Maximizing AI for BFCM

with Jon Tucker, CEO of HelpFlow

In this episode of Ecommerce AI Mastermind by HelpFlow, Jon Tucker explores how brands can use AI to prepare for Black Friday and Cyber Monday (BFCM). He discusses leveraging AI for accurate forecasting, automating customer service, and identifying sales opportunities. Jon provides a detailed walkthrough of AI tools to handle the increased customer service demands during BFCM and even automate ticket resolutions. The episode concludes with tips on using AI to increase sales through insights from past campaigns and upsell strategies, setting the stage for a successful and less stressful BFCM.

Episode
4

Maximizing AI for BFCM

with Jon Tucker, CEO of HelpFlow

As BFCM approaches, Jon Tucker, CEO of HelpFlow, dives deep into how AI can transform operations and drive more sales for brands during the busiest time of the year. Jon breaks down the AI strategies that can help businesses maximize sales while easing the operational burden of BFCM. With decades of customer service experience and several years of AI integration, HelpFlow has mastered the art of using AI to optimize both backend operations and customer-facing services.

Jon starts with the importance of AI-driven forecasting. He explains how brands can use historical and recent data to predict customer service volume more accurately and ensure staffing levels are perfectly balanced. By analyzing traffic-to-ticket and order-to-ticket ratios, AI can help brands prepare their teams, ensuring they neither under nor overestimate the volume.

Next, Jon highlights the role of AI in automating customer service processes. By integrating and training AI to handle customer inquiries, brands can offload up to 50% of their ticket volume. AI isn't a simple plug-and-play solution—proper onboarding, escalation guidelines, and stress testing are essential for maximizing AI's potential. Jon walks listeners through the best practices of training AI to manage real-time customer service volume, particularly during high-pressure times like BFCM.

Finally, Jon turns his attention to the revenue-generating potential of AI. He discusses how AI can help brands identify winning marketing strategies and campaigns based on past data, automate cross-sell and upsell suggestions, and pinpoint the best times to follow up on repeat purchases. Using AI for these advanced insights can significantly increase sales during BFCM and beyond.

By the end of the episode, Jon leaves listeners with a clear understanding of how AI can not only simplify the customer service process but also unlock hidden sales opportunities for a more profitable BFCM season.

SHOW NOTES
  • [00:01] Introduction and welcoming everyone
  • [02:50] How deep are you into AI? A fun audience poll
  • [05:20] Latest AI news and Jon’s story about Tesla’s AI capabilities
  • [09:12] The power of asking, "What might now be possible with AI?"
  • [13:40] Using AI to forecast customer service volume for BFCM
  • [16:02] Leveraging AI to predict ticket volume based on past and recent data
  • [20:14] Real-time tracking of forecasted customer service volume
  • [22:45] AI automation for handling customer service tickets efficiently
  • [24:50] Best practices for integrating and training AI for BFCM
  • [28:35] Escalating sensitive tickets to human agents using AI
  • [32:12] Stress testing AI systems before BFCM
  • [35:10] How AI can help drive more sales during BFCM through campaign analysis
  • [40:03] Identifying cross-sell and upsell opportunities with AI insights
  • [43:50] Timing follow-up campaigns using AI
  • [48:12] Wrapping up and taking audience questions
  • Notable quotes
  • “BFCM is the most important time of the year for brands, but it can also be the easiest to execute if you use AI to level up your operations.”
  • “You can use AI to forecast perfectly. This ensures your team is right-sized for the amount of tickets that are actually coming in.”
  • “AI isn’t just for automating customer service; it can be a tool to help you drive more sales through identifying cross-sell and upsell opportunities.”
  • EPISODE TRANSCRIPT

    Hey, everyone. This is John Tucker, CEO here at Help Flow. And we are going to get started. We've got people joining in from various channels.

    We've got this streaming on social. We've got it on the webinar platform and probably in other places I'm not aware of. But I've got Raecell on Slack over here. So I'm going to be kind of asking her questions to monitor as we go.

    But we are going to go ahead and get started. So you guys can be ready for BFCM. Raecell, let me know that the audio quality is good. Everything's good now that we're live.

    um and we will jump straight into it um but to start with one of the things I always ask uh is on a scale of one to ten how deep in AI would you say that you are? So put that in the comments. You know, depending on where you're at, some people are going to show up on the webinar screen. Some people are going to show up in the other social channels.

    Raysell, if you can let me know in Slack, the comments on social that won't be in this interface. But on a scale of one to ten, how deep into AI are you? Right. So ten being like super deep.

    We're building stuff. We're using AI. We're using, you know, custom solutions that we build. We're super nerds.

    Right. Like, that's level ten, right? And level one is like what I heard about ChatGPT, but I've never used it, right? Like somewhere in the middle, right?

    Somewhere in that spectrum is where we typically see people. So I'm curious on a scale of one to ten, kind of, you know, where do you see yourself? That way it helps me understand like how to frame some of the things that we're going to talk about. With that said, one of the things I always do on these webinars, because these are recurring things, so make sure you guys are able to show up for the future ones as well, I think this is our fifth one at this point, but I've kind of lost count.

    One of the things I like to do is share what's new in AI at the beginning, just so that you can really understand what's happening in the world of AI and new things that are now possible. So I share that each time. I'm going to share two. One is just a funny story.

    The second one is something you can actually apply, but I think both are important. Um, so I was at Starbucks this morning, uh, and I got an update on my Tesla last night for the new version of full self drive. And it has what's called actual smart summon, which will drive the thing across the parking lot and pick you up. Like literally I stood at the front of Starbucks.

    My car was, uh, three rows of parking back with multiple cars in there. It wasn't a very busy parking lot. Cause they didn't want to do it in like a super busy parking lot. Um, And it picked me up in front of Starbucks.

    I tried it. I was super nervous. I was like, dude, if this hits something, I'm going to be late for the webinar. This is going to be a mess.

    But it worked. I pressed the button. You have to hold the button on the app. There's a little target of where you are and then where the car is.

    And it backed out. And it drove all the way down the lane. And then it turned left. And then it drove down.

    And then it stopped for somebody. And then it turned left again. It picked me up right in front of Starbucks. And a lady did look at me weird.

    She's like, what just happened? Right. Like she was pretty confused. But I was amazed. And I share that as an example of learning what's possible with AI.

    That car has been a huge eye opening experience for me because there's something about the physicality of it, the physicality of AI when you're driving in a Tesla and the fact that like, hey, this thing's going like sixty five miles an hour down the freeway, like must trust this thing, right? And I do. I use it all the time, like, ninety percent of my driving. Maybe more, probably.

    And it's amazing. So the physicality of that is an experience that just opens up your mind of, like, AI is happening. It gives you, like, this different level of respect for AI. Second, there was something about that today that just having it, like, do it without me at all, all the way across the parking lot, and it was hyper convenient.

    Like, in that exact case, I can walk to my car from Starbucks, no problem. But What might be possible now? Right. Valet for everyone, essentially.

    Right. I get out of the front of the restaurant and the thing just drives off. Right. So it brings me to this overarching question that I would encourage you as a business owner to just always be thinking.

    Pay attention to what's happening in AI. Understand what's happening in AI and then just ask yourself, like, what might now be possible? Right. What might now be possible?

    Because in this case, the car drives all the way across the parking lot. A more actionable example for you guys might be something related to the GPT-Zero-One model that came out last week or the week before. I think it was two weeks ago. The GPT-Zero-One model.

    It's a new model from OpenAI. that can essentially think is how they explain it and how I would explain it as well. It can essentially think through a problem. And I want to show you a screenshot of an actual chat.

    There you go. So this is an actual chat. I actually don't know what the question is in this particular one. I think it's a math question.

    Let me go ahead. And I think I'm already sharing this. Raysell, if you can show that on the screen. Yeah, I'm already sharing my screen.

    So Raysell, there we go. So Basically, this is a chain of thought that the AI is doing behind the scenes to understand how to solve this math problem. And I probably should have found the actual question that was asked. I just wanted to pull this real quick before the mastermind.

    But notice what it's doing here. It's saying, okay, I'm going to define the variables of the problem, right? Then I'm going to map out the relationship between those variables. Then I'm going to calculate future ages.

    Then I'm going to look at the variables differently. Then I'm going to tweak the timeline a little bit and see if things still hold up. Then I'm going to map the age relationships, determine the future age, formulate the equation, and now I have the equation. Again, I probably should have found the actual question that was asked.

    But the concept I want you to understand here is that it's breaking down complexity into smaller pieces. And then it processes those individually, right? It takes complexity and breaks it down individually. And then it rolls that back up to solve the complex problem.

    And so again, like we could probably go back to the Tesla as an example. It drove all the way across the parking lot to pick me up. Amazing, right? But break that down, right?

    The first thing it needs to do is it needs to have a strong connection from the phone to the car, right? So that's one piece of the problem. And how's the connection signal, right? And there are situations where it doesn't work.

    um second it needs to know the surroundings are the cars behind them are their cars uh or to the side of them or their cars behind them what are the other cars that are around in camera's view etc right then it needs to back out now it's backed out it has new vision of like the the rest of the lane that is driving down and it needs to go slowly and see everything there then when it turns it's going to see more stuff so what I want you to understand here is that gpt zero one is able to break down complexity into single pieces and then think through, it's not actually thinking, but it's reasoning through the individual steps.

    And so this opens up the opportunity to do far more complicated things with an API call by basically sending a request to OpenAI and saying, please figure this out, right? So what might now be possible? That's what I want you to be focused on in this type of news situation is what might now be possible if AIs can think or reason. I would call it more reasoning because it's triggering when people say it's not actually thinking because it's not human, which is true.

    But it is reasoning. It is doing a chain of thought reasoning process or a chain of logic. So what might now be possible? A lot of things.

    So that's one thing I wanted to share with you guys today is some of the things that are happening in AI. And so if you're not experimenting with some of these things in the paid version of ChatGPT, it is available. You can use it there. It's like twenty bucks a month.

    Highly recommend having that because it's like you'll literally be able to see what's happening and give it some complex problems. I coach my son's flag football league. I'll close on this. I coach my son's flag football league.

    I know a little bit about football, but not enough of the strategy of plays and calls and routes and all this stuff. So I've been learning as I'm going. But one of the things that I did is I had it basically map out all our different plays and which players should be in each position based on their skill sets with passing and short, medium, long passing, et cetera. The down, first, second, third, fourth down, short, medium, long, all those things, right?

    And I literally had to grid out an entire call sheet, essentially, for all the scenarios. And it was accurate. And it blew my mind because I knew that there was a structure or a reason-based process to structure the plays based on all these different variables. But I just didn't know how to do it.

    I knew there was a way, but I didn't know how to do it. And it worked. It was wild to see. I'm not saying that's the best way to coach football.

    To be crystal clear, I think it was actually too granular. But the process worked, which was wild. So I would encourage you guys to dig into GPT-O-One, give it some complex problems, watch how it thinks. And I think it will start to unlock things in your minds of like, wow, that might be possible now, right?

    New things might be possible. For now, though, I want to focus on what you're probably thinking of, which is BFCM, right? BFCM is coming up. It's the end of September right now when we're recording this.

    And... you got to be ready for BFCM. And so what we're going to talk about today is how to do that with AI. And so let me click over to the slides here. And Raysell, can you let me know that you're seeing the slides moving around?

    Audio is still good. Everything's ready to go. And we will jump straight into it so black friday right black friday is the most important time of the year for brands but it's also like massively challenging ai or not right it's massively challenging time and the reason why is not only is it a huge opportunity to drive more sales but like for some brands like most of your sales come during this time like we we work with a ton of brands we run cs operations for brands um and some brands of ours drive forty or fifty percent of their revenue in november Right. On that five day weekend or four day weekend, everybody does it slightly different in terms of the schedule.

    But forty, fifty percent of your revenue in five days of the year. Right. So it's an opportunity for massive sales, but it's also a huge missed opportunity if you drop the ball. Which brings me to the second step is like it's a huge strain on resources.

    not only is it tough for your people, right? Like it's a lot of work, both for you as a business leader, but also like for your CS team, your warehouse team, like everybody, it's tough. Your ads team, it's crazy. Especially last year, Facebook decided to change a bunch of things.

    I think that was last year. Yeah, they changed a lot of their stuff right in October, I think. So it's just, it's a crazy time. And at the same time, customers have, customers have, high expectations, right?

    They're shopping from everybody. Amazon set this bar of like, it's going to be there this afternoon, right? So like you're getting, where's my order questions all the time. Um, and it's just stressful.

    Um, and so, uh, sorry, I had a little tech tech issue here. So, um, customers have those high expectations and it can be really, really challenging, but if you know how to leverage AI, BFCM this year could be the most successful, which is great, but it could also be the easiest to execute. So imagine if you could like significantly level up results, but also level up the ease or level down, I guess, the complexity of running things. It would be way easier to run things.

    And I think that's what's possible this year. So I want to kind of plant some ideas in your mind for this. So for us, just so you understand like the context of where this is coming from. So helpflow.com,

    we've been running customer service operations for hundreds of e-commerce brands for almost a decade, so a long time. And so we've been able to go through a lot of BFCMs. We've been fairly technical from the beginning and we've seen a lot of how this plays out with BFCM, but also just in general with customer service for e-comm, right? And over the last three years, we've been very, very deep in AI and integrating AI into all of the operations that we do.

    And so we're using a lot of AICS tools and we're building our own AI tools and we're using AI not just to handle customer service and help customers increase the volume of tickets that are done with AI. but also in like backend operations. So like when you train a new agent or you onboard a new brand, like we use AI for a ton of that stuff, right? So we're using it for a lot of different things.

    And in this webinar, I wanna just give you some ideas of how you can use AI to handle like the unique challenges of BFCM. That's what I want you to take away from this. And I think there's probably gonna be like two levels. I think one level is gonna be using AI in your customer service process just to handle and automate a lot of the customer service workload.

    That's going to be one. And then the other one, I want you to understand like what might be possible. You don't need to do all of the things I'm going to tell you about today, but I think it will give you ideas of additional things that you can do in the future. And so that's, that's kind of the frame I want you to be thinking of.

    So first thing I'm going to talk about is forecasting, forecasting with AI. A big challenge that brands have is they misforecast one way or another on their customer service volume. And the same concept applies to like demand planning. So I'll touch on that for a second in a couple minutes.

    But I'm going to focus on customer service here. Customer service volume grows a lot in VFCM. The challenge is you don't know exactly how much it's going to grow. And you don't know if your forecast of what you think it's going to grow to is accurate, right?

    If you did, you could be perfectly sized in terms of the number of team, not too many agents, but not too few agents, right? And so I want to walk you through how AI can help you with this. To be crystal clear, we do forecasting anyway. We've always done forecasting, but in the last two years, we've gotten extremely nerdy with how to do it with AI.

    And again, I want this to spark ideas for you. So first, you can export your year over year data and your recent data, your customer service data, tickets essentially, ticket volume, and website traffic, and order volume, all of those things. When you export the year over year, for the past few months, you can see the growth and you can see the most recent data for the last three months, not year over year, but last three months, is it growing also, right? What that enables you to do is to give that data to AI to recognize the patterns and better predict how much customer service volume there's gonna be.

    And the way that it can know the customer service volume is by basically calculating traffic to ticket and then order to ticket ratios. And again, this is the process we've always used. But with AI, it's way more robust and easier, frankly, to do. But the methodology here is essentially for every one thousand website visitors.

    How many chats come in for every one thousand website visitors? How many email tickets come in? So you've got your traffic to ticket ratio. You also want to get your order to ticket ratio.

    So for every one thousand orders, how many customer service tickets come in? Usually those aren't chat. We don't break it out for chat for order to ticket ratio. But for every one thousand orders, how many email tickets come in?

    So you calculate the ratios for the time period that you just had, which is way easier with AI. Right. And then you forecast customer service volume based on those ratios. So now what you know is, OK, for this amount of tickets here or this amount of traffic, here's how many tickets we're getting.

    This amount of orders. Here's how many tickets we're getting. Right. Then you can project into the future how many tickets you're going to get.

    And you can do that with recent growth estimates. You can do that with your marketing team to say, okay, like how much additional ad spend are we doing this year compared to last year? You give that to the AI, the AI will understand the already existing growth year over year. It will understand the trajectory of the last few months.

    And then when you give it this new future variable, I guess we'll call it, of ad spend, it will understand how that's going to affect the existing growth that's already seen, right? So again, you're leveraging AI to basically forecast how much volume there's going to be. And then you're able to know how many agents you're going to need, right? So again, you're using AI to forecast.

    And I want to touch on one more thing here. This is all for forecasting. You can technically do this in real time, near real time with the current volume coming in. So one of the things that we do is we break that forecast into daily forecasts, into hourly forecasts, and then real time, we're tracking how much volume's coming in right now compared to the forecast, and what is that gonna mean next hour, right?

    So you're essentially able to track the amount of volume happening based on your forecast, hour by hour, and make adjustments in staffing. So you can do real time stuff as well. That's more granular than most brands need to be, but we have to do that because we have so many tickets, so many agents, so many brands going. But the AI capabilities make that so much more robust to do.

    And so the end result of all of this, to kind of zoom out, the end result of this is you have a perfect forecast of how much tickets are coming in or going to come in, and you have the perfect right size team. uh you can do this with demand forecasting you can do uh for to know that you have enough inventory like the same logic applies there um a similar framework could probably work as well but the key thing I want you to understand is now you know the volume that's going to come in and you know if that volume is coming in as it's happening right so your staffing is perfectly right sized that's the first step of using ai for uh for bfcm Second is what you expect and what you may have seen in other webinars from other people in this space.

    And what you might expect is like, oh, use AI to help with the customer service, right? Which I agree with, right? You should have AI handling tickets automatically and having AI assist your team on the rest, right? So we talked about this on other webinars.

    I'm not going to do like an entire webinar, like you should use AI for customer service and BFCM. I want to provide more value than that, but I will touch on this for a second. You can absolutely have AI handle a lot of tickets if you know how to work with it. And so I want to walk you through like how to actually get results from AI, because it is not a set it and forget it tool.

    Like especially the SaaS vendors in the space, they say, oh, just turn it on and it works. It's super easy, right? Like that's not how it works. You need to know how to manage the AI essentially.

    So I want to walk you through how we do that. Number one, onboarding, integrate and trading the AI. Just like a human customer service agent, you need to properly integrate it into your systems, right? So you need to get it into the help desk.

    You need to connect it to the Shopify or e-commerce system. You need to connect it to your returns portal if it's a totally separate system. You got to give it access to the data, right? So integrate it.

    Then you need to basically onboard and train it for various situations that are going to happen, right? So based on the tickets that you already have, based on the SOPs you may or may not have, you really don't need them. You can create SOPs based on all the tickets because the data's there. And basically you are going to give it detailed training on how to handle the tickets And then you're going to make sure that it's handling those well, right?

    And so you basically need to onboard, integrate, and train the AI so that it knows how to handle those various situations. And part of that training process is running like side-by-sides, just like you would with an agent, a human agent. You would say, you process the ticket and I'll watch you do it and give you some coaching along the way. Or just draft a bunch of tickets and we'll get together for ten minutes every hour and just crank through the reviews, give you coaching, and then we'll send them, right?

    So you got to do the same thing with AI to make sure it's running well. Once you have that in place, you need to have clear guidelines for when the AI should escalate to a human agent. Part of that is from the knowledge base that you give it. Part of that is how you structure the instructions for the AI so that it knows when to recognize sensitive situations.

    If there's a high value order, that's asking certain questions, you might want those handled differently. Or if there's an unhappy customer, you might want those handled differently, right? And so you want to have a context of how it knows which ones are sensitive and make sure they get escalated. And then also, if something gets escalated, if it doesn't get handled by the AI for whatever reason, even if it's not sensitive, You want to have a monitoring process to know why did it choose to escalate so that if you don't want it to escalate in the future, you change it back.

    Right. Like you put you change the knowledge base to keep it with the AI. Right. So you need to have proper escalation and monitoring.

    And then you got to stress test it. Like Black Friday is unique because it all happens at once. And I know it's not completely at once, but like. Thanksgiving night for e-comm brand owners is not relaxing, right?

    Friday morning after Thanksgiving is not super relaxing. That entire weekend is not like, oh, so nice. It's Thanksgiving weekend, right? It's exciting, but it's also a ridiculous amount of work, right?

    So the ticket volume goes and just like gets crazy very fast. So you've got to stress test it. And one of the things you can do with this, it's a little more complex. as a brand owner you should test the ai with questions like test the ai with actual tickets give it the tickets see how it does right um but you can technically get crazy with this and use an ai tool to test an ai tool right so you can like put the ai tool in the persona of a customer with all the questions that might get asked or like give it your tickets and then you can have it generate questions for the ai and see how the ai responds you can do a simulation that's what we do on like big scale brands works very very well you don't need to do that but the bottom line I'm saying is test it before bfcm so that it is you know tuned and working well at that point and and where that gets you is not only are you going to be providing great customer service which is like super important um but also the ai is going to be Running like, you know, thirty for thirty percent of the tickets is a good like, you know, benchmark you should be able to get to pretty quickly.

    Fifty percent of the tickets is really like where you want to see it. And so you get to the point where the AI is doing it, you know, thirty to fifty percent of the time, the team's able to focus on happy customers and, you know, not just, you know, reacting to the ticket volume. Right. So now you've got great customer service happening.

    Partly because you have AI, but partly because you have the forecast we talked about earlier, right? The last thing I'll touch on, and then we'll open it up to questions. So Resell, if you can kind of compile some questions for anybody watching, if you have questions, put it in the comments of where you're at, that'll show up here. It'll show up in other places.

    Resell, if you can kind of centralize the list for me in Slack, that would be awesome. But basically, we'll hit some questions in a minute. The last thing, though, is you can drive more sales with AI. So apart from using AI to forecast the volume and apart from using AI to obviously handle customer service, you can use it to make the entire sales process more effective because BFCM is a time to maximize sales.

    right? So you can use AI to do that as well. A couple of things can work for this. And this is a little bit more advanced.

    I think this will give you guys ideas that you could apply with your marketing team. We don't have anything to sell on this. We can help our clients get data that fuels this, but like we're not a marketing agency. We don't like, change Facebook ads and stuff, right?

    But I think this will help you understand what's possible with AI. So identify winning campaigns with AI insights. One of the things that you can do with AI is use it to analyze past performance data from past campaigns, identify which promotions and marketing strategies drove the most conversions and use those insights to like double down on that during Black Friday, right? So you could take the different email copy that you sent last year, the different Facebook ads.

    If you did some price testing on some of the promos last year, you could take some of that. You basically take all that data, feed it into an AI process, start to query it with questions of like, okay, how did this perform? How did that perform? Here's an offer I'm thinking about doing for this year.

    How might this perform and why might it perform well this year or badly? And again, that GPT-Zero-One model that I mentioned, that's literally like ten days old. would be phenomenal at this. It takes a learning process.

    It's going to take a learning curve to go through to like really understand exactly how to use it. But it is great at complex reasoning stuff. And that opens the door to these types of situations. Take all your marketing data, all your spend data, all the conversion data, structure it into a JSON file.

    You don't even really need to know what JSON is. Like just say, turn this all into JSON and it'll turn into JSON. and then give it to the AI and see how it can help you mine that for insights. I'm oversimplifying, but that's the point.

    I want you guys to understand that you don't need to know deeply how all this tech works. You just need to know what's possible and then start trying it, start attempting things. Identifying winning campaigns, that's one thing. You can also use AI to analyze the past conversations or marketing content you've put out.

    And look at the conversion data and use that to be able to see when there's an upsell or a cross-sell opportunity, right? So you can train the AI to know when there's an upsell opportunity and then configure AI to basically suggest that based on what's happening in the tickets. And again, this is a little more advanced. This is not like the tier one, level one stuff, right?

    You don't start here. But again, I want you to know what's possible because everybody else is talking about the basics in the market. And we focus on being a really strategic resource for our clients to understand what's actually happening in AI. Because next year, BFCM, I think a lot of things I'm talking about right now are going to be table stakes.

    That doesn't mean everybody's going to be good at it. So I want you to at least understand it this year, even if you don't use it all. So for suggesting cross-sells, If you can use your data to know what was effective for cross-sells and upsells, then you can configure a tool that reads the ticket, suggests to the agent, cross-sells and upsells to offer as part of this conversation, right? So that's one of the things that you can do.

    Similarly, you can time your follow-up marketing. So kind of like the cross-selling part, you can use AI to gauge like when is the best time to follow up for repeat orders, right? So you could look at, I'm gonna sort of make this up, but the framework should work Looking at your Klaviyo data to understand, okay, for this skew of product or maybe category, you do it in different ways, but for this product group or something, right? Look at the Klaviyo data or campaign for when we sent out like reorder reminders, right?

    Those types of campaigns. And look at it to say, okay, if we sent it out on day twenty-one, here's what the conversion was. If we sent it out on day forty-five, here's what the conversion was, right? Take all that data, dump it in there.

    And then use that to understand when is the best time to follow up on repeat orders. That way you crush Black Friday, but you know, hey, I wouldn't normally follow up on January second. But based on the data, I think that's actually going to perform really well. So let's do that.

    Right. Or, you know, maybe it's January eighteenth. I don't know. Like in the data, you'll start to be able to see this is probably the best time to do it.

    Let's test that this year rather than waiting till some arbitrary number like February first. let's follow up on January eighteenth, right? So use the AI to understand the timing. And so zooming out like the end result here is not only are you using AI to forecast customer service, but you're using it to drive sales, right?

    You're using it to forecast volume of customer service. You're using it to handle customer service more effectively. And you're using it to drive more sales. And so again, I want to acknowledge, especially for anybody that just joined us, this is like level two, level three stuff.

    Don't start here, but I want you to understand what's possible with AI. And so as a quick recap, and we will be posting the recording of this. So if you haven't uh if you didn't find us from like a email or a social share that we sent out of like here's this webinar happening um contact us through the website from healthflow.com or follow us on social and that way you can see the recordings but we will be publishing this we're changing that process But as a review, use AI throughout Black Friday, Cyber Monday.

    Number one, perfect forecasting. You can use the historical data to know exactly how much volume is going to come in. And then you can use a real time process to see if that volume is accurate, that forecast is accurate and make changes in real time. Right.

    So use it to predict the volume of tickets. so you can prepare your staffing levels. Second, train AI to provide great customer service. Use it to process customer service tickets.

    Automate as many as you can while keeping CSAT really solid, and then use AI to assist on the rest of the tickets. Super, super important. Last is drive more sales with AI. So use the insights that are in your data give that to AI, it'll surface those insights and you can use that to identify additional sales opportunities.

    And again, like number three, drive more sales with AI, that's more complex. But again, I just want you to understand what is possible. If you're interested in pursuing these types of things, we are here to help. We're happy to have a call.

    You can visit us at helpflow.com. Where we fit in the market is we run the CS operation for brands, but we operate as a really deep partner in the business. So you can hire your own customer service agents. If you just think you need customer service agents, we are not the best provider.

    There's other providers for that, but frankly, just get good at hiring agents and hire them directly. Why would you ever pay an increased cost for agents if that's all you need is the agents, right? Or that's all you think you need. What we find is because the customer service part of the business is so integral to driving lifetime value and driving repeat orders, You need somebody that really understands the intricacies of how customer service works and how it relates to marketing and how it relates to retention so that you can leverage the work in customer service to drive sales, but also leverage the insights that are available in customer service to drive the sales of the entire business, right?

    Because there's goldmine insights in the customer service conversations. We help our clients to get those insights and we interface with marketing or interface with retention to help them be able to leverage that. That's one way that we fit in. Another way that we fit in is on the tech side.

    A lot of what we just talked about, I'm CEO and I feel like sometimes I oversimplify. It is very simple to me and I could sit down and do this stuff for any brand. But that's the experience we want our clients to have is like when we say, hey, we're going to do this forecast process. We're using AI.

    We need a couple more data points from your data warehouse or whatever it is. Here's what we need, right? It is easy feeling for our clients when we get to that level with clients because we help them get there. We help them get the right tech pieces in place.

    So a lot of the things I talked about today are very easy to do. when you have the technical foundation in place. And so we are a partner that helps you do that. A great example of that is there are other CS agencies in the space.

    And if you go to them and say, which platform, which help desk is best? they'll default to something, right? And a lot of times that will be because they're a deep partner of that help desk, or they're biased to that help desk for one reason or another. What we focus on is like, which tool is actually best for the client, right?

    Which tool is actually best for your use case? If we're gonna go hard at AI next year, Certain platforms are better for that, right? So that's another way that we fit is not only like the technical guidance and making sure that you know how to do different things, but also making sure that you have the right platforms in place, right? And that it's not a biased recommendation.

    And then the third piece, which I hope you've seen from this webinar, is we're very, very AI minded. We're very savvy when it comes to AI. We've always been tech forward as a company. But really, in the last three years, we've just gone so deep into AI because it's exciting.

    Like from the top, from me in the company, we have two hundred people in the company. I've got this weird relationship with AI where like, I know it's massively changing customer service. And I've told my team that we have hundreds of customer service agents and I tell them like, hey, tier one non-voice customer service is going to be completely automated away in the next eighteen to thirty six months, somewhere within that window. Right.

    So I'm conscious of that, even though that is our core business. Right. It was. And so that's caused us to go very deep into it where it's like, I know it's going to disrupt what we're doing, but I'm also so grateful to be living and leading through like the biggest business and technology change in human history, not just in my lifetime, but in human history.

    Like it is wild what's possible with AI when you really understand it. I understand at a deep level, but every week it's like new layers and layers where I'm like, oh my God, like if it can do this, then we could do all these other things, right? And so we're very, very tech forward as a company, but we're also all in on AI. And I think that we are the best shepherd for this space, right?

    The best shepherd to help brands navigate this space. Because a ton of things are changing and the tech is just moving so quickly. But we've tried to remove the bias from our model. And that's part of the magic, I think, is like, I don't care what help desk you use because we don't get paid by the help desk.

    But I think these are the ones that are doing AI right. And these are the ones that will be out of business in a year and a half or two years. And I know that's a little extreme. You can see my Twitter account of like why I think that.

    But, you know, we're not biased. And so that's my long rant to say we can help you. I love working with brands on AI. So if you're a brand that is interested in doing this, it doesn't have to be for BFCM.

    I know some brands already have their stuff dialed in for that or set. Like you don't want to make a bunch of big changes. Totally understandable. But if you're an e-commerce brand and you're doing like over a million dollars in revenue, that's kind of like the floor of where we can have a big impact.

    Contact us at helpflow.com and we can have a conversation about it. For now, I want to get to some questions. So looking at the questions over here, let me see what would be best to start with.

    I think this one's good. The personal touch one. I think this is what we did last time, Raysell, is I put it into the comments. So that's what I'll do this time.

    So the first question is, how do we make sure we keep that personal touch with customers, right? I think this comes down to... the tone of the AI, and then also the escalation process with the AI. Based on our testing, what we have found is that you should not disclose that it's an AI and say, you know, this is a customer service bot or something. There's regulations around that.

    So I'm going to speak like separate from lawyer land and just say, look, like, you know, figure out your local regulations. We can point you in some resources if you actually care at that level, like as you implement it. I'm happy to have that conversation, but I want to talk at a strategic level. If you don't disclose that it's an AI, the customer will engage in a natural way with AI and they will feel that it's very natural because it is natural.

    AI is very, very good now if you know how to configure it. Then make sure you get the escalation process right so that when the ai starts to not be able to handle something it escalates up to the human agent if you get that part right that personal touch scales you will have a situation where you don't have to have a personal agent human handling all the customer service tickets but it will feel personal and human to every customer right so it's more of a function of how you position the AI and how it relates to your human team, rather than some very specific setting in the AI. That's what I would say for that particular one.

    Next one is how long does it take, basically, is this question. Here we go. So if we're new to using AI, how long does it usually take to get everything up and running? It's like, when should we start preparing?

    How long does it take? If I had to put like a set number on it, I would say like a three to four week period is like the window you would want. That assumes you already have like a solid help desk in place, because if you have a solid help desk in place, we can basically use AI to go through all of those tickets, understand the questions people are asking, understand how your agents are already answering it, and build the knowledge base. So we can use AI to build the knowledge base for AI, right?

    So like a lot of brands get hung up on that where it's like, I don't have our process documented. And my response to that, as you do, it's not documented, but it's there. It's in the business. It's in the data.

    And we use AI to basically build that knowledge base. So you can do it pretty quickly if you have the help desk already set up and you're able to move quickly. Three to four weeks is probably the right window. Because you can get up and running quickly, and then you can use that escalation process to humans to over-escalate to humans as like a guardrail, essentially.

    use that to train the ai and increase the automation rate from there um so you don't have to start with targeting like a you know thirty percent automation rate in week one um you can do that if you take the time to train the ai up front more deeply but you don't have to right so I would say three to four weeks is kind of like the the time window that I would look for um let me see what else there is here I've got a couple more it's personal touch we hit uh This one's helpful, I think. The smaller brands without historical data. I think I probably answered this one, but I do want to touch on it.

    In short, you do not need training data for AI. And like the AI people on the car would be like, you're crazy. Like you do need training data. You don't need a knowledge base to train on.

    You don't need SOPs to train on. You don't need a list of FAQs to train on. You just need tickets. Like if you're already an actual business and you have access to the tickets, like all of it's in there and you can use AI to build that.

    The one situation I do see, we literally have a deal. We're working on this. I hate when deals get structured like this, like for you as a brand, but we have a deal, uh, where the brand is like partnered with the three PL like the, uh, the fulfillment provider. And for some reason, the three PL like owns the customer service stack and the customer service tickets.

    Like it's the, it's the brand's customers, but the three PL does the customer service and they have some ridiculous, you know, uh, policy or, or, or agreement of holding the customer data hostage. So literally the brand doesn't own the customer data. We do have a way we can kind of get at that before they sign up, scrape out all the data, use it to train. And then like, we're not stealing the data, we're training based on the data that's already in there and then they don't need the data.

    That's the only situation where it might get complex. But in that case also, just make sure you don't cancel that contract before you get the actual knowledge base built. That's what I would say in that one. That's the only situation we've seen that's kind of funky.

    Next one is, what's the best way to use AI across a bunch of channels, right? So if you're using, you wanna use AI on email and live chat and social, et cetera, right? You wanna have that consistency of tone, that consistency of information, all those things, right? I think that's more of a function of tech you're using.

    So the help desk that you're using needs to centralize everything in, right? Gorgias is really good at that, of centralizing everything in. Other platforms are good at that too. But if you don't have everything centralized, then AI can't get to all of it, right?

    So really important to centralize everything into one platform. And then if you do that and the AI uses just that core, that single platform, then it will be able to answer questions across the different platforms effectively. You can tune the AI to respond differently depending on channel. So like respond differently on Facebook than on an email ticket, right?

    And it can keep track of, If somebody asks a question on Facebook and follows that via email, technically that can be tracked. Most of the systems are not great at that. The cross-channel sequence is kind of what I would call that. They start on Facebook, they continue on email.

    They're not great at that in terms of automating those tickets. But as long as it's all in there, you can put some guardrails in to... handle that. Um, but bottom line, having it all in a single platform is how you get that consistency. Um, I think those are the main questions.

    Some of these are like variations of those questions. Um, the last thing that somebody asked, which I think, um, addresses like what else can AI do is, is almost how I would word this. Um, so can AI help with like all the returns and exchanges that come in right after BFCM and like, how is that different than the BFCM period? Um, AI can handle anything that it has access to.

    That's the way I word it to clients. So can AI decide if a return should be done? Yes. If it has access to the order, access to the customer service, communication with that customer, access to their past returns to understand if they're a fraudster, right?

    Like if it has access, it can, it can, uh, Consider that. And does it have access to the policies? How do you want to make decisions about returns? And that might mean very granular, here's our rules for returns, or it might mean, hey, if they're not screwing us, give them a return so that they come back and they buy more stuff, right?

    Make the customer happy, drive lifetime value. If AI has access to the data, AI can handle it. That's the same for basically any problem, at least in customer service. Can it execute the return?

    Yes. if it has access to the system to execute the return. What I mean by that is, let's say you're using one of the common return folders. You're using loop returns, right?

    You're using loop. That's where the returns actually get processed. If you can configure the AI to hit loops API for a return that you approve, can it hit loops API and say process this return? If you configure it like that, yes, it can process the return, right?

    AI makes the decision, it hits the API and loop technically processes a return, but AI tells it to, right? That is something that Gorgias is also really good at right now. Gorgias is very integrated into the ecosystem. For Gorgias' AI to decide on a return is trivial, super easy.

    But Gorgias' AI is already connected to Loop and all these other systems, so it's very easy to do. I think that's something that changes next year. I think that brands will be able to build tools that integrate with other systems very easily. It is not hard to tell the loop system, I'm picking on loop, but really any system via an API can do stuff.

    That's exactly how Gorgias does it. But I think what will happen next year is that AIs will start to have access to all these other tools and just do a bunch of API calls to action on things. But that's probably too deep for today. What I want you to get out of this is AI can absolutely handle returns, exchanges, rerouting of orders, changing of addresses.

    It can handle anything if it has access to the information. related to that situation and the system on where the action needs to be taken. So that is what I would say on that piece. In closing, if you need a partner that understands this stuff, we are here to help.

    We love working with e-comm brands. We've been doing it a long time. we are big AI nerds. We're just way into this stuff.

    Again, I think it's the most important change in human history that has ever happened. And I'm super stoked that we're in the position we're in, even though our industry is changing a lot. You know, if you were customers of ours, like we have had conversations, some of you have even spoken directly with me of like, what's going to happen with AI and customer service. And my answer is the same, like a lot of it's going to be automated in the next two years.

    And that is exactly what we're helping brands to do is figure out how to do that. Because I do believe that as a customer service agency, We are best positioned to help you maximize the percentage of tickets and maximize the results of AI handling your customer service. We are best positioned for that for a couple of reasons. Number one, you have to onboard and train the AI and manage the AI effectively.

    That is the same process we do with human agents. We've onboarded, trained, and managed hundreds and hundreds and hundreds of customer service agents, humans, over the last decade. Same process with AI. If you're technical and you know how to do it, it is the same process.

    Second, one hundred percent of tickets will never be handled by AI. It might be ninety, it might be ninety-seven. I don't know what the number is. For non-voice, it's going to be in the nineties, for sure.

    For tier one non-voice, to be clear. But it's never going to be a hundred. And if it's not a hundred, it has to escalate to a human. It has to know when to escalate to a human.

    And then the human really needs to be able to like deescalate it back to the AI to continue it for maximum efficiency, right? Escalation path is also a major part of how outsourcing agencies work, right? We have to know when to go from tier one to tier two to tier three, right? We have to know how to do that.

    We have great processes to do that. And again, we apply that to the AI. Lastly is you have to continuously train and improve the AI, right? You have to review the tickets to understand how it went.

    You have to update the knowledge base to improve how the AI did things. And you have to increase the percentage of tickets that the AI can handle independently while still having a good customer satisfaction score and revenue score, whatever tracking metrics you're using, right? That's exactly how an agency operates. We've been doing that with agents for years.

    We do that with AI now. We've been doing that for at least around eighteen months, I would say, like in production. we are really well positioned to help brands maximize ai because we've been doing it with agents for so many years and we are really the only outsourcing agency that is all in on ai knowing like hey a ton of these uh customer service tickets are going to be automated in the next eighteen months everybody else is just kind of fiddling with ai or they're in some weird partnership with a help desk and kind of slanging their ai products We are practitioners. We build our own stuff.

    We have a lot of weird tools that we built on the back end. Some of you as clients know that and we're deep in it. So if you want to be in it with us and you want somebody to help you shepherd, to shepherd you through like this next phase of AI and what's changing, we are here to help at helpflow.com. We're happy to dig in.

    We will go through your entire system, kind of give you a roadmap and explain, look, here's how we would do it. Here's how we'd approach it. And whether you work with us or not, you'll get a lot of value. So I hope to see you come through.

    My team is ready to chat with you either way. Pay attention to what's happening in AI because not every company is going to make it. So make sure you're paying attention. Make sure you're at least understanding the basics and start applying as much as you can.

    Thank you so much. I appreciate your time and I hope this has been valuable. Thank you.

    EPISODE TRANSCRIPT
    LET’S CONNECT

    Jon Tucker

    Website: https://www.helpflow.com
    LinkedIn: https://www.linkedin.com/in/jontuckerusa/
    Email: jtucker@helpflow.com

    Join the next AI Mastermind session and stay ahead of the curve in ecommerce!

    Past Episodes of

    Ecommerce AI Mastermind

    Details and requirements

    How The AI
    Mastermind Works

    This will be a laid-back, no-pitch session where you can connect, learn, and discover actionable AI tactics. Expect a lively discussion filled with actionable insights that'll give you those "aha moments." You’ll walk away with innovative AI strategies you can directly apply to your business, ready to spark real change.

    Next Ecommerce AI Mastermind Session

    11

    Days

    22

    Hours

    38

    Minutes

    32

    Seconds
    REGISTER TO THE NEXT SESSION

    Meet Your Host

    An image of Jon Tucker, CEO of HelpFlow
    Jon Tucker
    CEO, HelpFlow.com

    Jon Tucker is CEO of HelpFlow - and a huge nerd when it comes using AI effectively. Jon has been leading the charge at HelpFlow in guiding our client brand partners to leverage AI over the past 2- 3 years. In addition to client use cases, Jon has spearheaded numerous ways that HelpFlow applies AI behind the scenes in our operations - so he knows how to apply AI in a wide range of business scenarios.

    When it comes to the AI Mastermind, Jon is the perfect moderator to maximize insights and impact. Not only is he deep in the AI rabbit hole, but he's got years of experience being a member and leader of various business groups. With this nerdy knowledge of everything AI + ability to drive a group discussion, these masterminds are sure to be valuable for everyone that attends.

    Ready to dive in? Register today!

    Providing your phone number is optional and will ONLY be used to send SMS notifications on the event day.
    Your message has been submitted.
    We will get back to you within 24-48 hours.
    Oops! Something went wrong.
    Let's nerd out on AI together

    Be A Guest on our Mastermind or Podcast

    Contact us now if you are:

    • An Ecommerce Brand with an inspiring story and best practices to share.
    • Or a Tech Company who wants to showcase their technology as a useful ecommerce solution.