Insider Secrets to Amazon AI Listing Optimization - Max Sinclair - Ecomtent

And now your hosts , serial entrepreneurs David Shomer and Ken Wilson .

Prior to founding eComtent , max spent six years at Amazon , where he worked . I'm going to start that one over . Prior to founding eComtent , max spent six years at Amazon . While he was there , he worked on launching Amazon business in the UK , the country launch of Amazon in Singapore and led the launch of Amazon Grocery across the EU .

Very excited to talk to Max today . Welcome to the show , max . Great to be here . Good to meet you , david and Ken . Absolutely . To start things off , can you please share with our listeners a little bit about your background and path to founding eComtenta ?

My first few years I spent kind of directly working with the sellers so kind of progressed through the company to managing like the top five or ten sellers in the eu . So these are the anchors and the sun valley techs and kind of really big sellers doing hundreds of you know , millions of uh euros a year .

And then I launched , was part of the team that launched amazon business in the uk , was uh fortunate enough to get out to singapore and launch amazon in singapore , so that was really cool .

Spent months doing that , came back to the EU or to the UK , specifically in the Amazon grocery team and was responsible for scaling up supermarket selling for Amazon , so like Morrison's Monoprix , dia , like big big supermarkets selling 3P via Amazon . So yeah , that's where I left .

You've got a lot of years working at Amazon , so we have lots of Amazon sellers that listen to your podcast , and so my first question to you is going to be with all of your years of experience working for Amazon , what is your advice to Amazon sellers ?

To either learn how to sell better on Amazon or learn the inner workings of Amazon how to just navigate better in there . What's your advice on that ?

Amazon's algorithms and everything else is getting smarter and smarter and smarter , obviously with AI , and therefore , if there was a hack that two years ago you could do this and hack the , you know , the A9 or whatever . You know all this stuff it's getting less and less possible and it's not a long-term solution .

So really incredible products that people really love will be the surest path to success , and I think any . Obviously , the content . We focus on generating great content . So you know that would be part two great content that people can can easily understand and can see themselves using and everything else .

But , apart from having great product listings and great products , yeah , I don't think there's really any hacks to this game .

Maybe you've heard of Rufus , but there's a new LLM-based search which is coming where it works like ChatGPT .

People will put in their desired intent , like I am going skiing , and Amazon , rather than surfacing 50 products , will reply to you like an LLM , like ChatGPT oh , you're going skiing , you're going to need a hat , you're going to need several different items , cross-category and it's going to create that for you .

And it's my really strong belief that come the end of the year , you know the majority of customers will be using this rather than the traditional search . So I think optimizing for , for selling , you know , to an llm is going to be new and very disruptive .

So , my mice , you know , of course , what I said before to that , but I think ultimately there'll be you know , in the long term there'll be no hacking thisM . You need great products that people really love and the systems will work that out and it will surface it to the right customers who are shopping for that particular niche . Very nice .

I I had , as , as I mentioned , a small business was as a school nothing , you know nothing , um , exceptional , but kind of I ran like a , an agency , uh , helping my school friends to get like babysitting and dog walking jobs with local parents , so kind of you know , set up a website and all this kind of stuff .

So it's always really been part of me and I guess I had tried to do various side hustles whilst in Amazon , like two or three different things , and I kind of got to the conclusion that nothing would work unless I put all of my focus and intention into it .

Because you know I'm working a eight to seven job I wouldn't say nine to five job , right , I'm , you're working longer hours than that , maybe on average , I don't know . You know you're working eight till six on average , right , so not terrible , but you work like you're working a hard job , like I also want to be fit . I want to go to the gym .

I need , you know , I need to keep maintain , you know , myself , um , kind of physically , it's important to me and also , you know , I want to see friends and family and the rest of it .

So when , when you kind of put all those things together that I'd be working on these side projects and they would get you know two hours every other evening or something like it , just it just would was not enough time and attention to do anything , and that was kind of my feeling . And then my now co-founder showed me the first release of stable diffusion .

This is in september 2022 , where you could type in a text and generate an image . You know , this is before chat , gpt , and I'd never seen technology like this before and it was . You know , it was completely magical and I think maybe people forget about how insane this was when they first saw it . And I said to him you know he's a phd in ai .

I said to him could you know he's a PhD in AI ? I said to him could you place a product in that scene ? Could you , like you know , take a water bottle , put it in the image , generate that and generate like a lifestyle image .

And he built the first demo of that over a weekend and it had all these issues of distortions and hallucinations and all this kind of stuff that people associated with , like early LLMs and fusion models . But fundamentally I was like was like wow , this is , you know . You know he was confident he could get it right .

I was confident in him and I was like you know , this could be so transformative . And it was literally to me like being at the start of the internet and you know , it kind of all came very quickly in november with chat , gpt , but this is kind of even before that .

So it was , you know , it was like this is going to be just huge and therefore I thought that this , this was , you know , my once in a lifetime charm . So , uh , we went full-time . Basically , since then we've gone through a lot of different interactions and buildings we've now booked , had a patent on our technology . Like you know , we've raised money .

Like a lot of stuff has happened since since . Since that point , lots of happy customers , but , um , but yeah , that was kind of I mean the . The actual reflection point for me was I was considering handing in my notice at amazon . Like you know , this was in the few days after . Like , I've seen this demo and I I'm like very into my cold water swimming .

So I was with my friends and saturday morning we're going into the ponds I kind of jumped into like the lakes and I just had this epiphany that I was just like I just have to to do this , like this is the thing for me . And then that you know , monday I had my notice and then that was that .

So what advice would you give to people that are in a similar situation as you , where they're working long hours , they're trying to maintain some sort of balance in their lives , but they have this thing gnawing at them that they , they want to pursue what ? What advice would you give to them ?

He has this framework called regret minimization framework , which is basically you do the thing that you're going to regret least when you're , when you're when you look back on your life , when you're on your deathbed , and that's going to have to think about decisions and like , from that perspective it was .

It was really clear to me that , like you know , thinking about that would I regret not you , I regret not founding literally one of the first AI businesses . There's hundreds of that , but we were one of the first in this space . It was obviously going to be massive . Would I regret that enormously ?

Would I regret staying at Amazon churning out another year getting slightly higher on the pyramid ? No , I don't miss out for a second . So that framework really helped me . I my grandpa was still around then and I talked to my grandpa and I and I put it to him I said , you know , you know he's literally using this framework .

I was like he he was in his eighties Like what was your perspective ? Right ? You know , my parents were dead against me doing this , as parents often are . But he was like , do it , you know . Like do it for two years , if not you'll find another job . It won't , you know , it won't be a problem .

So that that perspective is like not just a framework , but then actually kind of like talking to um , talking to someone with that perspective really helped me to be like I'm not . You know my grandpa , who I had a lot of , I still have a lot of respect for .

If he believes that was the right that's why I would regret , regret light less in my life then then I'm gonna . You know , I believe him and that's the path I should take nice , very nice .

You know it takes big balls to do a major step like that and a lot of people choose against it , choose not . It sounds like you went out and you sought different advice from different people to collectively to make your choice . I really liked the advice your grandfather gave you . It's like , eh , go try it .

Worst case scenario , you go back to doing what you were doing before . Right , no loss there . So now that you're almost two years out of that , were there any things that you know ? Do you regret the decision ? Do you not like ? Are things better ? Are things worse ?

And I think back to amazon , and you know amazon was supposed to be one of these companies that move fast . And I remember , like trying to get a tech feature prioritized , you know we'd scope it up , we'd go to the tech team , we'd talk to them , then we'd go to end .

You know the us and we debate the us about this and and like the whole thing would take like six months to do , like literally what would honestly be about two hours work . And now it's like , ok , you know someone on my team , we're going to do this , do it , and it takes hours and it's done . No , there's no .

I mean yeah , like there's no part of me that misses .

I guess one thing is that you do miss the , you know , like the kind of communal sense of having kind of peers , I think you know , in a workspace and what I kind of peers , I think you know , in in a , in a workspace , and and what I've got , which I'm very fortunate for , is like a good community of founders that I'm part of , and I think that is super

important , that you know we , literally we just we have a catch-up call every two , every two weeks we just had it earlier and , like you know , you talk to everyone . How you know everyone's in a similar stage to us . How are they doing ? These people , my good friends as well , and and we get a lot of advice from that .

So I think the , the communal aspect is kind of maybe one thing you know which you don't get when you're starting out by yourself , but I mean like if you put that against , like the other , you know being able to stand up for yourself in the world on your own two feet and make something , it's just . It's just like not comparison awesome , yeah .

So I also want to get into e-comptent and kind of what you guys are doing in there , and so maybe we can kind of probably drive into both of those and so can you share with the listeners what Rufus is and for all of the Amazon and e-com sellers , how is this going to impact the next 12 to 24 months in terms of search and product market fit , things like

that ?

But basically , as I mentioned , like Rufus is an LLM that they've trained that you can type in questions and get answers . So I mentioned you can shop by occasional purpose . I gave that example already . You could like find recommendations .

Like you know , give me the best gift for Valentine's Day , give me a good gift for a six-year-old's birthday party , and you can also kind of , when you're on a specific product , ask specific questions about that product detail page . Like you know it was what's the size or whatever this of this product .

So it's basically like your shopping assistant on on amazon and my really strong belief is that this is just such a better , you know customer experience .

Anyone who's used ChatGPT will kind of intuitively understand how to work with this model and I think consumers like all of us who buy stuff from Amazon I guess there might be some categories which might be a bit different , but broadly , like if I'm buying a new kettle . I don't know anything about kettles , I don't really care about the kettles .

I'm ask ruth first , like , hey , what's the best kettle ? Ruth is gonna know like my average spend . It's gonna know . You know it can know the size of my house , like it's gonna know all these details about my purchase history .

So it's gonna say , hey , you know , hey , max , here's five that are good and I'm gonna say , okay , like I want one that like boils in a minute . Which one ? Which one does that ? Bang there , you go fine .

Like I'm not to spend hours and hours going through the different product pages anymore and like , so I think it's going to be , it's going to be completely transformative and addy content . Uh , to kind of in your second question , like we have been thinking about this from , you know , basically ground zero .

Like we're building an ai native so we help to generate the lifestyle images , infographics and keywords . But our real edge is going to come in optimizing for these llms , because llms are a difficult beast . They drift , they change all the all the time we are .

You know we've been doing pilots with slightly smaller live kind of llm based shopping assistants that are out . Now there's , you know , if you're a shopify seller , you can find you know these kind of plugins already for your store .

Like we've been working with some of these folks and we'll be monitoring how Rufus is evolving and what's needed to service and we'll be helping sellers to rank there .

So the whole framing that we have now of I'm going to go to Helium 10 or Jungle Scout or whoever and I'm going to put the keywords in and like that's gone right .

The LLM is going to understand the intent the customer has when saying hey , I want a six-year-old gift for my six-year-old niece , and when saying , hey , I want a six-year-old gift for my six-year-old niece . And it's going to match that intent with what it knows about the customer already .

It's going to work out what additional stuff to put in there that they're not even putting in there . Like you know , they might not put in she likes pink , but they might . Amazon might know that they bought , you know , tons of pink stuff , right ? So like there's going to be all sorts of kind of things going into this model model .

So number one quality content that explains what your product does , how people can use it . Number two visual content . This is more bigger at the content . But these ai models now can understand images . They don't just understand text , right ?

So if you have a like a gray shirt , like I'm wearing , and in the title it says blue shirt , but the photo is gray shirt , rufus is going to understand that that's a gray shirt and they're going to put more weighting towards the images in terms of color and other and other aspects , right ?

Whereas with the , you know , keyword based world we live in today , it's not the case . You put in a color , you're going to get all sorts of random colors come up on amazon . So , like images which kind of have lifestyle images that show how your product is used .

Again , in this example , the six-year-old girl having a , an image of like a girl who looks about six years old , with a product , is going to help with us to understand the context of that product , right ? So , using all of your images basically to sell to the AI , not just to the humans . Content Q&As are going to be super important .

The customer reviews and Q&As , like Amazon have openly said this is what they're training it on . So you want to have really in-depth answers and everything else . You're not just replying to that customer anymore or that particular customer you're . You're telling amazon , like how it should respond to customers when searching for your kind of product .

So the q a's would be another one and I think the last one , and I could go on , but I'll stop . The last one would be regular optimization .

I think people gonna either through software or manually , they're gonna need to track this , they're gonna need to work out what's what's working , what isn't , and they should be refining the listings because these , these , these models are going to shift right .

So what was working for you on in january and what works for you in in march is not going to do the same thing , and I think anyone who's used chat , gpt will have felt this that you can .

There were periods where they change a model or whatever and suddenly the problem that worked really well is now giving you really weird and bad results and their answer is like they've introduced new data to the model or like they've done some change and like it's gone over here , but somewhere way down somewhere else it's made a change and nobody really knows how

these ai models work like . In essence , like how they compress the information is still a black box to the scientists . So changes will have unseen consequences , which means models are going to drift . People are going to need to stay up to date .

I don't know how many sellers focus on this or not , but if you look in the back end of the category listing report there are hundreds of columns of descriptions or fields for you know age range , color , what is this product made ?

So do you think filling all of that information out is going to be able to drain that model more for your specific product , I mean ?

Or , if they don't exist in the Q&A , go and create as a customer saying , hey , what's your best water bottle usage , you know , or whatever , and then you can kind of go in with your phrase of , hey , this is boom , boom , boom , boom , things like that .

And so therefore , like , yeah , as you say in the q and a's , even if you're repeating what's literally like up the listing , they're literally training the model off your Q&A . So I would 100% do like , as you say , I think it's a very good idea .

You know I , I was responsible setting up the cataloging in singapore , right , it's my , my job to kind of do the browse notes and everything like this , and the reason that you do this locally is because there's different laws , there's different sizes , there's different all sorts like .

There's a hundred reasons why you know the Singapore layout of how people should browse and search needs to be different from you know the main one . But it is just a nightmare to talk even internally to the same teams . There's just so like Amazon is so big and there's so much tech that's been built , there's like so much change that needs to happen .

So I think that the people who are like I hear a lot of people talk about this in the community and say , oh , like they've done this change on Seller Central , the Seller Central team would like optimizing titles . Therefore the Rufus team are using , like these teams literally probably don't even know each other exists , right , like that's honestly .

Like people in this right . Right , like that's honestly like keep on this right . Like there's no like central meeting where , like where all these people come in . You know there's just like so many different teams and I think like , yeah , maybe a day they might report into the same manager , like a way up high in the vps and whatever .

But like you know the the tension , like I can tell you the tensions between different teams , and like we're both reporting in same same vp and one team saying , oh , like , we care about this thing , we care about , like I'm in charge of titles and I want all the titles to look uniform and that's my team's thing is like you know , we want to get rid of

characters and we would like , however , they're gold . And then you have the other team which is refus , like we want to improve search . You know they're scientists and researchers . Like I just don't think that these things are related and I will . Um , I mean , there was .

I'm not gonna give this example , but like I was talking to one person who's very one in the Amazon community you know very good speakers he was talking about this thing . That happened and I went and I talked to one of my friends who was an AI engineer in AWS , who worked on this , and they're like , oh no , we're not . We're not .

Like we've literally built our own model . It's completely separate from that example .

So I think that , yeah , that these teams are very disparate and I like , from being there , there's there's really not that much kind of common strategy , putting like there's obviously common strategy but there's not , like they're not talking to each other like nearly as frequently , as they should do .

I've got it now in the size of in the side of your browser so you can kind of always use bing .

So I imagine you will have the search and then next to it you'll have like a little Rufus , or maybe they'll rename it to something which is a bit easier for people to get on with and they'll just click that and then they'll start typing in and it will go on .

I think it will be integrated soon like into the heart of Amazon search and people will either be doing their old search or they'll be like clicking like you know two centimeters away to , to , to , to interact with rufus and those , and it'll be very seamless , but between them .

But we've we've kind of evolved to basic like e-content the name comes from e-commerce content and the vision is to basically help sellers optimize their content with ai across no-transcript , no-transcript . You know you'd have the 250 characters for your amazon listing , you have the 80 characters for ebay and you'd have the right images for each one .

So that's kind of where we're going . Where we are now is we're very much focused on Amazon . We're just focused on we've got optimized keywords , we've got lifestyle images , we've got infographics in terms of generating benefits and generating dimension infographics , so you can basically generate a very high quality listing in one click on Amazon .

I don't know when this is launched , but we're working on A-plus content generation now in one click and that will be a month or two from today .

They are now on page one when it comes to titanium kitchen knife we've helped them with . We've done some split testing on product pinion . If you know , product pinion it's like a , like an ab testing site and 88 percent of customers prefer the ai generated listings we've helped them with versus others . So that's , that's a case study that we've got live .

We've got a couple of others on on the website . You know we've worked with some of the aggregators . We've worked with some other . You know work with some other sellers . So , yeah , we've got a few , a lot of those case studies .

And this is something let me ask you this for somebody maybe a little bit newer to the Amazon game sometimes it's challenging to come up with brand new infographics and lifestyle images . However , if you see a brand that's doing a great job , oftentimes it's easier to observe somebody doing a great job , and so , yeah , silly Slick .

I guess I answered my own question . Silly Slick is a great example . Any other , like gold standard brands that you think do a really good job here , whether they're customers or not ?

account manager and I remember kind of meeting these guys and they just were so much more advanced than , like what I knew about .

You know , I was learning from them about selling on amazon rather than it , and they were so brilliant in the product development like I remember they showed me like their roadmap and they had like those projectors , which is like a mini projector we can project onto a white wall and it would like you know , so these make quite common now , but like back back

when , back then , you know , like six years ago , this was , this was the new thing , right , and they had those and they had like the Roombas that are quite common again now , but , like you know , they had all this like new anti-tech I mean , they're obviously famous for the power banks and this kind of stuff they were like just so focused on great product

development and , yeah , also , they look beautiful in terms of the infographics and the lifestyle and everything . So I think they are now they've IPO'd right Now they're an incredibly you know , they're an incredible business . So I think those are the that is a gold standard on Amazon that I've seen .

Those , the knives , the images are amazing , and what I think is pretty sweet is that you guys offer A-plus content infographics , product images , listing copy , lots of different services all for Amazon , which is super cool , and so I guess a follow-up question would be now the Rufus , not to get into that , but like you guys are obviously very well in tune with

what's happening and what's going to happen on Amazon . Now you're crafting this tool that you have to kind of go along with that , or yeah ?

We've been building , like everything we've built has been in the context of ai and of of these models , and we have been fine-tuning them since the week after it was possible to do it and we've done .

You know , I'm happy to go into this like tons of different experiments on what is possible with ai and what isn't , because , as we learned early on , like a lot of stuff , it's not ready for a lot of things , but we've worked .

We've kind of worked out over the the two years like what does work and you know what , what you can use these models today to produce great results . And yeah , that's where we're focused on , on on building .

I it sounds like a horrible title for a book now , but I remember reading it a year , like I was like a year into amazon and I always felt that I was , you know , hard working and smart and I just couldn't understand why , like other my , my peers were like getting better projects than me .

And then I realized I read the book and I realized like I was , you know , it was my fault and how I was interacting people was completely wrong . Yeah , it really transformed my , my and I would recommend everyone reading it .

But I mean the amount of just stupid trainings . Trainings like every week you're sitting through like four , five , six hours . These like ridiculous , like trainings . I've talked to my friends now . They're like lawyers and accountants and there's like , oh , there's sensibility training , that and it's like literally like you can't , you can't .

No one's doing any work , no one's focused on the customers . People are literally like sitting there . Like you know , yeah , I don't miss out at all okay , yeah , it's not very efficient at that side .

There's always going to be tough times , like there's .

There's times when , um , you know , me and my found , my co-founder , like what you know , weren't taking salaries like nine months , like , well , you know , when you know , me and my co-founder weren't taking salaries for like nine months , like you know , we raised , you know like 30k USD for the first , like you know , nine months .

So we had no money to do anything . So it's really tough and I think that you know you just can't .

You just have to believe in yourself and push through and you know , like someone , if you see a good idea , someone's going to do it right and you just have to believe that you're going to be personal or that your team is going to be those people , and just have to push through and get in there . Excellent advice .

You know she's also deep in e-commerce , so she comes from the e-commerce background as well . Yeah , we talk about the latest trends in AI . We get on , we get awesome guests . I really enjoy it . It's a great part of my day . I think what's great about doing a podcast is you can have all the interesting people and you get to learn from them .

You can sit there and ask the questions that are top of your mind , so I recommend people to take a listen if they're interested Absolutely , and if people want to check it out , what would be a great starter episode .

Well , what we're doing now , which I would recommend people to listen to , is me and Joe are sitting down and doing reviews of the biggest AI trends . So we just did one of Q1 . So we have guests on and the guests are great and you can see if you're interested in sourcing in China , there's Gary .

And if you're interested in like sourcing in china , there's gary . And if you're interested in like you know , like ai product ideation , there's amy , we . So we've got like all these people coming on so you can look at like specific ones .

But what we're testing is like you know , we both look back on a quarter or or a year or a month and we kind of work out what we think are the three biggest like things have happened and we do a bunch of research and we get on and and then we just talk about those like three big developments . So I think those ones are yeah , I , I think they're in .

You know they're interesting and a lot of more work goes into them than the other ones . So , like I hope people are , awesome , awesome , uh .