AI for Restaurant Operators: What's Real, What's Next, and What to Do Now

Transcript:

Christina (Host: Fourth): Hey everyone, we are live. Thank you so much for joining Fourth’s webinar today. We’re going to go ahead and get this started. I’m Christina from the Fourth team here as your host. So for today’s webinar, the focus will be on how AI is reshaping the manager’s role, improving day-to-day operations, and what operators should be doing now to stay ahead and drive real value.

Now to officially introduce our speakers you’ll be hearing insights and expertise from Christian, Fourth’s CTO, and Jay, Fourth’s COO. I know we’ve got a lot of great discussion to cover today so Jay, do you want to go ahead and kick that off for us?

Jay (Guest: Fourth COO): Thanks Christina. There’s a ton of conversation about AI in the market, which in our experience working with customers, differs from the reality for operators. We’ll talk about what’s real in AI, adoption, actually getting value out of AI enabled tools, and a little bit of what we’re working on and how we’re trying to use AI in our products. Lots to unpack here. Without further ado Christian, thanks for making time to have this conversation today.

Christian (Guest: Fourth CTO): Great. Thank you. Looking forward to it.

Jay: There are some big topics there to cover off. Why don’t we start with what’s hype versus what’s reality?

Christian: There’s been some great innovation happening across the whole space and you’re right there’s also a lot of a lot of hype. That’s across the entire world, in all industries, and it is no different in the restaurant industry. The two areas where we’ve seen real value delivered so far, and one area is close to our heart and that is all about forecasting demand – understanding what you’re up against. Demand Forecasting is a classic use case. It’s one that’s understood. It is one that’s been able to generate value and real impact. And so that has moved well beyond any kind of hype.

The other area where we’ve seen it, is in terms of drive-through. Voice ordering and interpreting what people are wanting to order. We’ve seen some great examples from our customers where they’ve seen huge improvement in their productivity. But that’s sort of only the beginning of it [showing value].

There are other areas where you want it to help you augment the decision making, that are in the early days.

Jay: We’ve seen forecasting be a classic example, as AI is good at pattern recognition, and pattern recognition beyond the scope or beyond the scale that a human can do. AI is helping make humans better in their role.

And I agree with your assessment. What’s tricky is getting AI to work in the realm of decision making. That’s a taller order.

Christian: Absolutely is, but I think it’s where it really will start complementing and supporting managers in delivering the commercial outcomes, and the goals for operators like guest service, and staff happiness. And they deserve the use of technology like this, it shouldn’t be reserved for other industries. Helping operators orchestrate the day, make sure that things run smoothly and having a feeling of satisfaction after delivering a great shift, ensuring the commercial outcomes you want and be able to grow your business – that’s where AI really has to come and help.

Jay: So maybe let’s talk a little bit of where we see AI helping today. An example like forecasting.

The industry is going through a massive flux at the moment, and for some there’s great uncertainty. Everybody’s looking for return on investment. Some businesses are growing, and others are just trying to maintain the profit margins in a very pressured environment. So the ability to know and understand what you’re up against in terms of a demand, and forecast more accurately. Helping even the inexperienced manager to know [how to forecast demand] means that you you’d give them a fighting chance.

Christian: I think the best managers will always do a good job. They are the single biggest influencing factor on the locations success. But that tenure [of a manager] is unfortunately going down. And it’s just the reality. And so this is where technology and forecasting and AI should help out.

For some it’s actually recognising that we can do more. And there is more money to be had, and not leaving anything on the table, because the AI will not have.

Jay: There is a lot of the rhetoric about AI in general – is it’s going to replace jobs but we’ve seen a number of examples where it’s the opposite. A better forecast, a better plan has led to managers actually having more labour, adding more servers, adding more bartenders, to capture incremental sales, so part of the story has been helping with sales lift.

Christian: We’ve got some great examples of that and sometimes it’s quite radical. Like it can really change – one or two more people at the right time to capture that [demand] creates the right experience for the guests and they suddenly come back and then you get into that virtual circle.

The other role it plays is when you’re under pressure. It makes sure that the manager understands where they can not have as many people on, so they’re still managing to deliver the numbers they need to do and that they don’t over forecast and get in trouble. So it plays just as big a role in trying to manage in a very tight labour market and labour costs.

It’s there to give them that much better starting point, and confidence, that they are delivering against a plan, that isn’t just theirs alone but it’s been supported and based on intelligence and help.

Jay: It has some external factors in it like events, and weather. And we’ve seen periods of times of change in demand where the tech removes the emotion and gives the manager a better starting point. So if you have shifts in demand across the day or across the week, rather than just averaging or using your labour plan from last week or the week before, you’re starting to use the AI to anticipate.

Christian: We’ve had some great successes in the UK. Labour costs has gone up dramatically as part of legislation changes and governments deciding to introduce new taxes. We’ve got customers who have been able to offset all of that, and still come through with growth, still hit the numbers and offset that terrible impact on their cost that’s been introduced. I think we naturally want to help grow businesses and do more, but I think its also great to see when it’s being used to effectively manage costs. So it can do both. And then it all depends what you need from it and the help your manager needs.

Jay: Because we’ll probably have a global audience for this what’s the percentage increase in labour?

Christian: It’s really significant at the lower end of the wage scale and we have examples of customers that were able to stay cost neutral. I believe it ranges anything from six to nine percent additional costs just being added onto your average labour from tax structures and legislation, and so for many businesses it’s eating into the very slim margin that’s there already.

The flip side of that is what we saw in the US, where a couple of our customers have bended the trend and really managed to grow business and deploy people at the right time, really maximize the guest experience, and the revenues and repeat custom, guest satisfaction, and it’s just fantastic to see. It’s such a wonderful industry.

Jay: Why don’t we talk about change adoption.

So for businesses on the path to better results with AI, managing change is, in my opinion, important and a challenge, especially in an industry where there’s a lot of focus on service. You’ve got a lot of processes in place to drive that type of thing.

Can you talk about your view of what successful change management looks like?

Christian: I think to compliment what you just said, Jay, is that the added complexity is the distributed nature of all these businesses. It’s hundreds if not thousands of locations dotted all over the country. So it’s in a distributed environment where you have to bring change in.

The key to it from our point of view is that the technology goes hand in hand with the change management within the company. And the belief in what is being done will bring a better outcome. It can’t just be used as this tool [implemented] and [assume] it’ll be better. The tool itself has to help out and show [the manager] that it is better, and be part of  the adoption process, because this is about trust. This is all about can I actually trust AI to help me be better? [Believing] it’s not here to catch me out. It’s not here to make my life more difficult. Actually, it’s here to make sure that I deliver a better result and outcome that is better for me, better for my staff, better for my guests and customers. And so the trust is equal to the adoption. And the best way we’ve seen this is when there is a strong belief from the senior management – operations, finance – everybody’s joined up, and they want to do this. And then finding the champions.

A very classic way of [implementing] it is to go to our best operators, give them another tool and they will ace it. And then you roll it out to the rest of the organization, and leaders will hang on this example, and ask what went wrong when they implement it to the rest of the organization. Rather we’ve seen some fantastic results on both sides if the Atlantic that took a very similar approach – we’re going to test whether it works and using the first as the change agent for the rest of the organization. Typically go for locations that are more under pressure or are not as experienced because you want it to work there. Because if it works there it you know everybody will believe in it. In all honesty, you also see a better return upfront because this is all about where you need to see value. When you bring it into your absolute best operators you may get those mini gains, because you literally got fantastic operators that are top of their game. If you bring it into the locations that are not necessarily as experienced or delivering as good results you will see a much greater improvement. And that means that for the rest of the business everybody can suddenly see this is actually fundamentally going to make us all better.

Another big part of it is when the manager uses the technology itself, and interacts with it, it has to be there to help them and explain what’s going on. And sometimes explaining what’s going on in AI is rather complicated because some of it is really advanced algorithms. How do you simplify that into a way that someone can understand it, interact with it and trust it, whilst they’re doing many other things, and are under pressure to do lots of stuff at the same time. They don’t have time to sit down and study all of this.

So every technology vendor has a real challenge on their hands to be part of the adoption and understand that the adoption and the change management is the key to it. The only way that AI becomes better is when it gets adopted, because managers can sometimes choose to ignore it and then that’s when things start going sideways when managers don’t understand why.

Jay: Maybe to pull out a few themes in what you said. We’ve seen both approaches: customers who really needed the quick return on that investment, focused on underperforming locations, underperforming stores. We’ve also seeing customers whose concepts were performing well and set a higher bar, take higher performing locations work with our most competent managers. One thing you said that I think is really important the best outcomes have come from when the managers are engaging in the forecast and using the tool, but not redoing the work. So they’re learning how the algorithm functions, learning how the tech works, and then gaining confidence in it over time. But there are going to be things that the manager is just going to know, the manager is going to catch that the technology can’t.

Christian: It’s the AI and the managers together, that’s the winning combination.

Jay: Absolutely. I think that’s a very important part of our approach in philosophy is that it’s the human and the… tech is the winning combination.

Christian: We hear too often, can your AI algorithm beat my manager? I’m not interested in beating your manager.

Jay: Let’s talk a little bit about ROI for a minute and then transition to talking about where we see the tech going. As we’re talking to customers, there are some notable exceptions to this, but it’s a difficult time for restaurant operators, there are profit challenges. The U.S. industry in particular is down in the same store sales. My take is there’s going to be a need for fairly rapid payback, and quick realization of ROI for AI technology. What’s your view of how we think about payback and how we think about giving customers assurance that they’re getting value for these products and technologies.

Christian: The key to this is actually tracking it in much more real time. To actually have and much clearer more real-time view of the benefit, the ROI, the what locations are generating what returns.

And that speaks back to your adoption and how quickly can you get ROI? It only comes with adoption. That is historically been not that clear because a lot of the ROI comes from delivering improvements to your balance scorecard. I use balance scorecard as a general term for how you track how a restaurant, how an outlet is doing, how a store is doing.

Jay: Typically guest service, customer satisfaction, and so on

Christian: And there is typically anywhere between 10 to 12. God help some managers, they might even have 20 that they’re trying to manage. E.g. like for like sales guest happiness, cost per labour hour, wastage, all of the metrics, staff turnover numbers etc.

To give confidence to the rest of the business that we’re on the right track is to actually explain against the balanced scorecard and track. Because it can become quite intangible – e.g. we saved the manager an hour, but what was that hour converted into? Rather you can use metrics like guest satisfaction. In that location guest satisfaction has gone up. That’s probably because the managers spend more time on the shop floor. But there are hundreds of actions that managers take, all of them translate into continuous improvements, and improving the commercial outcome. Why would you invest in technology if it doesn’t help deliver better commercial outcomes?

Understanding the KPIs on the balance scorecard and tracking that, and not trying to link it to a singular event is the way forward.

What will happen over time is that AI will start to identify what actions are being taken across your whole estate that seem to have a positive impact on your ROI and your commercial outcomes, which actions are taken that have very little impact on commercial impacts, and which ones potentially even detract.

Jay: Back to that pattern recognition. Start to build correlations between good things happen when the manager is doing X, or good things happen when X conditions are present.

Christian: AI is hopefully helping to enable that, because it can do things that traditional technology can’t. It can help the manager deliver a better service.

Jay: We have a saying, help the manager host the party, and not be focused on administrative tasks.

Christian: The way to get comfortable that these investments are generating ROI, is through tracking that balance score count.

Jay: You need that data. Ultimately you want to be able to recognize the patterns, the events, what are the behaviours that lead to the better outcomes, and give the tech a chance to learn that. So I think of it as almost the instrumentation of ROI is teaching the tech how to learn to better operate a restaurant go together.

Christian: It’s the only way that we can evaluate whether the recommendations that AI is coming up with are valid and good, and that over time they generate better outcomes. It’s the only way that we can run simulations, A-B testing, when we do experimentations and the data scientist looks at the effectiveness of the things we’re doing. It’s a quantitative measurement of what’s going on. It’s essential because it ties everything together. And more so than anything else, proves the value that was delivered. Tying the central business strategy and where you want to drive your company into direct actions in store is the name of the game. And that is where AI is changing this game, a central point of view you have commercial outcomes, and you can tell AI. And it can learn from that and it will then help influence every decision making in these distributed stores in a way that you could previously never really achieve because you can’t reach every operator when they’re in execution mode. We have the ability to do that now, and it’s a central control, central management, and a central strategy that I think is going to be absolutely game changing for the industry.

Jay: This is a natural point maybe to transition and talk about where things are going. So we have talked about what’s real and not, use cases that we’ve seen. Now let’s talk about what next?

Where we are headed is helping managers with decision making, both improving manager decision quality, improving above store manager decision quality and taking some decisions off their plate that can be automated.

Christian: Talking about ROI and the effort and change management, it all ties together. Operators will be thinking do I go through the pain of putting something in? Is AI suddenly going to start running the business for me? And how do you know this is going to work? It’s a massive orchestration of different elements of delivering fantastic service and profitable growth that has a ton of different technologies involved. All the way through from the ordering, the CRM, the guest recognition, the loyalty, all the way through to kitchen management, scheduling, getting paid, etc, etc. All of these technologies are all coming up with ways for AI to be part of helping and delivering. But the challenge with all of this is that you could have 22 different AI agents running around doing things. The technologies and the support to help managers are greatly needed, and it’s not going to come from a single source. How do you create a central control that you’ve given the guardrails and the strategy to, so that when the technology becomes more effective in doing the decision making, working on your behalf, you’re not just letting loose a whole bunch of AI tools that god knows what they’re going to think of doing.

The complexity of how it’s going to work needs radical simplification so that it is shielded from the manager, so they can focus on running the party.

Jay: Pulling data out into Excel in your office, where there is a stack of papers and the fax machine, having managers do a bunch of analysis, seems to be like a bad way to ask managers to operate.

Christian: It relies on them fully knowing and having the experience and understanding of what that next best action is that they need to take.

The other real aspect of this is that every restaurant is unique. We speak to many operators and they say my brand and my food is unique. It absolutely is. We’ll have the ability to treat each one and optimize each one individually. This has not been possible before.

If it was down to humans to do all the data crunching, all the real-time monitoring, all the adjustments. You would need an army of people. What AI can do is truly know and understand what effects that location, how to help that one manager with their level of experience and deliver in their location, with the challenges and the opportunities that that location has. The deal here is that it feels completely configured around them, and their day.

Jay: A motorway or an interstate location is going to be different than a stadium location, which is going to be different than a downtown location, which is going to be different than a suburban location. They are unique businesses.

Christian: It could be a new site that’s just been opened up, and there is a different operating model to the one that’s been there for 20 years. There’s all these nuances, which are tricky to manage.

If you have 20 or 2000 locations, that’s where AI and technology can help. It can give head office the confidence, and the guardrails, the controls, to touch an individual location if they need to. Giving the system the ability to learn and tweak parameters. What are the rules by which I deploy labour because my forecast for demand is one thing, how I turn that into a schedule, and how I think about my shifts is another.

I think there’s a lot of value to be unlocked from having some guardrails and common sense management at a brand level, but then letting the system actually start to adjust, and learn over time, because there’s just not another minute in the day for your manager to have the time to do this. You’re asking the manager to deal with gas in the kitchen, personnel problems – all these things. And then break away for 90 minutes to think about the configuration of my labour software? Just from a human and design perspective, it’s unlikely that that’s going to happen unless you’re dealing with a superhuman. So that’s one thing I’m personally excited about, and how we’re thinking about our approach into the future.

The other aspect to the success of this, is the fact It has to be self-adopting. There cannot be a learning hurdle.  These technologies now have to operate in such a way that you can just give it to the manager. You don’t need a steep learning curve to be able to operate these things.

The other side of the adoption and the change management is the data readiness. The tools and the way that managers interact have to be self-taught. It has to have micro learning. But it also has to help correct the data, because no operations has perfect data. And if you rely on perfect data to implement AI, you’re never going to get there. It has to be self-correcting, self-fixing. It has to figure out to what level is acceptable for us to see and start using that data to do real good at recommendations.

Because relying on a big change management process to go and clean up? Not going to happen. Operators are too busy. So the tools have to do the heavily lifting and figure out how to even do the data readiness and do that as part of the adoption.

Jay: 100%. We’re coming up on 45 minutes, let me play back a few themes.

Future of AI is about enabling decision making for managers. Which is inherently real time and in a way that works for the flow of the manager’s day. So serving up next best actions, serving up decisions in a way where there’s not a lot of noise and tons of places you have to go to look for the answers.

Another theme is this learning and reconfiguration. So having the technology actually work on labour models, ordering patterns. Parameters that historically have been handled at store level or above store, handing some of that over to technology so that it can improve.

The third is this getting past this imperfect data challenge with AI.

The aim of all three is ROI, lowering barriers to adoption, and enabling and helping humans, not replacing humans, do the best part of the job.

Christian: The first day you deploy this stuff is its worst day. Day number two, day number three, you just get better and better and better. All my experience from the last five years of doing this is figure out a way of getting started. However small, because it gives the confidence that some of this will actually work.

And there will be there will be winners and losers in this. And the ones that do figure this out are the ones that are going to be winning. And you might be wondering how they keep growing, and what they do differently, and a lot of the time there will likely be some technology enablement underneath the success stories.

I can understand the nervousness about whether it’s hype or real, but the only way to really figure this out is to try it out and get going. Talk to people who’ve got experience of doing it. Talk to vendors who have been doing it for a while, and who can share with you other customers who’ve managed to do it, so that you’re not alone. You don’t have to figure this out on your own.

As an industry, we need to figure this out, because it can’t just be the competitive advances for the few, because then we’re not going to all succeed. My recommendation here is find a way to get started. And talk to your trusted vendors because they will help you through this.

Jay: Absolutely. There’s a ton more we could talk about, but in the interest of time why don’t we go to questions? Christina, any questions in from the audience?

Christina: So one of the questions is, what do I need to have in place before I roll out AI in my restaurant?

Christian: Quite an open-ended question, but a willingness to try and fail. Irrespective of where you’re going. There needs to be a willingness. And to fail fast. It’s the only way to innovate. It has to be that mindset. As a general approach, it can’t be that it has to be a guaranteed success, otherwise we’re not going to get going. You have to have a willingness to try and fail, and the best do that. They fail fast. And they use it as a learning opportunity and that’s how you get ahead in this.

Christina: Awesome! We hope you enjoyed the conversation and learned something new. Have a good rest of your day, morning, afternoon, depending on when you’re tuning in. And again, thank you so much, Christian and Jay, for the really insightful conversation today. Bye everyone.