entire processes on the differentiation of who or what caused a cancellation, this differentiation can be crucial for reporting purposes. Also, we decided to put some thought into terminology for the new levels as “cancellations” might be broadly associated with the customer-wise kind of cancellations only.
Alternative Breakeven Analysis
So far, we suggest to run breakeven analyses as indicated in the transaction metrics matrix – that is to say: on the level of order value and revenue (= after discounts). During the workshop, we discussed that a second, alternative approach should also be considered best practice as it is relevant for several use cases as well: Running breakeven analyses on the level of GMV (gross merchandise value). E.g. businesses who handle discounts as marketing costs might consider this the more meaningful approach.
Separation of Taxes
The placement of taxes within the transaction metrics matrix is quite arbitrary so far – actually, it’s consensus that companies should be able to account for transaction metrics including taxes on all relevant levels, but departmental stakeholders usually don’t want to see taxes within their KPIs and reporting at all. Therefore, we decided to take the tax level out of the matrix. Instead, we will probably integrate a recommendation on the question for which metrics to consider an “incl. tax” version into the data model.
Cover Marketplace Business Models
As marketplace models gained such an important position within the retail market (and will continue to grow this position rapidly), we also talked about how to make sure that our Commerce Reporting Standard always covers those business models just as it does models of direct sales and came up with some ideas. We will get back to this aspect in one of our next articles.
Handling of Advertisement Subsidies
For both brands and retailers (from completely different perspectives, though) it has always been a never-ending discussion: How to best handle advertisement subsidies in terms of reporting? For our project, we came to the conclusion that there are three possible ways to built them into our model: 1) Handle them completely separated from our core systematization of transaction metrics; 2) Work with prognoses, i.e. expected values; 3) Directly set them off against purchasing prices. To keep our Commerce Reporting Standard framework an actual standard, the plenum leaned towards version 3.
Joint Reporting View of Discounts and Marketing Costs
As mentioned above, the necessity of considering discounts as marketing costs for particular reporting cases has been a discussion point during the workshop. Ultimately, consensus was to include a recommendation into our model on how to a) keep discounts separated as an important control element most of all for daily business decisions (like they’re already handled in the matrix today), but at the same time b) make a joint view of discounts and marketing costs available e.g. for high-level reporting that makes the total investment in marketing and promotions visible for strategic decision-makers or for an alternative calculation of single metrics like e.g. customer acquisition costs.
More Complex Logic for Open Orders
We also talked about open orders and briefly touched on the question, if (similar to the topic of splitting up cancellations into two types) a more complex logic should be considered for our matrix than just having one “open orders” column. The discussion covered several cases like standard open orders that occur during the short time span between placement and invoicing of an order, but also cases of ordered goods being out of stock or not even produced, leading to states of subsequent delivery resp. subsequent non-delivery, if goods cannot be (re)stocked at all. We’ll probably dig deeper into this question during one of our stakeholder-specific discussion complexes.
Additional Time References for Transaction Metrics
Though there was a broad consensus on order date, invoice date and return date being the most important time references for transaction metrics (heads up: not for commerce reporting in total, of course!), we also came to the opinion that there are four more time references that should at least be considered for inclusion into the basic transaction metrics matrix as well: 1) first touchpoint date (especially for the purpose of cohort calculations); 2) cancellation date; 3) shipment date; 4) payment date. As, for all these dates, different question have to be answered, we’ll take care of those step by step.
Consider Medians as an Alternative to Arithmetic Averages
Last, but not least, we covered the topic of transactional averages and came across the question, if for some cases (or under specific circumstances), the mathematic median could be a more useful kind of calculation than the arithmetic average. We’ll dig deeper into this question during our discussions on the reporting interests of various stakeholders.
That’s it for today – we’ll keep you posted with the detailed documentation of the workshop results in January as well as an outline of the next steps our project will take. If you have any thoughts on the topic listed above, please don’t hesitate to share them in the comment section!