Perform 2021 - inside the Planful Predict announcements, and how planning must change
- As Planful Perform 2021 kicks off, Planful announces their second predictive application, Predict: Projections. But why do finance teams need AI/ML planning tools - and how should they be designed? Here's my show preview with Planful's CMO and CTO.
I don't care if an event is virtual, on-the-ground, or both. In this case, Planful Perform 2021 is virtual, and you can still check it out. No matter the flavor, this holds true. Events always have:
- A big news story or two (chosen by leadership or marketing).
- A sneaky big news story (typically, this is the backchannel story you might not hear about in the keynote, and it might not be "good" news either).
- An underlying theme (the reason non-news-junkies tune in).
- A burning question - identified during the event, but can only be answered over time.
- Customer lessons to apply (if you can root them out).
Usually, you don't figure these out ahead of time, but for Planful Perform, I got around that - via a sneak preview with Planful CTO Sanjay Vyas, and CMO Rowan Tonkin. The news? At Perform, Planful is announcing the second app in their Predict portfolio, Predict: Projections: an "intelligent forecasting solution."
The underlying theme? As I see it: how finance and workforce teams are responding to the pace of change - and why they need more agile planning to succeed in these market conditions.
The burning question? Can the Predict solutions, including Predict: Signals, help customers to address shifting markets, and flat-out plan better? How can AI/ML tooling make a difference here? The history: in June 2021, Planful formally announced Predict: Signals - the first solution in the Predict portfolio. Now we have Predict: Projections as well.
Finance teams face the transformation pressure
But Tonkin began the conversation elsewhere. You really have to start with this: the pressure on today's finance teams - ratcheted up by the widely-postponed return-to-office, and the challenge of the all-remote financial close. As Tonkin told me:
The big thing we see right now is: the demand of the CFO to be doing a few things. Number one: doing their job more effectively, in a more agile way. And they are under intense pressure to do that, given what's been happening over the last 24 months. As you know, the demand of board ownership and shareholders is: more information faster, faster and faster.
Tonkin cited another problematic twist: many back office transformations were not complete when the pandemic hit. Tonkin:
A lot of organizations were either in the midst of their back office transformation, had already done it, or were lagging. The companies that had already done their back office transformation, or were well on their way to that journey... The increasing intensity that has occurred has certainly impacted them - but nowhere near as significantly as those that were in the midst of their journeys, or hadn't even started their journeys.
If you're behind on your back office transformation, your finance team is back on its heels also:
For those that haven't started, their journeys are playing massive catch up. They're looking to do that acceleration far quicker than they anticipated. They were the victims of their own slow-moving ship. Ultimately, they were effectively caught out.
Add the remote work complications:
Now they're suffering with the great resignation. As they shifted to work from home. CFOs probably never thought you could close the books from home, virtually. Now they're having to do that. All the pressure on finance and accounting has forced them to make changes they otherwise may not have made.
No surprise: Planful hears from these companies - and they're ready to make changes. But they don't just want to play catch up. They want to figure out how to move ahead. Tonkin:
What we're seeing is a strong appetite to start that transformation. Now, do it quickly. But secondly, do it in a way that allows them to gain an edge. Now that they're going to invest, they want to do it really quickly and efficiently.
And so they're asking vendors like us, 'How can you make this happen?' And subsequently, 'How can you do it where I'm hyper-efficient in this in this environment?' And so they're looking to us for automation, AI and machine learning capabilities that allow them to speed up without risking accuracy.
Why do planning and finance teams need AI?
The first step in Planful's response? Predict: Signals, which "Empowers users to take corrective action by surfacing variances and anomalies in data with AI-driven signal detection." Why do customers need this? Tonkin:
Going back to Predict: Signals, we asked customers, 'Okay, we're building this predictive engine. How do you want it to be working within the application?'
And the overwhelming response was. 'As we're harvesting forecasting data from the business and going through that process, we want to speed that process up.' It's fairly common for organizations to be hesitant to move to monthly forecasting, because of how arduous it is. It's a very big ask of the business, to update those forecasts.
If you try to speed up forecasts, manual errors can get you:
The manual errors that can get created are significant. Ultimately, it can end up slowing organizations down. But they all know they want to get there - and they want to get to that monthly forecasting process.
Tonkin says that Signals, now generally available, has been able to ease that forecasting transition. How? By focusing on five design principles:
- AI/ML must be built for purpose with finance teams, and a deep grasp of their functional needs.
- These solutions should not require in-house data science teams, which many companies don't have. Finance teams should be able to consume and act on AI alerts and projections.
- Wherever possible, these capabilities should be embedded into workflows and reports, so you don't have to access them separately - or externally. ("A lot of the AI/ML solutions out there are sidcars," says Tonkin).
- These solutions should augment finance decisions, not replace the human element.
- For finance teams, explainability is a must.
On role-based AI, Tonkin says:
Customers want to [access Predict] in a way that they would normally work, especially as they're going about their standard financial processes, whether that be in planning, budgeting and forecasting, or whether that be looking at variance reports or something like that. It's going to show them exactly what they need for that specific use case.
As for Predict: Projections, Tonkin says this evolved naturally from Signals. Early Signals customers asked for more native Signals functionality in their existing reports - extending alerts into more parts of Planful. "So anytime I'm viewing any report, I could see where there's a potential variance signal," explains Tonkin. And that's how Planful: Projections started.
[Customers] told us, 'Okay. I understand there is a projection there because Signals shows me a variance against an underlying projection. Wouldn't it be great if I could do my scenario analysis, or cast forward various projections to run a what-if analysis based on an upper-bound of the projection, the lower-bound, the midpoint, and then I can quickly modify that? That would make it more efficient for me.'
And so that's the next use case we're unveiling at Perform.
My take - what I'll be looking for at Perform 2021
I don't care whether it's AI or virtual reality goggles. If you can make life easier for your customers, you have my attention. Of course, the proof points are in the customer use cases. So during Planful Perform 2021, the diginomica team will be fanning out (virtually), and documenting those.
Predict: Projects is in early release status, available to pilot customers. But we're far enough along since June to take stock of Planful Predict thus far. Let's see what customers have to say this week. We already know this: if finance teams are under duress, then finance software projects can't be monster distractions. Tonkin shared this:
I saw in Slack yesterday, we got a customer to load three years of trial balance data, and live on reporting, I think it was in under seven days.
Granted, it's not always this fast:
That's an example of a customer that probably has a cleaner data source, for example. Doing that efficiently means rather than going from four or five days to pull the monthly board report or their shareholder report together. Now you're doing that in a few hours. Now they can spend that time focusing on other use cases.
Tonkin says that the go-live to Planful is a matter of weeks, at least for initial use cases. I asked: does that include the aforementioned customer that might be behind on their back office transformation, running on systems that aren't exactly API-friendly? Tonkin says when it comes to pulling data into Planful from legacy on-premise systems, "I haven't heard of customers where it's taking months and months to get data cleansed, and that's delaying the project."
The matter of AI explainability should not be taken lightly - and not just because finance teams need to trust the numbers. You're not going to address the potential bias problems in AI data sets without an explainable solution. (For more on why explainability matters, see Neil Raden's diginomica work, including The problem of AI explainability - can we overcome it?).
I asked Planful CTO Sanjay Vyas about his team's explainability approach. He said the main goal is to make the user experience transparent and intuitive. Give finance users the ability to drill in, and show them how numbers were derived. Example: if you're showing users a forecast, let them go into the account level to check how that data was sourced. However, when it comes to algorithmic explainability, Vyas said:
If we go down to the complexity of explaining the algorithms, it will be a challenge... We have made the user interface so intuitive that most people can say, 'Aha, that's what happened here. We didn't have sufficient data, or we had a missing period in the data.'
So, for Planful, as it is for many vendors, explainability is both a worthy commitment and a work in progress. Does that qualify as the "sneaky big story" of the conference? I'd say no - but explainability remains the sneaky big AI story overall, tied to the ethics issues that get all the sensational headlines. But I think explainability is the tougher hill to climb, and you're not addressing one without the other.
Yes, that means I don't have the sneaky big story of Planful Perform for you yet. With any show, virtual or not, it takes a couple days to draw that out. For now, the Perform 2021 keynote with Planful CEO Grant Halloran is on deck. Watch this space.