The principles of customer-centric ERP - how does AI fit in? Acumatica's Ali Jani shares field lessons

The ERP market category is out of step with the times. We need to assess ERP's value in an entirely different way, as in: Extracting value from cloud ERP in a customer-first world - what should we be pushing for?

ERP must either change, or be sidelined. These days, when I am interviewing cloud ERP customers, my go-to question is: how is your ERP system helping you to serve your customers better? It's not a question ERP customers - or vendors - are historically used to answering, but it's the one we must keep front and center now. As I wrote in my Acumatica Summit 2023 show review:

Whether we call it modern ERP, post-modern ERP, or something else entirely, it all comes back to this: "Is ERP helping to serve your customers better?" If not, then why did you bother upgrading to a new system? That may sound harsh, but I believe that's the bar now. ERP must be judged by its external impact. Operational efficiency is, at, best, table stakes.

Helping customers to serve their customers better isn't just about about flexible ERP platforms,  better APIs, or even composable ERP. ERP vendors - yes, even cloud ERP vendors - must change. During a recent virtual catchup, Acumatica CPO Ali Jani shared highlights from Acumatica's own customer-focused shift - and the top lessons learned. I've broken these out into a set of principles for ERP customer success.

The future of cloud ERP is vertical - can we deliver?

For years now, Acumatica's bi-annual releases have emphasized vertical industry functionality ( the virtual launch event for Acumatica R2 2023 is October 5), but I get the sense that even for Jani's team, the implications of a vertical EPP orientation have taken a while to truly sink in. To put it another way, Acumatica has clearly doubled down. As Ali put it:

We spent a lot of time at the Summit, obviously, and I think you got a sense of how much more we're involved with customers... Myself, Mike, and a lot of other folks that built the company had many years in ERP. So we knew what we needed to do, to get to the level playing field with all the other ERP companies. But once we achieved that, a couple of years ago, we started investing heavily in product management. We really felt to take it to the next level, we've got to really engage and understand, at the end of the day, the problems people are trying to solve.

I think it kind of shifted us from engineering-driven company, to a market-driven company. And I think we completed that transition, probably mid last year, where we hired up all our product managers by vertical, and verticalization was a big part of that process. Because if we don't understand the domain knowledge, we can't really understand and the deeper pains that the customers have.

The verticalization of ERP is not just an analyst mantra. This is about customer pain points. As I wrote in a 2021 review of Acumatica customers:

The traditional separation of industry systems and ERP is no longer viable. Yes, companies still toil to integrate the two.

But if you are serious about workflow automation, modern UIs, and embedded AI, you can't have green screens and operational data silos. I don't care if it's MES or restaurant management or construction - proprietary systems running alongside ERP make even the newest ERP systems look like clunky, back office dinosaurs. (The future of cloud ERP is vertical - Acumatica customers share proof points).

Keep your channel partners energized, but keep the end customer momentum also

This one is particularly important for Acumatica. Because Acumatica's partner channel is the sales channel, historically, Acumatica had a different set of tradeoffs. The energy in Acumatica's partner community is always on the high end of the shows I visit, but the risk of losing touch on customer relationships is there also. Jani says Acumatica has come a long way towards finding that balance:

ERP, because it's complex, and is typically installed by channel partners, it disconnects you from customers to a certain degree... From the very beginning, we wanted to co-own the customer - that was one of our big things. As we progressed, we interjected ourselves more and more, and engaged with the customer.

Acumatica's Open University, and transparency - you're not dealing with just a partner; you're also dealing with a software company. So that was kind of in our DNA, which helped us. As we got the product managers in, we really focused on teaching them [the same].

Nothing replaces the customer site visit

I could probably write an entire article on this one - there is a definite methodology to the customer site visit. For a vendor as serious about workflow automation as Acumatica, there is no substitute for observing the steps customers take. The classic example, from my own observation sessions, is the process that still contains a 'yellow sticky note.' The customer might be so used to doing things a certain way, they might miss the chance to automate.

Jani's team has honed a game plan for the site visit. Acumatica showed me the engagement letters they send customers before such visits, e.g. the goals and what to expect. As Jani explains:

I think we kind of morphed a methodology for ourselves, where we can really tune in by vertical on how we write up requirements - what are the root problems people have? And how do we capture those root problems? Do we allow the customer to tell us what they want? Or do we allow them to tell us what their problems are? Or do we observe? Each of them have their pluses and minuses. 

I think we came away with a really good balance of bringing the partner along with us, because a lot of times, the partner understands the pains of the customer more than the customer understands, because they've been with them for so long... When we do customer visits, we set the stage; we send communications upfront.

Customer expectations are important: is this visit a roadmap feedback session? A partner feedback session? Jani shared how they frame the visit:

We're really there to understand how you use the software; we want to observe you using it; we want to make sure we're looking at different personas in the company, and how they use Acumatica. So we really set the stage up.

These visits pay off:

Customer visits have been really helpful for us, because when we go on site, we see what motions they go through, and what they're clicking on... I get everybody to record. Because if they don't record it, I cannot translate the information well to engineering. And then we ask specific questions. It's very industry-specific. We don't go into solutioning on that visit, because we want to just capture pains, right?

Sometimes we do the second visit remotely. We can get on a Zoom call with them and just go through, 'Hey, this is everything that we captured; here's what we learned. These are some ideas we have - what do you all think about XYZ'? We do that individually with each person. And then we bring them all together in a focus group. And then we ask them across groups, because each person has their own agenda. And we try to find commonalities between.

My take - for customer-centric ERP, AI is up next

Obviously, these same principles apply to AI - including generative AI. One thing the generative AI hypemasters are overlooking right now: generative AI needs to go through the same customer adoption model as with any other emerging technology. Jani's team has experience with this, first with workflow automation, and then with AI/ML, which Acumatica has been pursuing for several years now. I strongly agree with Jani's take: when it comes to AI success, internal champions are crucial.

Customers are always giving us feedback: ML/AI is a big area for us right now. We believe in building champions across the different companies and finding which ones are going to really spend the time: do they have the power to bring the company along? Because sometimes you'll work with somebody, but they don't have the authority to try new things, and to get people to do XYZ. So we make sure whoever the champion is has authority as well.

Adoption is everything, and AI is no exception:

That way, we know when we release something, they're going to actually use it; they're going to push for it, and it's probably a real pain that we're solving. Sometimes it doesn't work, especially with AI/ML, because it's so superficial to people... We start with things like, 'Tell us all the different types of decisions you make when you're working with software.' We've learned that decision making is a big part of it.

And then we try to understand all the decisions they make, and then we say, 'Okay, would you have confidence? Is that a critical decision? Or is it a not-so-critical decision' - and then we focus on the not-so-critical decisions that they make, and work our way up.

I was struck by Jani's acknowledgement that "for probably 10 or 15 things we try with ML/AI, if we get one, we're happy." Why do people fantasize that sprinkling AI across the enterprise will work some kind of project magic, not subject to the discipline of maturity models? Others portray enterprise AI like opening a packaged solution out of the box, with wonderfully pre-trained LLMs ready to provide automagical instant value.

Wherease I see AI's potential via the kind of experimentation Jani described - and close co-innovation with customers. When that happens, we should see notable results. I asked Jani: give us an example of an ML/AI use case that stuck to the wall?

Jani explained that Acumatica's big AI breakthrough was realizing customers needed an ML/AI engine that would support multiple use cases. This way, customers could engage with Acumatica in an iterative process. They could try one AI scenario or approach, see if it works well. If not, try another:

So we created this ML/AI engine that's flexible. That can work with generic datasets. And then people can kind of slice and dice it, an d massage it into different types of use case scenarios. That's been a really big success.

We tend to over-mechanize AI thinking via so-called "intelligent" systems. The reality is we need a level of data and organizational readiness to plug these systems in effectively. However, in keeping with our theme of customer-centric ERP, Jani does note one thing: ML/AI requires a different pace of development for Acumatica as well:

If a year has passed, it's already stale in ML/AI. So we had to reinvent how we develop. On the ML AI side, we're very much agile. We had to build a different engine - and segregation on, what areas can we block off.

To be fair, I think all ERP vendors must treat customer-centric ERP as an ambition, not a reality. But when Jani speaks of how Acumatica has changed, you get a sense of how internal engineers can indeed be accountable from the outside:

We no longer want to put up a bunch of features that haven't been proven worthy by customers. When we come up with a new feature, we give it an experimental tag, it's kind of hidden. You go turn it on, and then we gather real feedback in production. Then we fine tune until we feel like 'Okay, it's ready now; we've got at least six, seven customers using it.'

We look at what percentage of the items are really customer driven, versus something that's competitive driven, or stakeholder driven.

We may not be there yet, but this is the right ERP conversation.