OneStream in the new ML-powered CPM World
- CPM/EPM vendor OneStream had its Splash user conference in Washington D.C. last week. The show had a lot of content re: Sensible ML, ESG reporting and the Splash product marketplace. Sensible ML was clearly the top attention getter of the show.
OneStream is a corporate/enterprise performance management (CPM/EPM) software vendor. They have applications for:
- ESG reporting
- Financial consolidation
- Planning and budgeting
- Lease accounting
- Statutory reporting
- Management reporting
- Cash forecasting
- Long-range planning
- Account reconciliation
- Tax provisions
- Operational planning
- And more
The company now has approximately 1,150 customers (around 250 were added in the last year). A significant concentration of these wins consisted of larger enterprise customers. This year’s Splash conference had around 2,200 in-person attendees and a number of virtual attendees, too. One executive speculated that virtual attendees probably numbered in the millions. That comment got the staid, mostly accountant/finance crowd laughing!
In a rarity in the application software world, OneStream publishes a list of all of its customers and makes this available to prospects to contact. All customers ‘must be referenceable’.
(One sad note, though: While last year’s Texas-based show featured an abundance of BBQ brisket, breakfast tacos and Dr Pepper, this Washington DC event offered a very different (but still acceptable) fare.)
OneStream executives have spoken about new AI/ML tools and capabilities for a few years now. At the 2021 (virtual) OneStream user conference, I noted how CEO Tom Shea spent time discussing how:
advanced technologies like AI/ML will impact companies in general and how these capabilities will impact software like the kind that OneStream sells. (see full diginomica article here)
At that event, Shea indicated that firms were passing through an inflection point where the expected value from tools like ML/AI was now approaching what could actually be realized. He also highlighted skills issues customers might experience around advanced technologies (I concur) and how early tools had shortcomings (e.g., poor transparency in how models/algorithms worked or bias in the key datasets used).
At that show, I also noted that:
OneStream previewed a series of solution capabilities to address modern business needs. Some capabilities include the ability to create thousands of ML models in parallel, reduce a user’s dependency on data scientists, and make AI/ML tools explainable to managers, auditors or regulators, etc.
The next year, 2022, in San Antonio, customers got to hear from actual customers using these new tools. In other words, the ML concepts discussed in the prior year were already incorporated into OneStream’s product line and customers were already using them. That show writeup noted that:
OneStream did put together one great joint OneStream/customer presentation re: their new Machine Learning capabilities. The customer was Polaris, the outdoor vehicle maker, and they are using new OneStream products to better predict demand for their various outdoor products.
This year’s event saw significantly more presentations, Q&As, etc. for customers to hear from other customers using these advanced tools to identify better, more likely sales forecasts. I personally caught three of these where I saw:
- How customers can select internal and external data files to improve the accuracy of the forecasting tool. Different external files (e.g., weather data, holidays, consumer price indices, etc.) are tested to see how well they help the models correlate with prior historical data.
- How actual customers have spent the last year improving their prior models with better data, more granular data and external data.
- How finance and operations personnel have to team up to more effectively use these advanced technologies. In fact, OneStream executives made a point to note that old CPM often was a transaction processing or consolidation tool that presented prior period financial results and operations systems were designed to record a lot of business events/product movements/etc. Neither system was designed to do a great job of analyzing granular, external and internal data. If they could, companies could have more accurate forecasts and better manage their businesses.
OneStream’s Shea also used some of his opening keynote time to describe how recent innovations in large language models, ChatGPT, etc. have left him pondering a number of thoughts re: how and what applications will be in the coming months or years. For example, he showed how an AI tool trained on OneStream’s documentation can answer far more specific questions, offer multifaceted answers, etc. than any prior generalized chatbot. On stage he asked the utility to show how a customer could create new working capital functionality in one of their CPM applications. The software didn’t just return a canned response. Instead, it formulated a 5-step method for him to follow and even suggested some highly accurate code to insert in the recommended solution.
This example was not some simple query-response chatbot (that had a number of canned responses to questions that the chatbot might encounter) or a search engine result. It had deduced what a correct approach might be after it had scanned, indexed and correlated all of OneStream’s product manuals, user guides, etc.
Later, a small group of analysts did get a chance to chat with Shea further. Approximately 40 minutes of their 1-hour discussion focused on Shea’s rapidly evolving assessment of the new market for AI-powered applications. More specifically, he stated that:
- OneStream will likely incorporate and use some of the newer AI services on the market today (e.g., large language models) as these have already been developed. Creating their own versions of these would take time and a lot of capital to acquire the compute power to develop these. These tools would provide services that OneStream might use as an adjunct for its application software. For example, Tom indicated that he could have one of these services translate all of their documentation from English to Mandarin for pennies of what it would cost in prior years.
- Older OneStream customers that are still running on-premises versions of OneStream might not transition well to these newer AI-powered applications unless they are willing to place some of the OneStream solution set in the cloud. Why? The computing horsepower needed to do AI-enabled forecasting, for example, may require huge datasets and lots of computing power. This type of application is perfect for a scalable, on-demand technology. A customer’s internal IT resources may be not be up to the task and it may not be prudent from a capital deployment perspective to make a big tech investment for tools that have lumpy consumption requirements.
- He sees AI as a big bet and one that vendors must to well and in a hurry. Transaction processing will become table stakes while the new AI-powered applications, insights and analytics are where the new, big value will originate. My shorthand for this is that transaction processing will represent motherhood, apple pie and the American way while AI-powered tools will be sex, drugs and rock’n’roll. The fees and fun may be in the latter camp, folks!
OneStream’s ESG tool is part of the OneStream application suite, is built on the OneStream platform and leverages that platform well. The software can integrate with financial and other systems via a number of pre-configured integrations.
In listening to customers interact with the OneStream personnel, it would appear that:
- Most customers are still trying to collect Scope 1 and 2 emissions on an annual basis. They are not currently capturing detailed operational data and, as such, lack the tools to determine which products, batches, plants, etc. are more/less green than others.
- Customers are trying to capture a number of social data elements so that these can appear in their firm’s annual, regulatory and other reports. Much of that data is not generally kept in a financial system. It might be found in HRMS, EH&S, time tracking and other applications though.
The current methods many companies are using to track ESG data (i.e., most notably spreadsheets) are definitely showing their shortcomings today. For example, all that one had to hear is that a firm is in its second year of producing sustainability reports and their team is getting swamped with a ton of information requests. What’s triggering these requests are that executives, shareholders and others want variances from one year to the next researched and explained. That analysis, in an overly summarized, non-granular environment can take days, weeks or months to uncover what are the drivers of these differences. While OneStream’s ESG application and architecture could help with a drill-down, drill-around capability, it’s the customers who will likely need to add more intelligent meters, sensors, etc. so that great data is collected. Allocated, generalized and highly summarized information is the enemy of great analysis and variance assessments.
(Again, for more on OneStream ESG, see this article from 2022)
OneStream also promoted their 3-component marketplace. Customers can acquire solutions from software partners, implementation partners and independent consultants/developers. OneStream’s has put guardrails in place to ensure products and providers are properly vetted before being made available to the customer. Products and vendors are also re-evaluated periodically to ensure solutions remain current, secure, etc.
It appeared to me that this ecosystem has grown in size and importance. The expo hall at the show was a proof point for this.
OneStream’s leaders have clearly seen the potential for machine language/AI and other advanced technologies. While executives at other companies have also expressed their interest in these tools, OneStream may be the most energized of all of the vendors that I’ve spoken with lately. In contrast, one vendor I met with this week was taking a very conservative, slow, measured approach to using these tools with some of their applications.
I’m not sure playing the ‘caution’ card when a market is making a fundamental shift is always prudent. (I’m also not advocating playing fast and reckless either.) Why is the caution card a challenge? Customers rarely shop for new application software if all of the new products look, act and behave like their existing apps. The value proposition for a customer to make a software change when there is little competitive advantage to be gained is pitiful. However, when new (or old) vendors pop up with highly innovative, value-creating solutions, customers will start buying in droves. That creates a market for newer/innovative vendors and this lets them take market share away from stodgy older vendors. This is when major shifts in market share occur. Smart vendors will get in front of the AI/ML/LLM/etc. train and get all of the market share possible (i.e., get the low hanging fruit NOW!).
And, yes, I did run into a number of traditional OneStream customers (i.e., accountants, finance people, etc.) at the show. Some were there to pick up tips/tricks to get more out of their prior OneStream purchases, some to party in Washington DC, and a surprising number were there to find out about ESG. I suspect ESG will be a big topic at other shows for some time to come. However, the current market interest in AI/ML/ChatGPT will likely crowd out/drown out some of the ESG content for a time.
One final thought, OneStream plays close to the vest with product roadmaps and visions. I suspect that next year’s show will offer up a lot more clarity around their Sensible ML/GPT plans. That should be really interesting to hear.