Wednesday, May 16, 2012

The Simplicity and Agility of Zero-Upgrades in Cloud ERP

I am coming to the conclusion that a primary benefit of cloud ERP is the reduction or complete elimination of version upgrades. This observation was reinforced again this week in my one day attendance at the Plex Systems user conference in Indianapolis. Plex is a great example of what a cloud ERP vendor can accomplish by taking what I call a “zero upgrades” product strategy.

The Goal: Zero Upgrades

Originally founded in 1995, Plex went through a complete product overhaul in 2001, when it completely rewrote its ERP system as a cloud offering. At the time, NetSuite was the only other product that came close to cloud ERP and even then, NetSuite was largely a financials-only service.

Interestingly, in my interview this week with CEO Mark Symonds, becoming a “cloud ERP” provider was not their primary goal, but rather a means to an end. Passing through the client-server era, Plex grew tired of the difficulty in getting customers upgraded to new versions and rolling out patches and fixes to its installed base. The move to an online system (Plex Online)—the term “cloud computing” had not yet been coined—was the means by which Plex would to solve this problem. Customers would not have their own installations of the system. Rather they would access one central instance of Plex, which would be developed and maintained directly by Plex. Customers would never have to upgrade.

Implications of Zero Upgrades

Actually, I've known about the Plex approach for some time. But the implications of this strategy became more apparent as I sat through the keynote and some of the individual workshops.
  1. You Like It? You’ve Already Got It. A good part of the opening keynote included the announcement and live demo of a new embedded report writer, called IntelliPlex. Now I wouldn’t say that the demo blew me away. It's good. It has an easy drag-and-drop method to allow users to create their own reports, generate charts and graphs, calculate new columns based on formulas, produce pivot tables and cross-tabs—all good stuff. Is it as powerful as the embedded BI capabilities that other vendors have demonstrated recently? In some cases, no.

    But here’s the catch. Every Plex customer watching that demo knew that they could immediately log on to their Plex system and have access to that report writer. They don’t need to order it, pay separately for it, install it, or be on a certain version of Plex to use it. Perhaps that’s the reason I didn’t see a single person walk out early from that keynote.
     
  2. Functionality is Front and Center. I attend a lot of vendor conferences. Many of the sessions are taken up with subjects such as “Planning for Version X,” “What’s New in Release 7.2.345b,” “Prerequisites for Migrating Product X to Version Y of Database Z.” The Plex conference has none of these tactical, infrastructure-type subjects. All attention is on what the software does, not what you need to do to get it to do those things.

    For example, I sat in on part of a presentation on work center production scheduling—not a subject that I would consider a major draw. In much larger vendor conferences, I might see 20 or 30 people in this sort of presentation. But, as shown in the photo nearby, there were about 150 (out of 800 total) conference attendees in this session. Because Plex users do not have upgrades to deal with and plan for, they can devote all of their time to learning how to use the functionality that they already have access to. There appeared to me to be a much greater percentage of line-of-business users than I see in many vendor events.
     
  3. User-driven Enhancements. Plex’s approach frees it from having to spend time managing multiple versions of its product, creating sandboxes, and phasing in customers from one version to the next. This gives its developers more time to work on product enhancements, which are largely driven by customer-funded requests. Although one customer may fund a change, all customers have the option to “flip the switch” and use it if they so choose, without having to schedule a version upgrade. All new functionality is delivered with the switch set “off,” so that individual customers can choose what and when to implement it.
Ultimately, the zero upgrades strategy enables business agility for customers. This point was stressed to me in an interview I did with Plex customer Ben Stewart of Inteva. His company has a growth-by-acquisition strategy and needs to be able to bring new plants and new locations on-board quickly. Plex’s zero upgrade approach and rapid response to his change requests (e.g. enable a new EDI partner) supports this strategy of agility to accommodate rapid change. Other customers report Plex implementations of new plants in timeframes of weeks, not months.

Are There Downsides?

I have discussed the zero upgrades approach with other cloud ERP vendors, and many of them disagree. They maintain that ERP is different from CRM or other non-critical applications, that cloud customers want control over their environments, that they want to choose when to upgrade and to be able to regression-test their business processes against new versions. In some cases, I think this is simply a legacy of the on-premises world: we’ve always released new functionality as version upgrades. In other cases, I believe that this position is taken for the vendor’s convenience, especially when their cloud ERP systems are using the same code base as their on-premises systems. Because the on-premises customers have to have version upgrades, the cloud ERP customers of the same system must also have version upgrades. Otherwise, the two classes of customers do not have the same code base.

This does not mean I am against the hybrid model—allowing a customer to run the same system on-premises and in the cloud, or to go from one deployment model to another. I have written that there are some advantages to the hybrid model. But forcing customers to do version upgrades is not one of them.

There is one scenario, however, where version upgrades are desirable, and that is in a regulated environment, an area I have some experience in. For example, current US FDA regulations require pharmaceutical and medical device manufacturers to demonstrate that they have control over software configurations that are used to support regulated processes. This does not necessarily rule out use of cloud computing, but it does make it difficult to claim that the user has control over the system if the vendor is changing it on a daily basis. In such cases, it is easier to defend the use of a cloud ERP version that is frozen in its configuration, where the customer can choose when to upgrade and can run testing to confirm acceptance of the new version prior to upgrade. In non-regulated environments, however, I believe that the Plex practice of delivering new functionality “with the switch turned off” is better, as it promotes agility.

Some may argue that the Plex approach may lead to bugs being introduced into the production system on a daily basis. For example, Plex may implement a new feature in one part of the system and it may affect processing in another part of the system—even though a customer may not flip the switch for the new functionality. One long-standing Plex customer indicated that this does happen from time to time, but still he is strongly in favor of the zero-upgrades approach. I would add, I have seen many cases where traditional on-premises vendors ship code that notoriously bug-ridden, where they are shipping “patch releases” for months, even years, later. At least with the Plex approach, when bugs are discovered, they can be fixed in a few hours.

One final issue has to do with the practice of letting customers drive new enhancements. This approach may have worked well when Plex was small, but I question its wisdom as Plex scales. If uncontrolled, this can lead to many one-off enhancements being introduced into the core system, which only pertain to a single customer.

Fortunately, I heard two things during the conference that mitigate this problem. One is that Plex is establishing a formal product management function to review and clear all customer change requests and to evaluate which ones have merit for multiple customers. Second, Plex has introduced a web services platform capability called VisionPlex, which allows customers and partners to develop their own enhancements to interoperate with Plex, but outside of the core systems. This capability is just being rolled out in several pilot projects, but if successful it will go a long way toward keeping one-off enhancements out of the core system. It also has the benefit of enabling an ecosystem of Plex partners to build on Plex as a platform—something that has been lacking to date in Plex’s strategy.

Plex is not a large cloud ERP vendor, having only about 750 customers. It is narrowly focused on a few manufacturing industries, such as automotive, industrial products, plastics, electronics, and a few others. However, it is showing strong and steady growth—30% revenue growth in 2011 and expecting 20% growth in 2012. Furthermore, it is an existence proof for the principle that a zero-upgrades product strategy has major benefits for both customers and the vendor.

Related Posts

Plex Online: Pure SaaS for Manufacturing
Computer Economics: Cloud Players Storm the Gates of ERP
Key success factor for SaaS suites: functional parity

2 comments:

Mark Symonds, Plex Systems said...

Frank, let me first thank you for attending Plex Systems' 11th annual users conference, PowerPlex. As we highlighted during the keynote, it was our largest customer event yet, with more attendees, more sessions, more partners and more community-building opportunities than any PowerPlex event to date.

Your observations are right on the money. Our adoption of SaaS was a means to an end - allowing our developers to produce more code that customers wanted, in less time and with the benefit of getting it into users' hands extremely. One of our best, and toughest, customers came to a realization some time back that prompted her to write a letter to our team.

She wrote, "This past Fall while preparing my budget for the new fiscal year, I was struck by something quite awesome. We have been using our ERP system for three years. Time for a patch? No. Time for an upgrade? No. Need to upgrade interfaces? Again no."..."We chose the software [Plex Online] because it was a good fit for our business, not for the SAAS delivery method. Yet, here we are realizing that had we traveled a different road, perhaps turned left instead of right, we would be starting to go through the cycle again. Instead we are focused on new opportunities, not living on the road that goes in circles. As we open new offices, the only ERP requirement is an Internet connection. That is the power of Plex."

Your point about medical device manufacturers is well informed and it is being handled successfully by our customers. While the software is improved daily, customers have the ability to “flip the switch” or "opt-in" to new features. This allows them to properly document the process change before it is activated, test it, and document the results, in conformance with ISO 13485 requirements.

Our unique approach allows our customers to use an ERP solution that matches their business needs much more closely -- at all times. The model reduces the cost and complexity to consume change. Our community of users, partners and employees all contribute to make a product that I am extremely proud of.

Thank you for your review. We appreciate your keen observations and informed perspective.

Matthew King said...

VisionPlex will be a catalyst for Plex to hit the big time in my view - if system performance is good (speed can be an issue when using Web services). Essentially a PaaS edge to complement a SaaS core.

One of the reasons SAP ECC is so popular is that functionally can be tailored and extended to accommodate the diverse and unique requirements of different organisations around the world. But at huge expense. SAP seems to have recognised this need and is now promoting new tools and technology that allows third party vendors and their customers to more easily build unique and innovative functionality. I see VisionPlex as being analogous to these tools and technology.

On the topic of upgrades, the switch on concept sounds great. But I'd like to know if new and old functionality can be run in parallel for a time (such as two ways of viewing or editing the same data). This would allow a progressive rollout of changed functionality in larger organisations.

And I imagine there is a distinction between changes that are controlled by the switch concept. Structural changes to database schema or tables would need to be more carefully planned more carefully (a formal notification to customers perhaps) than changes that affect user interaction.

Another approach might be to have two multi tenant systems instead of one, with the first system hosting the smaller customers and the second system hosting the larger customers. The changes then being applied to the first system a number of weeks before the second system receives the same updates. Yes this would add administrative overhead, but it might address risk issues considered important to larger organisations.