I have written about Celonis before and in some depth. Still, after a couple of recent conversations between Alex Rinke, co-CEO Celonis, Phil Wainewright, and myself, I've been struck by Rinke's rhetoric that positions Celonis as a service provider capable of delivering exponential value from existing software investments. The arguments Rinke deploys are compelling yet raise important questions about enterprise systems landscapes, their evolution, and present-day use in the context of a world of abundant technology. To understand this better, we need to step back in time.
What did we all miss?
ERP failure stories are plentiful. Many of them describe a litany of missteps, often characterized by a combination of internal and external mismanagement plus a good degree of scope creep. The bottom line is that cost overruns are normal, and expected outcomes are rarely achieved. The latest culprit to get attention is the evergreen topic of managing change.
But as I've listened to hundreds of case studies over the years, both good and bad, one thing stands out, and it goes back to questions that were initially raised during the ERP boom days of the 1990s. Where was the value supposed to be found in the first place? That was supposed to be answered by reference to the initial business case. Still, in all the years I've followed this space, I cannot recall a single case where the post-implementation analysis was carried out that compared the initial business case with the outcomes. As was told to me by a well-known ERP analyst back in the day, it is a very brave CIO who revisits the initial business case and undertakes that analysis. Why?
The illustration above puts it well, but it is only a start.
Phil reminded me about the heady days of Business Process Reengineering as espoused by Michael Hammer and as used by SAP as the cornerstone for making the business case of its then R/3 product. I can't help but think there was something fundamentally missing, the extent of which is only just starting to be fully understood. Hammer's argument was simple:
Most of the work being done by businesses do not add any value for customers, and this work should be removed, not accelerated through automation. He proposed that companies should reconsider their inability to satisfy customer needs and reengineer their processes.
The burning question - did it happen? Kind of, but not really.
During that period, I routinely looked for signs that ERP and other technologies improved corporate performance, largely without success. I can only recall a handful of cases where CEOs on earnings calls ascribed success to a named software vendor, although those same reports were quick to point the finger of blame at vendors when things went wrong.
As I've thought about this over the years, it's apparent that in many cases, organizations fell short of Hammer's ideal by some margin, choosing to digitize existing manual processes through replication coupled to a veneer of change and automation. Rinke argues that today, enterprise systems landscapes are incredibly complex, involving multiple systems, and therefore the potential points of failure are multiplied. Other than scale, I'd argue that's nothing particularly new except that the discovery mechanisms at our disposal are now good enough for us to clearly see where the real process problems lay.
Examples cited by Celonis include defective OCR, the pervasive use of undocumented spreadsheets as workarounds, manual entry and re-entry, and business process logic errors - to name a few. They are what Celonis describes collectively as process gaps. Here are some examples from a recent Celonis presentation about accounts payable.
Ironically, AP was one of the most frequently cited examples in Hammer's studies and, in particular, the Ford Motor Company where it was said that:
Michael Hammer proposed something radical to Ford: Eliminate the invoice. This meant a buyer would no longer need to send a copy of the purchasing order form to the creditor administration. Instead, a buyer would log an order in the online database. When the items appear at the store, the storekeeper would check they corresponded to the purchase order in the system.
This new system used information technology to change the accounts payable process. It implemented an invoice-less process. The purchasing department would create a purchasing order and update it in the database. As soon the materials arrive, a warehouse worker would check the materials received and confirm delivery. Next, payment would be made automatically without waiting for an invoice from the vendor.
Ford benefited drastically from this change. It resulted in almost a 75% decrease in personnel in the accounts payable department.
As I listened to Rinke talk about these many problems, I experienced a creeping sense of déjà vu. After all, hasn't the answer to all business problems over the last 30 years been a matter of throwing yet more technology at the problem that somehow magically solves whatever pain your business experiences? Rinke acknowledges that but references the tangible savings Celonis customers have found. Check this selected list.
In Celonis' world, the enterprise cannot expect to optimize business processes without first mining those processes to discover the execution gaps. In Rinke's words, process mining is 'foundational.' Once that work has been done, enterprises can then move on to considering the best execution strategies through a combination of machine learning infused recommendations and human intervention. This has clear implications for SAP customers, particularly following that company's proposed Signavio acquisition and was a topic raised in an earlier conversation between SAP ecosystem members and Juergen Mueller, CTO SAP.
At the time, Mueller said that while SAP wants to own the process mining element of the equation, it was far from clear how that dovetails to Celonis EMS. For example, I can't imagine existing joint SAP/Celonis customers splitting out their investments in process mining simply because SAP has made an acquisition. And while I admire SAP's ambition to provide best practice end-to-end processes, there is no getting away from the fact that SAP exists within an expanding ecosystem of applications, some of which it integrates into, and others where that work is outstanding. As Rinke said:
If your way of working now embraces services like Slack then you can't put Slack on top of an SAP system but you can integrate Celonis with Slack.
Another way to look at this is to acknowledge that while SAP remains very important, it is a shrinking part of a modern business applications landscape's totality.
A deep dive into AP
In the demo. Celonis talked about invoice duplication, a common problem. I asked whether its EMS system would flag potential duplicates in a situation where goods are bulk ordered and subject to call off and where the only real change would be delivery date and invoice number. Here, the company stumbled a little, ignoring the fact I'd deliberately omitted to talk about different GRN numbers but it is a good test case for retailers who might bulk order from CPG suppliers for later distribution via central warehouse operations. To its credit, Celonis was quick to point out that no system is perfect and that in the initial stages of process improvement discovery, there will be an inevitable degree of manual intervention and process rule checking.
As we worked through examples, Celonis said that its EMS could act as a layer above existing systems and so not touch existing processes. Rather Celonis EMS acts to optimize based on the customer's strategic objectives at a layer of abstraction that does not insist upon process change. Instead, its suggested actions can become templates for future automation, possibly augmented by RPA. That's fine as far as it goes, but I argue that where business process rules are defective, then change is required at some point. But logically, the way Celonis can help the enterprise begs the question for the long term that Phil raised:
If you can both save large amounts of money while essentially remediating broken processes, what does that mean for the future of the underlying process itself?
That's an interesting point because it brings into question the necessity for undertaking upgrades anytime soon for processes that can be improved without going through the pain of an upgrade cycle. Instead, it allows enterprises to consider alternatives at a pace that makes sense to them while generating the kinds of value that Celonis showcases.
But even if you buy that argument, how can Celonis guarantee the kind of exponential value it promises. Rinke said:
We talk to the CFO. They're the people who are most interested in saving money by any means possible, whether that's improving working capital or reducing rework. They understand very quickly and we now have hundreds of use cases they can check out. And you saw that we say if we can't save you $10 million, then we'll donate $100,000 to the charity of your choice.
And that is the trick to getting into the enterprise in ways that CIOs cannot coupled to a meaningful commitment.
The demo played directly into my finance background with examples to which I readily relate. Those examples raise questions but I was impressed by the candid nature of the company's responses. Rinke said:
Our intelligent algorithms go a long way but they do not necessarily provide a definitive answer. Human intervention is necessary, especially as you build confidence in the results and have those reflect the strategy you want to follow. But then we are only at the beginning of this movement to radically improve processes. We already see a growing ecosystem of partners who can help with that journey. Incremental change is not working. It's not what customers need and they know it.
As we closed out our conversation, I could not help but continue to admire the conviction with which Celonis presents its value proposition. While I focused on the SAP angle, Celonis is also optimizing Oracle, Salesforce, and ServiceNow processes. On that basis, you could be forgiven for thinking that with such a broad range of applications it can address, Celonis is something of a Swiss Army knife for process remediation. Time will tell if that is true, but then there are clear warnings for those many SAP customers who have yet to take on a S/4HANA migration project. The last word on this goes to Rinke:
Using S/4 as a way to simplify and reduce customizations while putting your application into the hands of a hyperscaler doesn't make you a cloud-based business and it doesn't necessarily solve the process problems for which an EMS is designed. You still need that. You can't easily integrate a lot of the newer classes of application that meet day to day needs so what have you really got?