How many press releases do you see that say "X - the world leader in Y is transforming...." and then you switch off? If you're like me then that is about 99.9% of the time. Not so with Celonis that last week launched what it terms an Execution Management System or EMS to give it the catchy TLA. My colleague Brian Sommer has penned his thoughts on the topic, framed in terms of the farce that is the fabled book of record. Sommer comes to the conclusion that:
In today's world, cost savings are a highly desirable outcome but, I'd argue that redeploying people to more strategic activities is at least as valuable. Why? Businesses need to reskill their workforce so that more people are intimately familiar with all-new smart technologies (e.g., algorithms, ML, bots, big data, smart analytics, etc.) like Celonis' EMS. A workforce may also need to reskill as the company's business model is shifting or changed dramatically. And, of course, the company may need people to help fuel its own outsized growth opportunities.
Growth, in my opinion, is the real outcome products like EMS can enable. A value proposition that only espouses tweaks around process improvements doesn't do the solution justice. The smart software buyer will see beyond the initial process savings and realize the potential of the products to enable a company to grow materially without necessarily growing headcount. In doing so, a company's margins will be greater and possibly surpass those of competitors.
Growth is the real endgame here.
The question then comes - does Celonis represent the future that Sommer envisages. Yes and maybe. Let's look at the background.
Mine deep, act on errors
Celonis is best known as a 10-year-old vendor in the process mining space. It has grown to $100 million in annual revenue with ambitions to reach $1 billion inside six years. Process mining isn't new. The concept goes back a good 20 years or more but Celonis takes the topic to an entirely new place with EMS, largely as a result of its Integromat acquisition announced 14th October and which has already been integrated into the Celonis stack.
Historically, Celonis is in the process discovery business. A good way to understand this is to think about how enterprise systems actually get used rather than what was originally designed or implemented. This is one of those 'best-kept secrets' where the help that Betty in the Corner (who knows all about workarounds for crappy process design,) has been giving to users since forever but is now exposed in graphical form.
Depending on the type of system landscape, processes can be used in an almost infinite variety of ways. The purpose of process mining maps that Celonis produces is to help identify problems on the road towards root cause analysis.
A common use case is where a long-run process straddles (say) an SAP and Salesforce environment and where the term 'customer' is defined differently in each and cannot be readily resolved through integration. Using some fuzzy logic, Celonis can match those systems. That's useful when sales (for example) are selling to a deadbeat customer. In an ideal world, the backend AR system would tell the salesperson that the proposed customer is on stop, most likely through their mobile Salesforce app. How often does that happen? Another example is where priorities change and those managing (say) payables need to understand where delays in settlement are occurring. This could be for a variety of reasons such as incorrect pricing in invoices or late entry into the system. Of itself, Celonis does not create processes but claims it can automatically correct errors of the kind just described. This sounds an awful lot like RPA and, on a call with analysts, Celonis didn't push back on that perception. Rather, their take is that a combination of process mining and their take on RPA is creating a fresh software category.
The RPA problem
In theory, RPA solves for manual rote processes but in reality, it's lipstick on a pig. Why? Because the better-known players in that market - think UIPath, BluePrism and the like - need constant attention, especially when processes are changed as happens all too frequently. When a process changes, the RPA breaks because most often, RPA is really just screenscraping by another name and then automating via the UI. As such, RPA doesn't scale. Analyst firm HfS Research, which was early in talking up RPA now says it's a dead end. At least in one definition:
The RPA that "died" is the poorly-defined "RPA" that got hyped up to create hockey-stick growth excitement for investors. It wasn't defined correctly, was a mash-up of desktop automation with pure-RPA (unattended back office) and all the deals that got signed were "attended" so weren't even "robotic".
The pieces of RPA that survive are the process orchestration tools (discover, design, automate and mine) that form part of what we see as the evolution towards "Intelligent Digital Workers" which augment human experiences and help with real customer-to-employee intimacy. Let's also not forget these apps also need to be enterprise-grade, compliant with ITIL and security factors etc. Scale only occurs when the business designs and IT enables... The winners in the future are smart enterprises with leverage technologies to anticipate where their customers are going... often before their customers even know themselves.
Celonis avoids this basic criticism by not touching the process operating at the UI levels but by manipulating the data based on what its machine-learning backed algorithm tells it should be happening, suggesting fixes and then providing the user with the opportunity to let the EMS do just that. To make that set of activities actionable, Celonis is doing two things. On the one hand it is developing applications that mine, surface, analyze and then suggest actions in AP, AR and opportunity management. RIght now the main focus is on SAP and Salesforce systems. On the other hand, it is providing a low-code (ish) studio that partners and customers can use for examining a variety of processes including those that contain custom code. Taken together, the company (mostly) meets HfS preferred view.
So what has Celonis done that takes this set of capabilities to the next stage? Taken to the next logical step, Celonis might be seen as the process improvement engine for which you'd otherwise consider re-implementation of business systems. Does that sound too far-fetched? Yes it does but that's not not necessarily what is/will happen.
I want to be clear that the demo that Celonis put out is super attractive and highly compelling. At first glance it's almost like magic - except I know that there is nothing magical about software. I understand that where late entry dates occur - as an example - then an automated fix is relatively easy and safe (compliance issues aside.) But I draw the line at 'pricing errors.' Immediate questions in my mind go something like this:
- Does the system know about special terms for X customer or customer group?
- If so then should this be classified as a pricing error?
- Where did the data come from that suggests a pricing error?
- Who is responsible for that data and what are they doing?
- Is pricing information entered manually and if so why?
- How come there are so many pricing errors?
- Are they really errors?
- Will I open myself up to lawsuits if I start manipulating prices inside my order/invoice/AP system?
- What are the compliance and governance implications?
- Why are we not looking at this process in more detail to discover the root cause of issues?
Where does this leave us?
The way Celonis presented the solution leaves me wondering if they're trying to be all things to all people. The company has made its bones by, among other things, partnering and appearing on the SAP price list. But it's not sold to every enterprise customer. Is that because Celonis could catalyze an S/4 project at risk of being deferred? We don't know but it's easy to see how that might happen, Celonis also emphasized its EMS store and the opportunity for partners to drop applications there for purchase or bundled as services. It talked about partnerships with the well known RPA vendors. In answer to the obvious question about EMS as a competitor to its RPA partners for example, Hala Zeine, chief product officer said:
I think they do see the need for the insights to continuously feed their RPA, or their automation. So I actually see this will be very symbiotic in a way where they're using our insights. And then it's very important to focus on the point of action. Because sometimes you get an insight and you want to just alert somebody, that is one form of action. Another one is you can assign a task to somebody. And then the third one is you want to automate. How you automate depends a lot on the systems that you have. So if all you want to do is change the price point or change the date of the offer, you can do that very quickly. But if you're running more, let's say scripts on older systems, then you want to trigger the insights into the action in an RPA. So I think there's room for a lot of different ways to automate. But it really depends on the decision and the landscape that the customer has in mind.
Are you buying all of that? Up to a point. If RPA is as HfS described then what Celonis says today must surely be an interim position.
If I have the choice to remediate or create new applications that avoid the pitfalls of screen scraping then shouldn't that be my direction of travel? And, if I'm content with my backend processes in SAP and Salesforce, then why would I consider upgrades or transitions to new platforms when I can mine what's already there and use that data as the basis for both better execution and the development of revamped business models supported by modern processes? And, if I can get all that done in a low code environment, which by its nature is 'build fast, fail fast' then maybe I end up relying on Celonis as my future state partner of choice.
At the very least, I see what Celonis offers today as very much a case of promising quick wins where it matters most: in the management of enterprise level cash. That is provided the demos are representative of what Celonis can do in the real world.
In the medium term I am unclear. As HfS pointed out, RPA has not really automated much and very little at scale, rather it has augmented existing processes, making them more efficient. That has delivered value but it's a one hit wonder. There is obvious value in discovering how messed up your systems really are and getting a bead on what needs fixing can only add to that.
But if the direction is one as envisaged by Brian Sommer, then that's a whole new thing. That view sidelines discussions about what systems need upgrading and allows leaders to focus on the strategically important things that matter. It is however telling that Celonis launched with partners first. They need convincing that the new capabilities introduced through Integromat and Celonis Studio will not mean a sidelining of their existing value proposition. So far, Celonis has been successful in that regard but it has to tread carefully. As one executive told me, promising what appears to be something truly exciting is fine but you have to deliver. That's how you win trust.
In its presentations and in the support materials, you can see what Celomis is trying to articulate - smart, fast, modern, cheap. But can you have it all? That's been elusive in the past but maybe possible tomorrow. But let's be clear. One thing it is not and that's magic.