Planful Perform 2023 - CEO Grant Halloran on the real impact of AI for finance - and why generative AI hype doesn't deserve a press release

What would you do if an enterprise CEO, on the verge of their biggest event of the year, basically said to you: "We have no AI announcements to make."

Would you consider them out of touch? Or refreshingly hype-resistant? One day before Planful Perform 2023, Planful CEO Grant Halloran didn't mince words with me:

I don't think finance and accounting people need to worry about their jobs. This is super hyped right now.

AI for finance teams - "It behooves you to start learning quickly"

I'm already wary/weary of each vendors' apparent obligation to release "we're doing AI too!" news, no matter how half-baked it might be. This is starting to feel like a contest to see who can write the spiffiest press release. This flood of AI announcements puts customers - and analysts - in a bind. We must somehow quickly validate:

• Which vendor is truly innovating in AI, and which is better at press releases?
• What's ready for prime time, and what's not? What are the risks?

On the other hand, Halloran does not downplay the impact of AI. He quickly followed his hype warning with an exhortation for finance teams:

I think it behooves you to start learning quickly what this technology can do, being part of the conversation inside your own company. At some point, [your executives are going to say,] 'How can we get more productive?'

I prefer to be the finance person in the room that says, 'Well, hey, boss, I've been studying this for six months, and may I lead the program to see what we can do?' I'd highly encourage people to do that.

We got two thirds through the interview before Halloran finally said: we should probably talk about AI. As I said on Twitter:

The Planful Perform '23 news - not jumping on the generative AI bandwagon

Planful's first batch of conference news makes it clear: Planful wants to speak to a broader set of issues than "we're jumping on the generative AI bandwagon." (Planful Announces Major Product Innovations That Support Customers in Driving Peak Financial Performance). Earlier this year, Planful shared the financial side of their news: Planful Announces Record Profitable Growth with Major 2022 Milestones and Wins.

One thing that frustrates me about enterprise software vendors: for all the platitudes about customer success, the detailed public maturity models are few and far between. Planful has taken this public aspect further than most. Leading up to Planful Perform, Grant Halloran wrote a provocative piece for Planful's diginomica channel: Finance leaders - use this one start-up lesson to navigate uncertain times.

One big takeaway that will surely come up at Planful Perform: yes, finance leaders need to manage risk. Yes, they need to run a tight ship during economic headwinds. But if you lose your capacity for bold innovation because you're hunkering down at every sign of disruption, Halloran doesn't like your chances. But can a more agile approach to planning embolden finance leaders to take bigger chances, knowing they can quickly adjust budgets and scenarios? Perhaps. It certainly changes your ability to collaborate across departments. Halloran writes about a Planful customer that can't take it slow: the Boston Red Sox.

For example, the Boston Red Sox organization adopted an approach that leads to faster decisions and more frequent course corrections; it's a quick-iteration, money-crunching strategy similar to the famous "Moneyball" discipline the organization uses to reduce financial variance and volatility.

I've spoken with the Red Sox CFO, Tim Zue, about how their organization's financial analysts are able to collaborate on budgets and quickly model various outcomes to manage cyclical uncertainty that impacts crucial unknowns like future ticket sales, concessions, merchandise, and even TV revenues. Planning technology helps analysts align with department heads and make spending adjustments quickly where necessary.

In terms of maturity models, Halloran's piece we published during Perform 2022 holds up well: Four steps CFOs can take to build business clarity. Re-reading it now, Halloran's first two lessons really pop for me:

• First, CFOs must embrace cross-departmental collaboration at a personal level.
• Second, they must move beyond financial and commercial data to embrace operational information.

Beyond spreadsheets - collaborative planning expands into workforce planning and marketing

Add those topics to the Planful Perform 2023 talking points list. I'm in London at diginomica's tenth anniversary event; ace contributor Brian Summer is on the ground at Perform. I'm sure Sommer will tell us what resonated, and what didn't. Another change in Perform this year: Planful has ramped up the dialogue across finance, workforce planning and now marketing, bolstered by their September 2022 acquisition of Plannuh (Planful buys Plannuh, connecting finance to marketing performance management).

Timed with Perform 2023, Planful CMO Rowan Tonkin wrote a tough love message to marketing leaders on diginomica: Budget agility is king. Here's how CMOs can achieve it. Tonkin's field lessons bring the piece home:

Budget agility is a superpower right now, and investing in it gives you immediate access to budget and spend performance. The proper technology can go a long way to help your marketing team prove (and even improve) ROI, invest in what is working, and quickly divest the programs that are not, especially when you're under-resourced.

And yet, as of 2020, 95% of companies were still relying on error-prone spreadsheets for some form of budgeting or planning. I'll bet that thorny/persistent topic surfaced this week also. For now, I want to get back to AI, and Planful's AI direction. Halloran is right: customers don't want to be flogged over the head with half-baked generative AI hypotheticals. But they do want to understand how vendors will harness this tech to solve real industry problems.

Planful certainly has plenty to say on AI - Halloran says the adoption of Planful's Predict AI solutions, Predict: Signals and Predict: Projections is "fantastic." I previously updated on Planful's AI strategy in 2022: Achieving finance planning at scale - Planful CTO Sanjay Vyas on how AI can change FPM). As Planful's AI solutions mature, Halloran sees another phase: AI augmentation.

That's the interesting nature of this next wave that we're in - it's that we're now starting to move beyond basic tasks, right? And to free us up to do high value stuff. The AI is going to augment us to do higher value stuff - higher cognition work.

But as for jumping on the generative AI news bandwagon, Halloran had strong words:

One of our competitors, God bless them, came out with a flashy marketing video the other day and said, 'Join the waitlist.' Waitlist - what? C'mon.

I remember a time when SaaS vendors used to differentiate from on-prem by only announcing generally available solutions. Alas, that happily transparent time is long gone. The urge to be perceived as trending properly is too great for any vendor, SaaS or not, to resist. Halloran piled on:

I've seen some really silly videos...  They're just trivial, silly sorts of things that are almost technology for technology's sake. It's like, 'Let's ask the AI, you know, why is this variance number off so much, or whatever?' By the time you type that, and it gives you the next answer, the realization is that there was just literally a cell click through, and you would have had your answer. It's actually faster to do it the traditional way. There's too much gimmicky stuff going on.

But what does Planful think of generative AI's possibilities for finance teams? In the near-term, Halloran says we must nail down risk management, and data compliance:

We just don't understand what the legal landscape is going to look like. And then finally, of course, there's security. Planful operates with sensitive corporate financial data. We cannot ship that out of our data centers - that can't go out to different places. So you've got to figure out, 'Okay, if you're going to be using third party technologies, OpenAI and whatnot, then you've got to figure out how you're going to be able to do that in a secure way.

Those risks require deliberate steps:

There are so many major issues around this topic that are not settled yet that it makes it difficult as an ISV to bring product to market that you can feel very safe about and promote. Now, that said, that doesn't mean we're not doing R&D and working with customers on it, because there are significant opportunities.

Such as? Halloran outlined three areas of near-term finance impact for generative AI:

1. Enhanced task automation.

2. Natural language analysis - one intriguing aspect here? Help finance leaders construct better narratives:

Finance folks do need to turn numbers into stories. Narrative generation, explanation generation - all those sorts of things have some merit.

Just pulling numbers together won't be enough:

It has to be smart. If it's just telling me that year-over-year sales of this product changed by 17%, well, I can already see that in the report. It's actually faster for me to visualize that data in a table and a chart than to read that sentence. So again, you've got to be very careful that you're bringing some kind of utility to the users through this.

3. Natural language insight generation - the conversational aspects of asking questions of the data.

But here's where Halloran puts on the press release brakes:

Now that, all that said, I don't think that I would be saying anything different to what other CEOs of ISVs would already have conjured up.

And that, folks, is why you aren't going to see a generative AI press release coming out of Planful Perform 2023.

My take - a forward look at generative AI for finance

You won't hear me complain about less generative AI press releases! Yes, life hackers on Twitter are doing nifty things today with ChatGPT. But to get beyond the gimmicks Halloran rightly threw under the bus, FPM vendors are going to need time to sort the data risk mitigation aspects, and determine how a company's own data can be fused with the proper external data sets and AI models, without losing the adherence to the regulation (and customer trust) a vendor like Planful must put as job one.

If I were a Planful customer, I wouldn't want to hear about generative AI specifically. I'd want to hear that Planful is aggressively making investments to enable collaborative planning - including AI and workflow automation. An event like Perform gives a chance for customer to get that gut check. When it comes to AI specifically, I do think Planful will have to find the right balance between baking AI into the core of everything, and offering additional AI solutions (e.g. Planful Predict).

Customers will want to know what they get access to and when. Some may need help with the AI business case. Generative AI will doubtless fit in there somewhere, but Halloran is right: enterprise finance leaders need to hear more on generative AI risk mitigation than the typical ChatGPT enthusiast on Twitter.

As always with these generative AI conversations, it's the longer term possibilities that are the most interesting to talk about. When it comes to finance, I'm interested in how these tools could eventually take trusted data and fill in a narrative, such as a cumbersome slide deck, and given a visual overview of how a new division or region has performed.

Or, consider an AI finance assistant, like an alerts infrastructure on steroids, pro-actively letting an executive know about a potential issue in APAC or a suddenly underperforming franchise - without having to crack open a dashboard. For Halloran's part, he sees AI helping to navigate variance issues:

Let's say I'm looking at a variance report. A variance report is often significantly more complicated than people imagine. We're talking about very, very large companies that have  very complex variance reports. I think the future is: just tell me what I need to know about this thing - a seven percent variance on a particular line. You have no idea whether that variance is material or not without the context, right?

 At Walmart, a one percent variance is massive for them, because of the sheer volume of the numbers. But for [another company], a seven percent variance might not matter.

That's where AI can someday excel for finance teams:

The big, big future is finding a way for the AI to truly tell you what you need to know, without you even having to think what the question is.

Yes, generative AI could serve as front-end user interface for querying data. But Halloran wants it to go one step further:

The level I'm talking about is a level above, which is: I don't need to look at this report. It's got 200 lines on it. There's lots of red; there's lots of green, but I don't understand it. It will take me hours to analyze it. 'Tell me what I need to know.' Now you've told me the five things I really need to know. What did we do last time this happened?

How does the data intersect? How does this compare with historical trends? Halloran believes AI could excel at supporting more contextual decisions.

When you start to look at the intersections of the data - a lot of the data that you presented within finance is what we call intersectional data. It's a calculation of a whole bunch of other inputs. Actually, the insight lies deeper down.

Often, that can be given historical context. I think these LLMs give us an opportunity to create a lot more historical context to capture decisions. 'What did we do? We introduced a sales spiff last time, and that worked. Or: what if we introduce a sales spiff? Well, we did that last time, and it didn't work.' So you start to leverage the AI to actually create this transcendent corporate memory and intelligence that you couldn't otherwise.

That's a good conversation to have - and a direction well worth exploring. I'll be interested to hear what Planful Perform attendees had to say about all this.