I’m so excited to share with you my guest today, Ali Najafian. Ali has had a storied background. He’s worked in web development for ESPN and Clear Channel, in the music industry for Lenny Kravitz and Madonna, in the film industry for the Matrix and Kill Bill, and now as a pioneer in AI and ecommerce.
Ali is the CEO and founder of Alfred Intelligent, and he’s a true innovator and a client. Alfred Intelligent is better known by the brand it owns and operates, Trendy Butler. Trendy Butler is a men’s clothing subscription box service that curates looks and sends them to customers on a monthly basis. The stylist working behind the scenes at Trendy Butler is an AI algorithm. Think of it as a matchmaker, but for your new style, instead of your new spouse. In this episode Ali and I talk about how to use AI and other innovative tech not to curate looks, but to enhance your marketing. Stay tuned for an information-packed and inspiring episode!
In this Episode
- [00:29] – Stephan introduces Ali Najafian, founder of Alfred Intelligent, better known as Trendy Butler.
- [01:39] – What is Trendy Butler and how is it powered by AI?
- [09:58] – The difference between machine learning and AI and why they’re vital in optimizing a modern business.
- [11:37] – Tools you can use to start making AI algorithms.
- [19:07] – Using search results data as an opportunity to recommend and upsell products.
- [28:58] – Ali shares how Trendy Butler uses AI for customer engagement.
- [36:58] – Why content creation is important in building customer relationships.
- [44:31] – Ali shares where Alfred Intelligent’s name came from.
- [45:36] – Robotic Process Automation used in business optimization.
- [54:12] – Visit trendybutler.com and follow Trendy Butler’s social media accounts for updates and news.
Great to have you on the show, Ali.
Thank you, Stephan, for having me here. It’s a real pleasure to have me here. I listen to your podcast. It’s great to be on it.
It’s great to have you on it. I’ve been trying to get you on this show for a while now. I’m really happy to have that finally come to pass. I’d love for our listeners to understand a bit more about the under the hood of Trendy Butler. How is this powered by an AI where all your competitors–the different subscription box companies–they seem to rely on human stylists, and you’ve got an AI?
Most of them use a lot of AI to make their decision making. Except with us, we rely heavily on our AI. Obviously, there is still a stylist because as much as you could throw as many AIs as you want on to this, still there’s an element of creativity that’s needed for a person to be applying to this. You never could control, let’s say the system picking an orange short with a purple shirt, you got to sometimes stop it. Most of it is actually done through our essentially wired AI system that picks these clothes for people now.
The way it works under the hood, it’s a little different from how others do it. I’m sure you heard of things like Stitch Fix, Trunk Club, and all these guys. The little difference is that, instead of having stylists being assisted by this AI to make these decisions, we do it the other way around. We have the AI make the decisions, and the stylists really look and see if they made these maybe improper decision making.
The way it works, usually how people attack the problem or others attack the problem, is that each person they bring them up, and then look at the historical goals that they’ve… Because throughout the process when you signed up, you essentially are asked multiple questions. Those questions essentially help us guide our styling methods and questionnaires.
The way we do it is we turn this process upside down. When we acquire a customer, we ask them a set of questions. Those questions could be from your sizing, your preference in patterns, the type of pants that you like. Then we go a little further, we look at your occupation, we look at your location, the location tells us a lot about you. That tells us the type of clothes that you want. If you send someone in Florida a heavy jacket, I’m sure they’re not going to like it. The location makes a lot of sense. Occupation makes a lot of sense. A lawyer will dress differently from a student, and so on. That gives us an understanding of who you are.
But on the reverse side of it, when we enter products into our system, we describe it. It’s a system that we came up with called synaptic links. Synaptics is essentially a method for you to describe a product. For example, the shirt that you have on has buttons, long sleeves, color, and a certain pattern. Well, each one of those is essentially entered as a synaptic into our system. It goes a little bit further than that. It looks at stretchability. This is something that as a human, you just guess and add a synaptic to it. The material is probably 95% cotton and 5% poly. Some material gives us a good understanding of how much it stretches and the sizing. Then obviously, the brand. Each brand from brand to brand has different sizes that are associated with that mandate. Each one has its own measurement.
We use all that data point. Think about you having a cluster of synaptics. These clusters of synaptics are basically put into a vector. We take a whole group of people that have similar preferences, we call them tribes, and are able to analyze those people. Look and analyze the products and the synapses attached to them to be able to make smart decision making to see which products match to which person.
A lot of AI comes from throughput. In order for the AI to be accurate and pretty powerful, it’s the data that comes in and the data that is put through has to be accurate. Throughout time, we’ve had tons of tons of inputs from our customers. The feedback from returns and exchanges basically makes our process of picking clothes more accurate. Meaning the longer you use the process, the longer you use the system, and the more people use it, the more accurate it gets based on the throughput.For the AI to be accurate and powerful, the data that comes in and the data that is put through has to be accurate. Click To Tweet
For example, when we get a new brand or a new product, we may send out 33% of that product to our member base. Based on the reactions, let’s say some people say, “Oh, this fits a little bigger,” or “Hey, the match…” We took that throughput and that understanding and reflect it on the next set of people that get it. Now we’re a little bit smarter about how we send the item. We use this really religiously. How we come up with these KPIs, it gives us an advantage of seeing what works and what doesn’t. We have certain KPIs. For example, in the clothing industry, the return rate online right now is between 25% and 30%. That’s what they say it is. Initially, when we started the company, because we didn’t have enough data and we weren’t as refined in our methods, we had higher return rates. But now we’re looking at between 3.6% to 6% return rates. That tells us whether the models are working.
Now, every time we build the model though, things change. Basically, the fluctuation and those KPIs will tell us whether we have to rebuild the model and something that went wrong within the basis of how we made our decision making.
The other thing we look at is how we purchase. The same process that we use to pick clothes for our members, we use to purchase products. When a vendor comes to us and has a blue shirt, we’ll enter that, send those synaptics into our system. Our system gives us a confidence score, how many people in our database would want to have that product. Based on that, this is how we make our decision making on our purchasing.
The way we look at if we’re doing good is the industry standard on a product being on the shelf and being profitable in less than 90 days. Anything over 90 days, you’re at a loss. At the onset of the company, we set 45 days. In 45 days, everything on our shelves or in our warehouse that we purchase has to be gone. We look at that rate, how fast inventory is turning, as our KPI for our prediction algorithms and seeing how well we’re assigning products. Right now, medium, large, XL usually turn faster. We’re looking at about a five-day turnaround. On double XL and smalls, we’re looking at 20 days. It means that we’re doing a good job as far as how fast we turn those inventories. It shows us that we’re predicting our purchasing patterns in a better method.
That’s how we use AI to optimize ourselves. We also use it to pick and do all the other things that we do in the background. Well, it goes further than that, there are multiple steps. We even look at it as far as when we should email someone. Let’s say you’ve been a member for three months. You haven’t really heard about us or maybe you haven’t had interactions with us, we try to invoke that interaction because we know that if you don’t have those interactions, people tend to drop off. We increase our LTV by making sure that there’s constant communication.
Based on our past historicals, we know that the people that are on the platform, and entering the system, and changing their style, and constantly updating the information tend to stay longer. We’ve had people from day one that we launched that are still members. We’ve analyzed those guys, we see that they are logging in every month or so, and they’re modifying their style. It’s great to have that, but it’s the data that makes us better at our job.
I was just at Abundance 360 a few weeks ago and heard this quote that I thought was quite profound. It goes something like, “There are two kinds of businesses that will be around at the end of the decade. Those who are using AI in their business, like a lot, and those who went out of business.”
It’s true. It’s a necessity. It’s a tool that you need to use. It gives you that edge that you need over others. We’re right now talking to a lot of traditional big-box retailers. We always ask them, it’s the funniest question for us to ask, what are the tools that you use to predict what to buy? We have this buyer and we look at his rate of return on inventory. That’s how we know. I’m like okay, you could use the data. You have it. You don’t have to look at guesswork. It’s interesting.
You’re using your data and collecting more data so that you can use that data for training an AI. Machine learning algorithms are important. People are just clueless about that. It’s important for the folks who are listening here to understand how vital machine learning and AI as to the longevity of their business.
You could optimize so much about your business if you know better, and you can make fast decisions. Sometimes you have all the data, you could do it with that data, without AI. Sometimes you could pull the data down, you could go through it, shift through it. It’s just a shortcut to sometimes that those decision makings be done by a little bit of help.
Where would you suggest somebody listening start with playing around with AI? TensorFlow? Would you suggest they just use an app that has AI as its basis?
TensorFlow, you’re going to go very deep right off the bat. Sometimes it may not apply. We use a couple of things. The guy that actually wrote the algorithms for our initial one was the guy that wrote the algo for Netflix. He has a system called LightFM that we use. Basically, it’s decision making on multiple levels. We use that. If you’re going to go that deep right off the bat, it may not be too helpful for you. You can start off, for example, on an e-commerce company.
There’s a lot of tools out there that you could use to start your training models. For example, one of the things that we do is looking at what are the products that you should display in front of your site. For example, if you go to a website like Zara, you’re not going to go through every page. What they do is they do a ranking algorithm to basically see what are the products that people are going to be looking for. There are certain key metrics, or certain trigger points, or signals that they will probably have a certain product present you on the first page of your browser.
There’s a company called Sailthru. You could use something like that. You could use multiple tools out there that you might not have to do the initial engineering needed until you get a hypothesis.
The first version of Trendy Butler, when we launched, was really based on logic basis. It was certain logic and if those logics met, then cool. We started getting more and more in-depth. Understanding, what are the data points that are needed for us to actually do this in a more optimized method? It took about eight terabytes of data for us on our customers to be able to train a model that was accurate enough for us to be able to pick. The MVP itself has to be a lot lighter and could be a lot of tools out there that could be used to be able to help you with this.
Just figuring out what to feature on your homepage, especially if you have a lot of traffic, is something that should not be based on a gut feel. It should be based on data and intelligent algorithms to help you with decision making. I know that Zappos, for example, assign a dollar value to every square inch, every pixel on their homepage. Each department, business header, or whatever, it’s holacracy. There’s no hierarchy to the company but you have to make it right in this case that you want a link on the homepage. Even if it’s going to generate a lot of revenue for the company in terms of SEO, you have to not just make a justification, you have to use the internal budget to buy that space on the homepage. It’s crazy. I don’t believe in that. I think that’s silly. It shows how strategic you need to be in thinking about what goes on the homepage and in what way.
Yeah. For example, Zappos is part of Amazon. If you look at some of those AWIs AI tools that they have, you could use a lot of those tools there. One of the things we do, we run these NLP algorithms through our interactions with the customer.
So Natural Language Processing, not Neuro-linguistic Programming.
The other one, the AI side, Natural Language Processing. We have certain open areas where you can enter notes, or we look at interactions. We pull down all the Zendesk interactions through an API, and we’ll run NLP on through it. Sometimes sentiment analysis is a little hard on those things because you sometimes have people that are very sarcastic about certain things. Those guys may not work but we look at keywords and we’ll try to see if something works or trying to find essentially signals of products, signals of people asking us for a certain product type. It helps our product development. We use Amazon Lex for that. It was easier for us just to use Lex to do this. Amazon Lex is the same algorithms that you use for Alexa.
You can actually use Polly from Amazon, AWS. That’s the same thing as what Alexa uses to speak back to you. There’s a lot of tools out there that you can use. You don’t have to over-engineer everything. It’s a different world. If we were talking maybe 10 years ago, I would tell you, “Okay, it’s going to be a little harder.” Now it’s becoming so much easier for you to use some of these tools out there.
It’s not just Amazon Lex. Google Cloud has its Natural Language API that you can sign up for.
Exactly. We use different ones. We use AWS for certain tools. We use Cloud for their vision algorithms and those are great. We do some other type of analysis. This one failed, but we tried to do analysis on the product to see if a certain set of people like a certain set of products. We’ve tried to see if we could predict that based on imagery. It didn’t really work out. Some things have a limitation to it. We try to get too cute with it. We use Google Vision algorithms to do this.
Now on the Google Cloud Natural Language API sales page, it talks about AutoML. For our listeners, who are not familiar with AutoML, do you want to give a quick explanation?
On that one, I haven’t really used the AutoML because we have our own set of things that we’ve been using ourselves, but it’s interesting, something I have to look at. I haven’t really used it. It’s probably something that you can train and you don’t have to worry about under the hood as well as what it is that you have to do in order for you to train an algorithm.Data eliminates guesswork. If you know better you can make fast decisions. Click To Tweet
It’s like Watson. It builds tools to see if you can make it easier for you to just upload a set of data points and see if you could find patterns through it. You have hints of that in your Google Sheets. I don’t know if you ever used it on the right side of it. They try to figure out certain links between data and see if they could find a pattern. It’s a big topic. It democratizes machine learning for everyone. It’s great.
Another thought that you talked about how much data we have. If we’ve got a website, we’ll probably have an internal search engine. Even just an informational site will have an internal search engine, then people are conducting searches that yield zero results. Those are failed searches. Why not go through your top failed searches – the ones that are most popular and keep showing up, create content around those or add those different synonyms and so forth into the pages that are most appropriate? If people are searching for a bespoke suit and you don’t use that term, you use a custom suit, put the word bespoke in there because you’re returning a failed search result for that and enough people are searching for it. Is there any AI to help with that?
We have another site, not in Trendy Butler, but we have another, DesiLux, which is for more luxury goods. For that, we use the failed search result as a method for you to basically suggest new products. Suggest products that are common and close to it. We use something like sales through to sell through to do this. The same algorithm you use to show what’s on the front page, you could use that as an opportunity for you to upsell another product that they were searching close enough for.
You could actually use the same as Elasticsearch or something like that. You could see something that’s close enough and be able to bring back results, but do it in a reverse way where you’re almost there looking for something. Although they didn’t find that specific item, you could suggest something that’s close enough. That’s an opportunity. I think we will use that as an opportunity for those search terms. That’s what we’ve done in the past.
It works. Four pages and not find on searches, those are all great opportunities for you to upsell them on other items. We use those all the time, especially on four pages sometimes. We know what they’re going for. The rule of thumb is never expire pages. You’re the SEO guy so you know this better than everybody else. Even those, you have that sometimes, there are certain pages as it gets in there, you don’t remember that you should have kept this. If that ever happens, let’s just build a tool that we could predict what they were trying to go after and be able to somehow optimize those pages. We’ve done that before. It works well.
That’s cool. I remember this was ages and ages ago. We had a client called Designer Exposure. What the founder would do is she had these relationships with all these different celebrities in Hollywood, Beverly Hills, Bel Air and all that. She would go into their homes with a big wad of cash. She would buy stuff out of their wardrobe that they wore once at the Grammys or something. They couldn’t be caught dead in that again because they’ve worn it once out there in the world. She’d buy these $10,000 or $50,000 dresses and things that would be a great deal for her, but then she could sell it at a nice profit. Still a lot less than what you’d pay for new. They were one-offs.
Once that item is sold on the site, it’s gone. It’s out of stock permanently. We didn’t want to expire those pages because they were bringing in lots of traffic, designer Chanel dresses or whatever. We would use those pages as an entry point that would say, “I’m sorry. This one’s out of stock. It’s sold but we have these other ones that you might want to check out. These are also Chanel. These just came in last week.” That was very successful. This was well over a decade ago.
You were ahead. Look, limited quantity products and things that are exclusive for now, the big hot topic, right? It’s the secondary market. There are more and more companies that are offering limited goods. I think that’s a great method to be able to capture products that are sold out. It makes sense. In those limited goods, we have another product that we’re coming out within another vertical that’s in luxury goods. There’s just a limited supply of those goods. That’s how they keep the prices of those goods up. If you just simply put sold out, you’re going to lose out on a sale. If you lose out, it’s hard to acquire customers. You have to keep them very engaged. That just makes more sense of offering them another product. You do so much work to get people and that’s what keeps them.
Right. They’re a bunch of different companies that specialize in AI for retail. They do different things like deep tagging or computer vision, all sorts of interesting stuff. One of them that I heard about in the Abundance 360 event was called Syte. I know they have competitors like LeapMind and Deep Vision and so forth. What do you see happening in this space? Are you familiar with those guys at Syte?
Syte, I’m not familiar with, but I’ve seen so many kinds of these guys. Think about it. I’ve seen companies that are trying to tag products in the video. For example, on YouTube, there are a lot of products that are being displayed and you want to tag. This deep tagging, understanding what’s in the imagery and trying to figure out those products. It’s a big business, and this happens over and over again. A couple of companies actually approached us. I forgot the name of those guys, but they were looking at videos. They were saying, “We have a group of these bloggers and they’re going to put up these certain videos. Come to us and give us your product. Should something be close enough, we’ll tag you guys.”
In Trendy Butler’s businesses, it doesn’t work because we’re styling service and we’re not selling individual products. If we were selling individual products, it would make sense. There’s a user adaptation period. People are not used to shopping that way yet.
Remember the whole television QVC and all these guys, that’s what it is, but clickability and through content. There’s a couple of companies that are going after that. There’s one, for example, that I just talked to, it’s called Live Rocket. The way they’re doing this is instead of doing the whole boring QVC type of videos, they actually have live performances.
For example, someone like Ariana Grande is performing and she has something on. What she has on, they tag, and you could purchase it. That could change a bit. It’s interesting. Auto tagging with Vision algorithm, it’s a little bit hard. You’re not going to get an exact product, you’re going to get something similar but it’s interesting. I don’t know if people’s habits and shopping behavior are going to go that way yet, these are things that people are trying to experiment with. It’s a problem that people have been trying to go at this for years.
It’s inevitable. Whereas augmented reality becomes just part of our lives, we will experience a broadway show. We’ll wonder, “What’s that main character’s shoes? Where can I get it?”
Somebody has to perfect the medium. Just on your TV watching it, that’s cool. You just have to get used to it, but in augmented reality, amazing. If I see something on someone, I want it and I could click it and see what that product is. That’s amazing. It’s one of those things. Augmented reality, once it’s perfected and the medium for it has been done great. I know Apple and a bunch of people are working on this. I saw the contact lenses for example at CES that they were pushing and all those. Once that becomes something that’s a consumer product that you could use, it’s not the test, I think augmented reality will change everything, especially e-commerce. E-commerce changed by so much.
At CES, one of these guys actually we’re partners with. It’s like these magic mirrors. There’s a company called ULC that we’re partners with. What they try to do is they try to transpose a garment on top of you in a mirror, which is pretty darn cool. What if you could, tomorrow, get up, go for a mirror, and be able to sift through all your products? Those technologies give you a glimpse into the future of how e-commerce is going to be.
It’s not far in the future. We’re talking a few years from now. I got a demo of Mojo, which just came out of stealth a week, two, three weeks ago, something like that. It was just right out of stealth the previous week. They were demoing at Abundance 360. It’s a contact lens that you can get this augmented reality experience. There’s a whole UI. You stare at something, then it can tell that you’re staring at it, then it pulls up different menu options and things. It was wild experiencing different aspects of this thing.
It’ll be out in the next couple of years. It’ll be expensive but it’ll be around for everybody. Affordable probably in the next five years or less. That’s going to transform how people shop online. Everywhere is going to be an opportunity to shop. You’re going to see an outfit that somebody is wearing on the subway and you’re like, “I like that. What is that?”
See, that’s what makes the difference. I think it was Mojo that was out there. That will change the whole experience. You’re no longer stuck in this box of your monitor or your phone. You’re now interacting with the world. Everybody is waiting for hovering cars or skateboards. This is the future. This is exciting because now you get to interact with the world around you through a digital lens. That’s exciting, it changes the world.
How are you changing the way that you market based on AI? You told me a little bit about how you email based on AI to maximize the lifetime value. What else are you doing that is at all machine language or AI based?
Obviously, targeting is the key one. Understand who are the audiences that have the highest CLV. What we did is there’s a great professor out of Wharton, Peter Fader that came up with a concept CLV. It’s the segmentation of your customers and understanding how long they stay and what are those customers that are your highest valued customers. By understanding those customers, you understand that “Hey, we need more of this group of people.” Therefore, using that to essentially enhance your marketing efforts. We have a subset of people, the most obscure things, like regions and all that. Sure, those are given, everybody does that but search really obscure things. People that are into certain music or certain type of events tend to have higher CLV.
Now, why? It’s because maybe they have something in common. These are the data points that we try to retrieve from our past historicals of what’s been successful for us. Understanding the lifetime value of your customers and segmenting them correctly will create a sub list of people, then we’ll try to find more of those people through audiences and other methods of targeting.
That’s through Facebook and Google. It’s really easy to do that. Obviously, you could upload them and create look-alikes. We’ve gone a little further now. We’re looking beyond Facebook and Google. I’m looking at what are experiential things that we could do? What are places that people are going? How can we be able to communicate with them? What type of blogs are we writing about those topic points? How can we acquire articles in those blogs? Using those data points to change our targeting methods.
Usually, if you ask us, who’s your target market? This is our most basic answer, guys between 25 to 34. It’s such a bad answer. The best answer is being so targeted that it’s not that. Maybe it’s that 26-year-old guy that’s walking on Hollywood Boulevard that has a Herschel backpack or something like that. It has to be more targeted and that’s how you know.
With a bulldog.
He had a bulldog. Exactly. Those data points really give us the insights we need to target that right person.
Just starting with uploading your customer list or your email list to Facebook, and generating a look-a-like audience, which is pretty much as easy as a click of a button. That’s a machine learning type of algorithm over at Facebook. It’s a black box. We don’t get to see what is inside it.
That’s the problem.
It’s pretty powerful to see that now you have 10 or 100 times as many people that were on your list that you can now market to through Facebook ads because you’ve created this look-a-like audience. It’s so darn simple, so many people still haven’t done it.
You could do the same thing with Google. That’s the simplest method. Then when you want to get a little bit more granular, you start to go and dissect your data more and understand your data more. We use machine learning to basically retrieve those data points that humans will just look and gloss over. The commonalities between people, as a human, you’re going to look at it in a list somewhere, you’re going to even look at a chart, and you’re going to gloss over it. Sometimes our system will say, “Hey, we see this set of people have the highest LTV with them. Good for you. You should find more of that.” That’s how you should target better.You don't have to over-engineer everything. It's a different world, and AI is eliminating all the excess work by providing smarter processes. Click To Tweet
Facebook and Google are very expensive and are getting more expensive as time goes on. It’s really important to do the right targeting and have the right ad copy and so forth so that it performs. Otherwise, you’re going to lose money or you’re going to have to stop advertising, then your competitors… Have you found that you’ve had to massively increase your ad budgets over time?
Oh, yeah. The difference between 2015 and now, it quadruples the price now to market. As it becomes easier for everybody to start selling… Shopify made it so much easier. There are a million e-commerce sites out there. There are more people selling products. People realizing that, “Hey, I could use Facebook to ramp up real fast or Google real fast.” That is one part of it, then you realize, “I’m competing with like, 1000, 100,000 other people and millions of other people. I have to look at how I market.”
You could throw money, sure, at Facebook and Google. You could keep doing this, but your cost of acquisition and your CPM are gonna rise and it’s not building a brand. You’re going to end up pricing yourself out of the market. We used to do this in 2016. We used to throw tons and tons of money. We still do. I mean, put a lot of money into Facebook and Google.
How much is a lot from your standpoint?
We were doing it, but maybe $600,000 a month minimum? It’s a lot.
$600,000 a month?
Yeah, $600,000 a month.
Wow, that’s a lot.
That’s a lot.
Are you spending six figures a month with Facebook and Google?
Yeah, but not as much. We reduced it by a lot.
Still, it’s six figures a month.
Yeah, we still are.
That’s seven figures a year of ad cost. Holy cow, that’s a lot.
That is a lot but then when you look at it, like, “Oh, there’s a diminishing return here. As ads get more expensive, your payback periods get longer. We look at how much does it acquire a customer and how long does the customer have to stay on for us to get a payback. If your payback period is more than two months, then you definitely are not making money.
What we’ve done is looked at that and said, ” We should reduce the amount of money we spend, and spend the time building value and building brand recognition.” The fewer people know you, the more expensive it is to market to them. What we’ve done in the past year and actually this is our trajectory now is looking at how we market, who are the people that we want to market to, and what is our ethos behind the company? Why do people want to spend their money with us? That’s a big factor.
You could go into this real-world point of spending money on Facebook and Google because it’s so easy for you to acquire customers; but at a certain point, it’s going to get so expensive that you’re going to say, “I wish I did some brand building.” Look at Elon Musk with Tesla. Look at Apple and all these guys. I don’t think they have to spend much on Facebook and Google. They have brands built that you go actively and look at what they’re doing and trying to be a part of. We’re looking at ourselves saying, “Okay, we want to be a brand that people look at and say, ‘I like to be synonymous with them. I like to use their services because I like them. I like their ethos. I like their product. They show me value.'” We look at that a lot. We also look at NPS, Net Promoter Scores.
If we provide a service that you go tell 10 of your friends, then that’s the correct way of doing it. At the end of the day, you could buy 1000 customers, but if they’re most of our detractors, then you’re not really building a brand; nor you’re building a business that’s going to flourish in the future. That’s the key part that we’ve realized, we should spend more time thinking about those things.
What’s one thing that you’re doing that is helping build your brand?
Content creation. We’ve looked at historicals. We’ve noticed that if you’re not resonating with people and you’re not providing value over a product, then you’re not on the top of their mind. We’ve actually looked at this. We’ve done A/B testing. When you’re constantly communicating with your members or customers, you’re actually creating a relationship with them. That relationship helps you establish a longer relationship as you go on. It’s not just your product. You’re not just sending them something every month and saying, “I’ll see you next time I charge you.” You’re also providing value. Content creation is big.
The other thing is ethos. I’m sure, everybody’s talking about how much waste we have in this world and how you are contributing to landfills and all those other things. A lot of research in the last year has been how do we not be a contributor to landfills? About 8% of everything that goes into landfills are fashion. About 60% to 70% are not used. Most of the goods they use, like poly and all this, takes about 300-400 years for it to biodegrade. Be conscientious of those. If you are genuine about your concerns, and you’re actually as a company doing something, create ethos with that and people resonate with that.
I actually read an article recently about how bad it’s gotten with fashion contributing to the landfills. It’s because of fast fashion like H&M and all that, so much is going into the landfills. It’s affecting other industries as well. This entire city in India that survived and subsisted on taking donated clothing and turning it into blankets. They have to find another business to be in. All these people are out of work because the cheap blankets that are brand new out of China are undercutting the recycled blankets that they’re making out of donated clothes.
It’s a crazy world.
We had to sit down and say how are we offering value, what are we doing as a company that is contributing? It’s not all about the P&L and what your end of the year financials are. We have to look at it like that too. If you’re truly offering value and you’re offering something else beyond just financial gain, then it resonates with people. I think you should do that. As a company, this is a step that we’re taking to make sure that we’re conscientious of those things.
Actually, we were looking at breaking down faux leather. Everybody says vegan leather. We looked at the process of how you make faux leather. It’s actually really bad for the environment. Sure, it’s not leather, but you’re killing a lot more other things by making it. It was interesting. It’s been eye-opening for us.
We’ve been doing so much research to understand how we impact around us and our environment. Also, people, right? We were making all the factories that we use for manufacturing our clothes to sign an oath that they’re not using child labor, or sweatshops, or any of those things. Actually, we found through this process of actually visiting them and seeing some of these people, we actually cut out some of our factories because of that. We don’t really talk about it, but we had. It’s better for inner company culture to know that you’re not just a big, bad evil company, and you care.
That’s awesome. You talked about content creation being a differentiator for you guys. What sort of examples of content campaigns or pieces has really made a difference? Just standout examples that you could mention.
Articles are obviously really easy, like how-to’s, but then we’ve gone further. You were a member, you had some of our clothes. How do people know how to wear something? One of the things we’ve started doing is looking at the combination of clothes that we are sending some of our members, then sending them examples of how they should wear it. Super simple but that increased our LTV actually. It’s because we could send a piece of garment, they won’t know how to put it together and see if they think that “Oh, this fashion is not for me.” By doing that, creating instructional videos and creating informative content for them, has actually helped our members. That’s one good key.
Sometimes, really short-form videos are better because we know men. They don’t have the patience for lots of things, so we make them short. Actually, it works better, really fast cuts and instructionals. Instructionals for us have been a really good key part of us being able to provide better service for our customers.Augmented reality is becoming more and more a part of our lives. We, as humans, need to evolve for us to cope. Click To Tweet
I remember bringing in a stylist to do a closet cleanse. The stuff that was remaining after the closet cleanse, she was like, “Alright, well, you can match this up with this. You can actually wear this under a sweater.” I was like, “I had no idea. I’d only worn it one way.” I could match it with all these different colors. I could wear it in different ways. I could unbutton it and wear a T-shirt underneath. I was like, “Wow, there are five totally different looks from that same shirt.” That’s very helpful because I didn’t have that sense of style or creativity. I had to take photographs of all these different options.
Now, you’re stylish. No, but you do good. I always see you very stylish.
Well, it’s my wife. She’s in charge of my wardrobe.
Well, you’re cheating. Okay, cool. Just having that information or method and just giving back, it helps our members. Therefore, it helps us because they use our service longer. I think that’s why they came to us. Just simply putting a bunch of clothes in a box and sending it doesn’t cut it.
Can a member converse with a bot and get styling advice? Is it real humans that they’re conversing with?
Yeah, we did the whole chatbot thing. Chatbot thing is great for very vanilla, simple questions. When you get closer, a little bit of a human touch will help that. Although a machine picks for you, sometimes you want to have a human to interact with. Needless to say, chatbots, filters, a lot of those simple questions. When do I get my box? Here’s the tracking number. Connection or how do I do it? Even creating return labels and stuff. Those are simple automated things that we do. It just doesn’t need humans. Humans in those cases frustrate people. They rather have what they asked right there fast. You don’t want to wait around.
For more detailed things… We do get a lot of emails with people saying or chats about people say, “Well, how do I wear this? I got this and I wear it.” We’ll have people actually talk to them. It works a lot better and increases our customer’s engagement.
I’m curious, why is your AI named Alfred?
The Batman. Alfred always took care of Batman. I always loved Batman as a character. Alfred just made sense for me. Now we do have an investor, keeps on calling him Jeeves. I don’t know why Jeeves because of AskJeeves? I have no idea, but he keeps calling Jeeves. I keep telling him, “No, it’s Alfred.” It’s funny. That’s what we call him. Also, it’s called Alfred Intelligent, AI. We try to play around because sometimes AI is used too much. Everything is not AI all the time. I say “Oh, that one’s not AI. That one’s Alfred. He’s doing it. Some guy we have in the basement doing some things.” I always use that as a joke too.
Yeah, occasionally you slide a pizza underneath the door, then he’ll slide the empty box plus some code underneath the door back at you.
A few more things I wanted to talk about here before we close out the episode. Robotic Process Automation, RPA. What can you tell us about RPA that’s important for our listeners to know?
Sure. We talk about AI a lot, but RPA’s are becoming as important. We’re at the very beginning of it. They’ve been around for a while, but it’s becoming more important to our day to day in our business workings. If you really look at it. I always tell everybody, there are three things that make a business successful if you could take care of two of those. The third thing is easy for you to overcome. One is marketing, the second one is inventory, and the third one is overhead. Your OPEX and CAPEX usually become commerce. As your business grows, you hire more people and you lose track of optimizing. We use robotic process automation to optimize for the repetitive work that people have to do in our office.
One of the things we do is actually creepy, but we do it. We record someone’s screen for a week. We look at what are the repetitive items that they do, and we analyze them. Literally sit there and analyze them. Not the whole week, we won’t sit there for a week; but we will take snippets of things that they do and see how often they do this. When we see that, “Hey, is there a way for us to automate it?”
The reason RPAs are great is because you have two types of RPAs. You have UI RPAs and you have ones that work backend or backend AIPs. The UI does certain things like that don’t have APIs or it’s hard with, we’ll use the UI to do this. It’s mimicking human clicks and mimicking human interaction. It’s doing certain things that are repetitive. The back-end ones are as simple as looking at invoices. For example, we’ll produce a PO, we send it to one of our vendors. The vendor will confirm it and say, “Okay, I’m sending 10 of this shirt,” for example. When the shirts get to our warehouse, they’ll count it in, and they’ll send us a report back how many they counted in. If there is a discrepancy in that process, there is an interaction that has to happen. Now your flow is broken.
What we do is an interception of RPA will intercept that email from the warehouse. It looks at “Hey, there were nine that were counted in this shirt. There were actually 10 in the PO.” There’s a problem. It’ll send an email out to our supplier, saying, “You only sent me nine. This is what we were counted in. Therefore, we’re going to pay a little less or pay one less.” Then email accounting, and essentially give an updated invoice amount. That whole thing is also mimicking our production person’s interaction. It’s coming from our production person’s email. Everybody thinks it’s human, but really it’s a system that’s in between that’s interacting with everybody.
That right there is a simple example of how you could just cut out a minuscule thing that somebody doesn’t have to do. It’s annoying, they don’t really need to do this, they don’t need to do the math. Interaction is there, and it’s great.
There’s an API for this or some other?
Actually, we built the platform ourselves. We built our own framework to do this. The reason we built it is part of Alfred. We’re trying to build a cohesive application that does all these different things.
One of the things we use it, other things we do, which is you could do it easier. There’s a couple of applications, but we do it this way because of our warehouse. Your common denominator in all these businesses as warehouses are always behind, not Amazon, that’s what they do. But usually, when you have a third-party, 3PL, you usually have some of them that are not as advanced. Then, you have to make up for that.
One of the things we have is an IPA that checks every tracking number of every package that was sent. If a package is not moving from a certain time… For example, it says, “Label created,” but the label doesn’t have a movement, we’ll then create a tally and it has trigger points. In 48 hours, if it doesn’t have a movement, email the warehouse manager saying, “Hey, why are these items not being moved?” It’s tedious for a person to do it, but for RPA, it’s easy to do.
We do a lot of those. Within every one of those interactions, we mimic a human. It’s coming from one of our staff members to act like it’s someone really paying attention. It’s actually committing to a task. That’s how we use them. It’s great. The more of these we could use, the less repetitive work our employees have to do, then they could do more creative things.
That’s great. From an SEO perspective, what are you guys up to? We worked together for a while; I’d love to hear what your experience has been for working together?
It’s awesome. You did a whole breaker. I think you do more than SEO. You say you do SEO, but you do more. You had a strategy for us with the type of content we need to make. You looked at what is out there, who are the type of people searching us, what content we need to produce in order to capture them. You do way more than that. I looked at it as a marketing proposal more than anything else when you did it for us. We use that as a basis to even market. You showed me the Old Spice stuff that they were doing, it was great. They created this whole website around this campaign.Offer something else beyond financial gain. This resonates best with people. Click To Tweet
Oh, yeah, the flattering man.
Yeah. It’s two-sided. It’s SEO to create content so you could capture those positioning points, but it’s also just traditional marketing. Creating the type of content that attracts people and keeps them engaged. It’s great. It’s not just putting a bunch of words on a page hoping that it’ll work. You’re the authority on that topic point. It’s great doing it this way. That for us was amazingly valuable. It works out great.
Obviously, the technical stuff, like the structure of the pages and all those other things and using all those syntax and structures that we need to do in order for us to be able to crawl better. That’s perfect. My experience was a marriage made easy.
You don’t do SEO, you do marketing.
I was known for my SEO but thank you. I do see how all the different parts fit together.
Absolutely. I think it took you some effort. That’s what it is.
If you don’t have remarkable content that is worthy of remark, you really don’t have a business.
Absolutely. Yeah, but sometimes it takes help to be creative. All those suggestions were very creative.
Thank you. What did I not ask you that I should have in this interview?
Oh, what did you not ask me? That’s a great question. We went through everything. Honestly, I don’t know. I don’t know if you didn’t ask me something that was… You hit all. Do you know what’s going to happen right after the podcast? I should have said this. It’s always hard to think of something on the spot.
Yeah. If I had to pin you down to share one last pearl of wisdom that you haven’t already shared that’s going to make a real difference for our listeners. What would that be?
Customer relationships are important. Creating those relationships is one key take away from me. It’s more than AI, it’s more than the product. It’s those relationships that give a human touch to it. You use AI to cheat yourself to have a better edge over other people, but at the end of the day, it’s having those interactions and having a good wholesome customer experience. This is why Amazon bought Zappos; probably, because of how they were fanatical about their customer experience. That’s key, that’s what sets you apart. That’s what makes people think, “Ah, this other guy’s charging a little less, but you don’t have a great experience with these other guys.” That’s what for us is the important part, and also the ethos of the company.
About the customer’s brands, one thought that I’ll share just to piggyback on what you were saying is that a website is something that you do, not something that you view. This has been true since the web began. A good website is something you do, not something that you view. If you have a website that’s passive, brochureware, you’re just sitting back and reading page after page of content, that’s not a website, that’s brochureware. You’re going to get eaten alive by your competitors.
All right. Well, thank you so much. They should totally sign up for Trendy Butler. If they wanted to sign up, if they wanted to follow what you’re up to, and some of your musings online, where should we send them to?
First of all, it’s trendybutler.com. The second point is social media. On Instagram, it’s trendy_butler. On Facebook, it’s Trendy Butler. We’re actually relaunching the site next with a better blog area, that’s part of our initiatives. It’s going to go through all the things that we found, like sustainability and all our reports. That’s going to be a great place to see all the content that we’re going to be updating there. We’ve been researching for a year, there are tons of content we’ve been working on. It’s going to be a great place to learn a little bit about what’s going on out there.
Okay, cool. It might be live by the time this episode airs.
It might be. We just actually made one push. I think there’s another push in our sprint that’s coming up on Friday.
I think it’ll be live by the time that our listeners are listening to this episode. Awesome. Thank you, Ali. This was fabulous. Wow, what a cool, innovative…
It was amazing speaking with you. It’s always great. It always lights up the room when I see you. That’s great.
You have tons of tons of information. I’m an active listener to your podcasts too. I learn a lot from here.
Thank you, awesome. You’re just doing such innovative things with Trendy Butler and Alfred Intelligent AI. It’s really cool what you’re up to. You’ve got a really future-focused sort of business. That’s not one that’s relying on old trends, but what is coming. That’s good to see. It’s refreshing.
- Ali Najafian
- Alfred Intelligent
- Trendy Butler
- Facebook – Trendy Butler
- Twitter – Trendy Butler
- Instagram – Trendy Butler
- Youtube – Trendy Butler
- Alanis Morissette
- The Matrix
- Kill Bill
- One Tree Hill
- Maverick Records
- Warner Brothers
- The Red Hot Chili Peppers
- Green Day
- Michael Bublé
- My Chemical Romance
- Stitch Fix
- Trunk Club
- Abundance 360
Your Checklist of Actions to Take
Invest in machine learning and AI if I want to stay relevant in the fast-paced world of tech. Not only will it help my business’ productivity, but it’ll also help lessen overhead costs in the long run.
Organize how I collate my customers’ data to provide better service for them. Map out the best questions to get me the answers I am looking for.
When integrating AI on my website, make sure it has accurate data to work flawlessly.
When selling multiple products, integrate suggested links that will help inform my customers about similar items. This type of marketing strategy helps promote other merch in my shop as well.
Keep communication lines open and accessible to everyone who wants to reach me. My email and social media engagement should be on point, so my prospects and customers stay happy with my service.
Implement conversion strategies on my homepage. It is the frontline of my company. Therefore it needs to make a great impression.
Utilize the API features of online marketing tools to improve convenience for customers.
Be smart about running ads. Make sure what I spend on ads isn’t breaking the bank because paid promotions can get costly.
Focus on outstanding, high-quality content creation. It shouldn’t just be something that fills my website. It should also be engaging and thought-provoking so people will want to share.
Check out Ali Najafian’s brainchild, Alfred Intelligent, and his website, Trendy Butler, for a game-changing way on how I can dress myself.
About Ali Najafian
Ali Najafian is the founder of Alfred Intel. In 1994, Ali created one of the very first proprietary online shopping cart systems. After the company selling for a considerable fee to a reputable firm, Ali then bounced around from ESPN to Clear Channel fulling his dream in web development, design and programming for prominent clients worldwide. He then settled at the New Media Productions division of Maverick Records. There he conceptualized, designed and created websites along with online content for widely known artists/brands such as Madonna, The Deftones, Alanis Morissette, The Prodigy, Michelle Branch, The Matrix, Kill Bill and One Tree Hill just to name a few.
Ali then transitioned from Maverick Records to Warner Brothers Records where he accepted the position at the New Media Department. There he created and implemented new technologies for artists including Red Hot Chili Peppers, GreenDay, Michael Bubble, Talib Kweli, E40, Lil Jon, My Chemical Romance and more. He continued working with Warner Brothers until, he accepted a job at Gold Inc., owned by Guy Oseary. There Ali worked with artists such as Madonna, Chris Rock, Lenny Kravitz, Katherine Mcfee, and others. Most recently, Ali was the founder, head of web development and Engineering for Go Merch and 8bit Inc. There he managed the design, strategy, testing, and development of all technical initiatives for over 350 clients including Tiesto, Pink Floyd, Fall out Boy, Blink 182, Good Charlotte, ABC Lost, Breaking Bad Sony, Fringe TV Fox, Machinama, EMI, Young and Reckless, Pac Sun, Shout Factory, Train and various other artists and brands.