MIKE DUBOE

Hey everyone. We just had a great conversation with Adam Turner, CEO at Postscript, which has been a special relationship for me as I’ve known Adam for the last handful of years being on the board of Postscript. Adam and Postscript really created the category of SMS marketing for commerce, and they are now creating the category of one-to-one sales for B2C businesses. AI has been a complete catalyst for what is going on with Postscript if you think about all the two-way conversational data they’re sitting on between merchant and consumer. And so we had a great conversation with Adam on how Postscript is leveraging this to really create what is a totally new marketing paradigm for eCommerce. Hope you all enjoy it. Excited to hear feedback.

Adam, thanks for joining us. This is going to be a fun one. You’re the first portfolio founder to join the pod. We’ve known each other for a while now. I’ve always admired your focus and belief in the power of SMS, not only as a marketing channel, but as a platform. When you started Postscript, SMS was really just an afterthought tactic for abandoned cart texts, and over the past few years, Postscript has really become a household name in the Shopify ecosystem and beyond, and the revenue growth has obviously been significant. You’ve also fully embraced AI as a way to transform both your product and internal ops, and so it’s going to be an interesting set of topics to get into with you. Thanks for joining.

ADAM TURNER

Yeah, excited. Good to see you outside the boardroom.

MIKE DUBOE

Yeah, so maybe to start and just lay some foundation for listeners. Let’s rewind back to 2019. What did the eCom stack look like when you set out to build Postscript? And maybe what was the core insight or opportunity that you saw that was left unaddressed by existing channels at the time?

ADAM TURNER

Sure. So before Postscript, I worked at a company called Stack Commerce, which we ran hundreds of brands. So we would run brands for publishers, so like the CNN store or the Mashable store. You would have these online publications and they would push articles to their stores so that they could sell. And it was a better model for them because they didn’t have to kick out affiliate to Amazon. They would kick out to their own store and they would make a much higher rev share on that versus Amazon. And we would have to run all of these stores, which meant we ran their marketing, we ran their newsletters, we ran the vendor selection, we did the drop-shipping for them, everything like that. So I was on the brand side before Postscript. The stack that we used was Responses, Oracle Responses, and Bronto, and these types of things. It was all just list building and list sending, rudimentary segmentation filters, but almost no automations or different things like that. And that was pretty brutal, but at the time, we were just starting to evaluate tools like Iterable or Cordial or things like that, and it was just starting to be a thing. I would say that none of those tools did SMS and SMS was not a thing. Everybody was terrified of SMS because of consumer fear. I would say, funny enough, what I think unlocked the SMS channel was service businesses starting to use it. So dentist’s office using SMS to remind you of your appointment coming up or to reschedule or something like that. And I really think that that was the start of consumers being like, Oh, business can text me. And it’s not just an annoying thing, it’s a helpful thing.

I was building a lot of different stuff at the time. A buddy of mine ran a Shopify store and he was like, “Hey, how come I can’t text my customers?” And I was like, “Oh, what if we just … If I built that, would you pay me 50 bucks a month?” And he was like, “Yeah, sure.” That was the early bootstrap version of Postscript. And as we were building that, we saw on a couple different Shopify stores that people were putting in their phone number instead of their email address more than 50% of the time. They could put in email or phone, and they put in their phone a lot more, which signaled to me that this was becoming the primary identifier for a customer versus email. And that was the major, unique insight that nobody was seeing yet as consumers were saying, Oh yeah, you can text me, it’s fine.

RISHABH JAIN

Wow, that’s pretty wild. I’m curious, upon having that initial insight … This was at the time that Shopify was starting to take off and there were some existing players and basically there would be … The other side of this is like, oh yeah, email is doing this thing, like you said. Email is doing this thing and Klaviyo is this player that’s doing it with email. What were some of the things that led you to start to have conversations with people, saying like, Hey guys, yes, you have your email channel, but you should try this new thing? And when you’re creating this category, not only does it require the insight from the founder — like you — it requires that communication between this new company and the customer to realize this new thing is being created. So when you’re doing this category creation, what does that conversation look like?

ADAM TURNER

It’s a good question because in the early days that we … We got 90% nos from people. This wasn’t a VC-funded thing, it was a bootstrapped kind of nights and weekends thing, which I think allowed us to have more patience into the adoption curve of SMS. We would reach out to people who ran sophisticated brands on Shopify and they’d be like, “Ehh, no thanks. I don’t want to bother my customers.” We heard that nine times out of 10, “I don’t want to bother my customers.” And then, one time out of 10, people were like, “Yeah, we’ll give it a shot.” And they would send their first SMS campaign and it would make 10 times more revenue than their last email campaign made. And they would be like, oh, this is a huge unlock and I just got promoted and this is great. And so the people that were in the seat to take the risk and want us to take the risk definitely benefited early on.

And while it was tough to hear so many nos from people, you couldn’t argue with the results from the people that were saying yes. And so that’s when you could really get conviction, even though the whole market was basically saying no, you could get major conviction around the future of the channel because you were just like, oh, for the people that are trying it, it’s becoming a number one revenue channel for them. It’s becoming a top revenue channel within just a few weeks or months of trying it. And that really helps a founding team get conviction because you’re just like, oh, I don’t need someone to tell me that this is a good idea. Our first customers are basically saying, I can no longer live without this thing. And so it’s only a matter of time. That seems super obvious for SMS in retrospect, because I would say that 80-plus percent of brands are using SMS today, even if it’s just for basic automations and things like that. But before, it was totally, “I don’t want to bother my customers.” And certainly the top brands, like the top 500 brands on Shopify, were not. They were like, no, we are a top-tier brand. We don’t do that. We do email. It’s definitely laughable these days because of how much it’s taken off. But really, the revenue impact is what gave us conviction.

MIKE DUBOE

So let’s spend some time in the early growth playbook. Having observed it at the time, I mean, I know the early traction there was outstanding, and there’s a couple of tactics that I think you leaned into. The agency partnerships program is now part of the playbook in the Shopify ecosystem, but I think you were one of the early ones to really realize that. And so maybe talk about that. The other topic that’s interesting is just leaning into the rise of the SMS marketer as a role. So you were part of creating a new role and persona within companies who adopted this. Where did that work for you and against you?

ADAM TURNER

Yeah, so they say that second-time founders think about distribution, and that was definitely on our minds. We had a failed gaming company before that, and one of our bigger problems was distribution. And so for Postscript and even the ideas before Postscript, we were focused specifically on the distribution channel. And so we saw the Shopify app store. It had less than 1,000 apps, probably a couple hundred apps when we first launched, and it was pretty nascent and new, which gave us a massive boost. And so everything was product-led in the beginning. Part of that was because I still had a job, and so we needed to be able to close customers without being able to get on the phone with them. And so everything had to be product-led because they would just be doing this on their own. And we didn’t want … We weren’t VC-funded and we didn’t want to spend any money on acquisition.

And so the Shopify app store was a natural place for acquisition where we knew at least some people were searching for SMS in the Shopify app store, and we would pick up some of those installs and then we would ideally convert those and boost our reviews through really great customer support. And so getting that distribution channel started was really important and cheap and it worked. I would say another thing that worked was something that a lot of people do nowadays, which is really, really smart, and it worked for us, which is getting a whole lot of angel investors in the seed round. So coming out of YC, getting folks who are in the ecosystem invested in one of the first SMS players because they knew SMS was going to be good because they were in that 10% of people that really believed, getting those folks on board really helped us get into the eCommerce ecosystem more so than … We were definitely outsiders before that.

And with that, we became insiders. Agencies, to begin with, were actually just more super customers for us. We didn’t have an agency program, they were just customers that had five different accounts, but we learned early on, one of our first features that we built in YC for folks was the agency view, the agency God-mode view, which before we wouldn’t have gotten to if we hadn’t been talking with these folks. Things that are obvious now were kind of hidden back then. I would say at first it was all Shopify app store and then it was all word of mouth because SMS was so new, and we definitely rode the wave there. I won’t give ourselves too much credit, but we rode the SMS wave. Agencies … And then I would say that for anyone in the Shopify ecosystem today, what they know is that the Shopify app store gives you almost no distribution today, it’s basically like the iOS App Store, and so you need your own channel. I would say one of the bigger things that we did for ourselves early on was we created our sales-led motion, and that drives 95%-plus of our revenue today. And so we’re not relying on the Shopify app store, but back then it was different. It was kind of like early iOS App store where you can make an app and get a bunch of downloads. So yeah, that’s a brief overview.

RISHABH JAIN

I was curious on, Mike asked this question about the roles starting to exist, the SMS marketer role, and you often see this with great SaaS tools that are in this bucket of creating a new category where all of a sudden there’s now a person whose job it is to manage the thing. How did you interact with that archetype starting to get created? And then are there things you did from a community perspective to continue to promote that or do you just let it happen? How did you think about that?

ADAM TURNER

Yeah, it’s funny. One of the early things we focused on, because we didn’t have that role yet, was that we wanted to get email marketers promoted. That was the idea. We wanted to basically say, Hey, I know that this is a risk to bring in SMS, but if you just start collecting phone numbers with us, you’re going to build up this asset that’s going to become really important. You’re going to seem like the really smart person in the company to have started building this asset of compliant phone numbers. And then when you send your first campaign, it’s going to make so much revenue that ideally you get promoted. I think that that was interesting because it was certainly a risk to take on something like SMS for a brand and for an email marketer or for a retention marketer to stand up and say, “I’m going to initiate this channel that the CEO may not like, but I think is going to be really important to our revenue, and the CEO is going to like the revenue.”

And so yeah, we just focused on helping those people make the jump. And I would say that focusing on compliance, focusing on best practices, helping show where other people had been successful is super important. And really, you just couldn’t argue early on with both the revenue from that first campaign that they sent and I would say also the conversations that they would have over the channel. We have this one customer that I really love, they’re called Old Bisbee Roasters, and this is a small eCommerce store, and it’s this guy in Bisbee, Arizona, who roasts coffee in his house and has a couple hundred customers that he talks to every single time they buy. He knows who their kids are. He says, “Hey, how are things going?” And things like that. I think people realize that it’s much more than just like this is a newsletter channel. They were like, oh, this is a relationship channel and the relationships can be fundamentally different when I use this for something like that. And so I think that people stopped thinking of it as just this promotional notification channel and started thinking of it more as this conversational channel, which helped that marketer make the case to their brand. And then that marketer eventually became the SMS marketer as well as email and retention and different things like that.

MIKE DUBOE

Yeah. Let’s go deeper on SMS as a channel, and so you’re kind of talking about the origin story here of this channel when people realize like, holy shit, obviously attention is moving from inboxes over to my SMS inbox, I want to be in there. That led to a bunch of spam early on as well. And so I think one of the areas Postscript stood out was more of an emphasis on compliance and quality controls. Maybe talk to us about that. And also, related to that, it would’ve been tempting to just do the easy stuff, take your email proxies and just run the same type of programs in SMS, treat it as an extension of email. You did more of the harder work upfront and really re-envisioned this channel net new. Maybe comment on those two points.

ADAM TURNER

Sure. Yeah. So for some background, compliance is huge in the SMS space, it’s this massive thing. And that doesn’t exist within email really, right? There’s canned spam stuff, but there are plenty of people who send emails to folks that haven’t signed up or anything else like that. There’s almost no retribution or anything else like that or damages that they have to pay. With SMS, there is this thing called the TCPA. It’s no longer federally relevant, but there are a lot of states that have then picked up this law. And what it means is if you send a text message to someone who hasn’t signed up, you have to pay $500 to $1,500 in penalties. And that’s per message and that’s per person. And so you can imagine, if you send a campaign out to 100,000 people who have not opted in, that could ruin your entire business as an eCommerce business. It could put you out of business essentially overnight. And there were a lot of early SMS platforms that were not telling you that this was the case. They were like, Hey, you’ve been collecting phone numbers on checkout forever, why don’t you just upload them to our list and you can text them? And it was a huge, huge problem in the space. And we basically said very early on, Hey, we’re going to collect numbers compliantly into Postscript. If you want to bring numbers into Postscript, they have to be compliant and we’re going to double-check that. And what that’s going to do is it’s going to help you, but it’s going to hurt our near-term growth. We took a … We were like, Okay, this is going to slow things down, but it is the right thing to do. And one, I think it protected brands a lot, but two, it really started our … what we call our gold standard of compliance. And that’s important not just to protect the merchants from these frivolous lawsuits and different things like that, but it really helps build your reputation with the carriers such that we are able to have a seat at the table with the carriers and they say, Hey, what kind of tools are you building to make sure that people don’t spam on your platform? And we’re able to show them all the AI tools that we use so that when someone signs up, we scrape what they’re selling, we scrape any messages that they’ve sent previously, we make sure that there’s nothing going through the pipes, and the carriers love that because that’s way less work that they have to do. And that becomes really important. I would say that where this helps us stand out for people that have maybe bolted on SMS as a promotional channel, someone like a Klaviyo, is that when a customer gets sued today from one of these frivolous lawsuits, Klaviyo will say, Oh, we’re just a platform. We just help you send SMS. And we say, Click this button. Here’s the entire legal packet that you will now send to this person that will make them drop the lawsuit tomorrow. Because we have every bit of data, every bit of compliance, everything from the opt-in to the opt-out to make sure that the brand isn’t in any trouble at all. And I would say that this is becoming far more important for brands because of that … the scale of damages that can exist within the ecosystem. And so for us, we made that tough decision early on, but I think that this can happen in a lot of different businesses, not just SMS businesses, but just making the right decision early on, even if it may hurt your growth, will actually become a massive competitive advantage in the future.

RISHABH JAIN

So something I wanted to also double-click on from your earlier comment about that story about the roaster where it’s much more conversational, now that we have these AI tools where you can start to do things like that, that were previously non-scalable, at scale. And I know you guys have done a bunch of testing with various formats of how to use AI to personalize your outreach. What do you think are the things that have already shown to be the most effective and then over time are going to be the most effective from a product point of view for this channel?

ADAM TURNER

Yeah, so it’s interesting because we see … With conversational over the channel, we seem to be going through the exact same kind of early-adopter loop that we went through with SMS, which is today when I say you should be having conversations over the channel, 10% of people are like, “For sure.” And 90% are like, “I’m making so much money over this, who cares?” It’s like, it’s working for me, I send a campaign, it makes way more money than email, and they’re less interested. But at the same time, what they’re asking us is, How do I decrease my unsubscribe rates? How do I increase my retention rates? How do I increase the revenue over the channel? And what we’ve done, like you said, is a whole lot of experiments over the channel about how conversations increase retention, decrease unsubscribes, and increase revenue per message over the channel significantly and over a very long period of time. I should not have to explain how this works, but I have to because so many people don’t understand. If you’re sending someone a message and they respond and you say, “We are not monitoring this channel,” that person is never going … They’re just going to unsubscribe, right? It’s like going into a mall and you go into a store and it’s just empty and you have a question and no one’s there to answer it, and it’s just empty. And so just by putting someone in that store, just by helping someone with a question that they have over the SMS channel, you’re much more likely to convert a purchase. This just makes intuitive sense, but some people just don’t understand it or they don’t want to do it at scale or something else like that. And I would say this is the beginning of what is going to be a major conversational change. I’ll tell you why. Because consumer expectations have shifted over the last two years because of this thing called ChatGPT. It’s like, consumers are ready and willing to put in anything into a text box in order to make sure that they get a good answer. And not just that, but they know that the more context they give, the better the answer is going to be. And so all of a sudden, the text box that was in the SMS channel that was pretty much worthless four years ago is now one of the most important interfaces that a consumer has with a brand. And not just that, but they understand that the more I put into this SMS box, the better result I’m going to get back. And so they will tell you everything about why they’re purchasing. “Hey, I just had a kid, I’m struggling with this thing and I need this solution.” They’ll tell you exactly why they’re shopping, all the competitors that they’ve evaluated, a question that they have about your product, all through the text channel. And I think that this has totally changed how the channel behaves just from primarily notification channel to the front door to your customer, essentially, and something where the customer is naturally incentivized to give you more information. We’re going from this breadcrumb world where, oh, the famous story that was from Target where it’s like, they found out that someone was pregnant through their browsing history and they sent a note to the house and the father didn’t know, and it’s like, whoa, isn’t that exciting? That’s the best version of pixel data that ever existed. Now people just say, “Hey, I’m pregnant. These are the five things that I’m worried about. This is the research I’ve done about it. Well, how does your product help me in that case?” We’re going from this world of just inferred data through these breadcrumbs on the internet through pixels to this direct zero-party data where customers are just willingly giving you this information so that they can get a better result for themselves. It’s a total paradigm shift.

MIKE DUBOE

Yeah, so that’s very well said, and I think transitions into some of these AI topics we want to get into. As you say, there’s so much rich context sitting in the two-way thread between customer and merchant that Postscript owning or having visibility into that full thread, there’s a lot that you could build on top of that. And I think people who are just viewing this as one kind of retention marketing channel are kind of missing the point or they’re missing the full potential. AI has supercharged your ability and others’ ability to build some of those products faster. Talk to us about, maybe just to start, how you are infusing AI into the core product today and kind of where it’s taking you.

ADAM TURNER

Sure. So I said this to our team early on, which is that, how lucky are we? How lucky are we that these things called LLMs came out, and what they’re naturally really great at is two-way text conversations between people and we work in a two-way text channel? How lucky are we as a company that we get to be in the middle of this paradigm shift? And I think that it’s truly something special because I think a lot of companies have struggled to be like, How does AI apply to my company? And they try a bunch of different things. And I think a natural thing for email or SMS was like AI copy generation, but then it’s like, what’s the difference between plugging it into ChatGPT or plugging it into your tool? There’s no real alpha there. But for us, we were like, well, what about this conversational aspect? Or what about the ability to generate hundreds of different variants of text messages at a time? We thought that was very different. And so the first thing that we did was like, okay, where is all the volume in Postscript? It’s in outbound messages. And what do people already do today that works? We know that they A/B test, we know that they try two different things, one is better, that’s great. How do we 50x the power of people’s A/B tests? And we said, well, why don’t we create an A/B test where we create hundreds of different variants at a time and then the best variant may win — and hundreds based on their traffic, it’ll depend on stat sig and everything else like that — but why don’t we just do the A/B test for them is kind of the thought. And then after that it was like, okay, well once the A/B test is concluded, why don’t we just get the LLM to learn about what worked and what didn’t? And then after that, create a hundred new variants and do another A/B test. And so we came up with this product called Infinity Testing where we said, why don’t we just continue to A/B test in the background, hundreds of different variants at a time, to make sure that the customer is sending the best thing for their audience in a given flow. Obviously that sounds like a really good idea. In practice, it’s a very hard thing to do, and there are a lot of different learnings in there. The first thing was customer voice. You can’t just have a generic LLM that provides the copy because the customers have things that they want to say, they have particular verbiage that they want or don’t want, and they’re very particular about how they communicate with the customer. So how do you build up the trust for that? Well, you build a brand center where the brand can put in all their voice and emojis and if they are a professional or a casual tone and all these different things. Also, giving them visibility into the different messages that we were creating and letting them vote up and down, giving feedback like that. And so this one idea led to this fully featured product set called Infinity Testing that drove an incremental 20% lift for our customers with any flow that they put it on. And for campaigns, it’s like a 30% lift as well, which has been a boon. I would say also along the way, we learned how important incrementality testing is. So basically, how do you prove that this lift is relevant? What we did there is we just always kept a holdout test going. So we always said, this is the infinity test and then this is whatever the brand was sending before, and always made sure that we were comparing what we were doing versus what they were doing before, which helped build a lot of trust with customers. A fun thing that we ran into there, though, was that the model found that urgency always basically got people to click and buy. And so the model would try and inject a lot of urgency into the different messages because those were converting the best. And we definitely had to curb its enthusiasm about urgency because it would always be like, Oh, we’re running out of this thing. And the brand would be like, No, we’re not. We’ve updated the model a lot since then, but even though something is the highest performing, it may not be what a brand wants because of that dynamic. And that was a big learning as well where it’s … Maybe it’s not always optimizing towards revenue, maybe it’s optimizing towards something else. And for us, it’s what the brand wants, and then also long-term revenue — subscriber LTV is what we call it. So not just optimizing for one purchase but for multiple purchases is important.

RISHABH JAIN

You’re starting to lay out this vision. It sort of reminds me a little bit of when Zuck used to go on his earnings calls and say, Hey, you’ll just come in and tell us what you want and then we’ll be able to take care of the rest. So it’s like you — in the context of the ad channel. And so, as you’re starting to say this, at least what’s invoking for me is like, hey, you’ll just come to us and say, Hey, here’s what your objectives are. Here’s what you don’t want to do. Here’s what you want to do. Here’s what your tone is and let us handle the rest. Do you see an end stage where everything from the construction of the flow, when flows are implemented, all of these sorts of things are actually decided by your platform? Should the brand want that? How far can we take this idea?

ADAM TURNER

Yeah, that’s exactly it. And there are hundreds of roadblocks before we get there. I think that it’s totally technically feasible today, but I think that there is a lot of trust-building that has to get there in baby steps that you have to do with customers to get there. Also, I think that there’s a lot of fear around this idea where maybe brand owners are like, Yeah, I would love this, but marketers are like, This is what I do. This is what I do every day is go into Postscript and optimize things. And so if it’s doing it for me, then what does my job look like? And for us, I’m incredibly bullish that the marketer will get to do far more things of why they became a marketer, the things that they’re passionate about, like what are the actual objectives that I’m trying to push? How do I actually change the strategy? How do I tune this model so that it’s both accomplishing my goal but doing it in a unique way to our brand? And I think that there’s going to be a lot more of that. There are some interesting things that go into that outcome that you’re talking about. Everybody wants to do this AI brain, the AI brain where you can just log in, give it your voice, give it your objectives, and it’s able to do the thing. But there are a lot of different steps along the way. So today, for example, we live in a highly segmented world where people think of themselves … They think of their audience as last 30-day clickers or last two-hour purchasers or different things like that. And obviously any shopper knows that they’re not like that. I don’t think of myself as a 30-day clicker or something like that. I think of myself as having a need and that’s why I went to the store and that’s why I signed up for SMS. The biggest shift that’s going to happen here is the shift from that previously structured data — which is someone clicked in the last 30 days, someone purchased — to unstructured data, which you’re primarily going to get from the conversation that you have with the customer in addition to the normal pixel data that you get from them. And so if someone goes to a hair care site and you text them and you say, Hey, why did you come to the site? What kind of hair do you have? How can I help you? And they have a whole conversation with you to say, Hey, I live in Colorado and my hair’s really dry and when it’s really dry, it gets extra oily and that’s a problem for me. And I love to go hiking, but then I’m like, my hair gets really knotty and yada yada. That is already so much more data than you would’ve ever gathered about that consumer in the history of eCommerce, right? Previously, it would’ve been like, they’re going … Maybe they take a quiz and they tell you that they have oily hair, but you don’t know with context around that and everything like that. What we can do now with LLMs is we can take that data and we can, one, give them better recommendations in that first conversation and say, cool, if you love hiking and you’re active but you want something that’s easy to use, this is what I recommend. But even more importantly is, we store all of that data as memory for that subscriber. We now know — and they willingly gave this information because they wanted a better result — they live in an area, they love going outdoors, their hair is dry, all these different types of things that get stored on the subscriber profile and, in the old world of marketing, were not previously accessible. You can’t make a segment about people who love outdoors, because how did you get that structured data in the first place? And so actually, this data that I have is unstructured and unique to that customer. So they’re like, it’s all this information that I have, this memory that I have about the customer. And so the next question is, what type of system is able to use that data to craft a better campaign for that person the next time so that the next message they get is not just, Hey, we’re having a 20% off Labor Day sale. It’s, Hey, I just gave you that great recommendation. I wanted to let you know that if you’re not headed out for the weekend on an amazing outdoor adventure, we’ve got a 20% sale going on and I reserved this item for you and I think that it’s going to be great.

That's just a much, much better experience to get there,and previously not made possible because it wasn't structured data and amarketer couldn't use it in a segment or couldn't use it in a targetedcampaign. But the LLM is able to use that very, very easily to determine the next best message. Again, there's a lot of trust to build with brands before weget here. They're going to want to go through almost every single customer andsee, okay, for this campaign, what's this customer going to get? What's that customer going to get? Why are they going to get that? And showing the reasoning behind it, showing ... time after time, you're going to have to dothis for a while, but I think we will get there. I think the biggest bottleneckis trust-building. That's on us to do, that's not on brands to do, and it's going to take a lot of product-building and everything else like that. So even if you have the revenue engine today, the AI brain today, I don't think that'sactually the hard part. I think that the hard part is deploying it in a way that builds trust with brands, that proves incrementally, and that drives the results they're looking for.

MIKE DUBOE

There’s a lot here. And maybe expanding on this topic of trust, one of the things that has always excited me about the potential here is, as your network of merchants grows, you actually have a record of an individual customer-level behavior and preferences stated across merchants. Obviously there’s — right now, Shopify is sitting on its own dataset that they’ve only been able to utilize really around Shop Pay — but there’s a question of, as this … You are sitting on a potential network effect, how can you start to let that network manifest in how you actually help merchants more deeply personalize to consumer-stated preferences across other merchants? And what are some of the lines that you have to be careful about walking there?

ADAM TURNER

Yeah, it’s both something that totally excites me and also I’m just like, how do you do this in a way that works for everyone? How do you create this rising tide that lifts all boats? I would say that, before anything, everything comes down to: do merchants consent and does the consumer consent. That is everything here. I think that consumers is pretty obvious. Once you are able to deliver an amazing consumer experience that helps them shop, they’re happy to give this data. A good example is, what is the old-world model of this? Just to give some more background, I was talking with a YC partner the other day and they were like, “I love businesses that take something that was only for the ultrarich before and bring it to people.” So a great model is Uber. People used to have private drivers before Uber, but it was only reserved for a small portion of folks, but Uber made it available to everyone.

The version for Postscript is, people had personal shoppers before, they had concierges that would go out and shop for you, or you had your person at Nordstrom that would text you when something special was in and that sort of thing, but it was only reserved for those types of people. And so how do you create that version for consumers where we will always text you when something that we think you might like is available. Or when you are asking us for recommendations, we will always give you the best recommendations in the same way that your person from Nordstrom would. And the person from Nordstrom knows what you like because you’ve interacted with them and all this type of stuff. And so I think that, for consumers, this is an easy one down the road because it’s going to make their experiences better. And we know from ChatGPT and from how many conversations they have with us that this is something they’re ready for.

MIKE DUBOE

So on that point, because I think you’re hitting on an important part of Postscript’s future is evolving from the retention into acquisition programs. And as you say, this vision of having a personal shopper forever. I mean, I used to work at one of these companies, I think now it is a lot more possible today than it was several years ago. I think it is nontrivial to go from helping merchants on retention to actually getting into the universe of helping acquire and convert customers and push them down-funnel. These tactics work, we know that. So maybe, what are some of the lessons you’re learning in that transition? And maybe talk about some of what you have cooking on shopper and conversational commerce.

ADAM TURNER

Yeah, there’s a lot here, and certainly moving from retention to acquisition is hard. One of the most difficult things about acquisition for brands is scale. I think that we saw this definitely with the AppLovin thing when it first started. People weren’t like, Oh, does AppLovin work? They were like, Does it scale? That was their question. It’s like, Am I going to be able to put $1,000 into this or am I going to be able to put $1 million into this a month? And that was a big question. And so that’s something we think about where it’s like, okay, we can definitely be an intermediary for consumers to help them find the things that they want. The question is then, how do you get that to scale such that it’s worth a brand’s time to have the conversation with you to do that?

And I think that for us, it goes back to how many consumers are you interacting with on a daily basis and are you providing value to them such that you’re interacting with more and more every single day? And that is a question of turning the channel from a notification channel into a conversational channel. That’s it. Because the amount of messages that we send outbound is far more than the amount that we get inbound. And I think that that is now a consumer behavior that Postscript has been teaching for six years is like, here’s an outbound channel, here’s a notification.

Now I would say that the brands that are most successful in getting this information from consumers and benefiting from Shopper — so let me explain what Shopper is first. Shopper is our conversational agent. It takes in that brand voice that I talked about from the merchant before, and it has all of the knowledge about the Shopify store, what’s in stock, what’s out of stock, what’s been selling recently really well, what the consumer is interested in, what they’ve browsed, what they’ve purchased, everything like that.

And it’s able to be there for a consumer any time they text in. Right now, the biggest thing holding brands back is the amount of conversations that they’re having because the default behavior on this channel is notification. We are seeing the merchants that are having the most success with this product are the ones who are changing their existing flows to add a question at the end. This is something that ChatGPT found out probably a couple of years ago, which is that if you actually just add a question at the end, you will have a much longer conversation with someone. And instead of an abandoned cart just saying, Hey, here’s 20% off, it’s, Hey, here’s 20% off, was there anything in particular that you were looking for? Or, Was there a particular question that you had about this product that’s in your cart?

And that’s more likely to drive conversion, more likely to get you information about the consumer that they willingly give to you, and more likely to retain them and drive value over time. So yeah, that’s a little bit about Shopper. I think that the biggest thing that we’re going to see, just like how people were hesitant to go from not using SMS to using SMS, we’re starting to see the transition from using SMS as a notification channel to using SMS as a conversational channel. The top 10% of brands who are doing that today are benefiting by far the most.

RISHABH JAIN

This makes me think two things. The first thing it makes me think is, there’s all of this rhetoric around you use B2C tactics in B2B marketing, and this is now a world where you’re using B2B tactics in B2C marketing. Asking a question at the end of a message, that’s what my head of sales yells at me for. Every time I send an email and he’s like, “Rishabh, you know why you don’t get responses? It’s because you don’t ask a question!” I’m like, Okay, okay, I get it. So it’s like we’re getting to do quote-unquote ABM at a scale that has never been possible before in some ways. So that’s one dimension of what it makes me think.

And I’m wondering, when you look at the customers that belong to that top 10%, do those marketing teams have more sales DNA in them, and is that what is moving them toward that approach? Or are they just more experimental? What are the characteristics of those teams that are bringing them into that 10% of people who are taking advantage of this new way of thinking about marketing at the consumer level?

ADAM TURNER

Yeah, that’s a phenomenal way to put it, and that’s exactly what we thought early on when we called our initial product that was human-powered at first, we called it SMS Sales because we had this philosophy where — exactly what you’re saying — what do great SaaS businesses do well? They have this sales motion that allows them to objection handle, that allows them to do follow-ups, that allows them to do all these things. And why hasn’t D2C gotten there? And it’s obvious. Because in the past, this was not economical at all to do, period. And with LLMs, (snaps fingers) overnight it became economical to do the same type of sales motion that you would do for a high ACV SaaS product for a incredibly low ACV bar of soap on the internet, right? It’s incredible.

And so that was the exact philosophy behind the product was that, what did the best salespeople do and how do you actually make that economical for merchants to do?

In terms of the DNA of the merchants who adopt this, do they have more B2B DNA? That is a phenomenal question because that’s people who understand sales and are able to do that. If I had to guess, I would say yes. I would say that the main thing that ties our merchants together that are using Shopper really well, that used previously SMS Sales really well, is that these are incredibly high-performance marketers that are willing to experiment with anything and have in-depth detailed internal processes to understand if it’s incremental. It’s folks that are, they’re like, Yes, I’ll try out this thing, but if it doesn’t prove out in my models, then we’re just going to turn it off if that’s okay.

The good news for us is that the models end up panning out. We do holdout tests, it ends up working. But I would say that the folks that are getting ahead are for sure the ones that are both experimental but not wildly so, what they have [is] shiny object syndrome. They’re saying, We’re going to do a two-week test, we’re going to prove incrementality, and then we’re going to understand how to deploy dollars into this thing based on that.

I’d say those are the folks that are willing to experiment with this and doing the best job. The folks that I see not willing to do this, for example, are folks that are probably under-resourced and understaffed. They’re having a hard time right now, and they’re just like, I just need to save time. I understand that it might be an incremental 20% over the channel, but that’s not where I am right now. I’m in treading water because I’m wearing five different hats. Those are the opposite sides of the spectrum.

And that’s certainly, I’d say what we see within eCommerce today. The folks that are winning and doing really well have more room to do this stuff. And so there’s this virtuous cycle where it’s like the more room you have to do, the more you figure out that Shopper is working for you, the more you deploy into it, all that type of stuff. And the more you’re trying to tread water, the less likely you are to try new things.

MIKE DUBOE

Adam, as we’re nearing time here, I think one of the topics I wanted to wrap on is your usage and transformation of internal ops via AI. You’re now running a pretty at-scale marketing company, and you got started before this whole wave, and I’ve seen you lean into it quite heavily reasonably early, and I think it’s not just in the core product, but in how you are running Postscript right now. Talk to us about some of the most impactful things that you’ve done there and maybe where you see this going.

ADAM TURNER

Sure. Yeah. So there is the obvious stuff that I won’t talk about, which is the Toby memo and the importance of writing your own memo and also getting your team on board with using ChatGPT … Those are the table stakes. Everybody understands that now. I actually wanted to share the stuff that’s actually interesting. So there’s the idea of LLMs will make your engineers 10x, and this is a fight that’s happening between the engineers and other folks, which is like, how are they actually 10xing, all this type of stuff.

I posit a different approach to this, which is that mildly technical people are now the 10x engineers. They’re 10xing their own output. I’ll give you a specific example of this. Our sales engineers are now some of the biggest builders in the company because these are folks that understand what JSON is, they understand how to their way around SQL, all this type of stuff. And also they deeply understand business problems because they’re on the calls with customers, they’re in it with the VP of sales, they understand the marketing challenges, everything else like that.

And so our sales engineers have been able to build internal tools, going zero to one very, very quickly with LLMs, and they don’t need to worry about things like — it’s all internal tooling — so they don’t need to worry about security infrastructure for customers because these aren’t tools for customers. They’re internal, they’re within our own infrastructure, only Postscript people can log in.

An example is … So I don’t know why Gong hasn’t done this, but Gong has not leaned into the AI. You cannot really search — deeply search — calls. We wanted to bring Deep Research to our own Gong calls, like actual o3-deep-research. We’re going to add ChatGPT-5 deep research to our calls.

And so our sales engineers exported all of our calls with the API, put it into a database, ran all this analysis on the calls, and then created an easy Slackbot that anyone in the company can ask, Hey, what are the top five features of popups that people have brought up on recent calls? And it’s crazy for PMMs, for PMs, to be able to get this information in five minutes, right? Or a BDR can ask, while they’re on a call with a customer, Hey, I have a customer that’s on a headless shop. How have we dealt with that previously? And it’s like, oh, here are the last five calls. Here’s what we say about that.

And it’s not just this Guru knowledge base, it’s actually taking live calls that have happened, live recorded calls, what are the more recent takes of this? You don’t have to worry about updating and verifying Guru and all these types of things. And one of our sales engineers built this in just a few days, and they’re continuing to build these tools.

I think that internal tooling is the name of the game right now, and I think that having slightly technical folks with a lot of business context solve internal problems is actually the way that you reduce headcount and end up doing more. I think that there’s the obvious stuff with customer support, all the type of stuff, but I think that arguing the pure engineers can all do 10x is not the efficient way to do it right now. It’s all focus on the low-risk, high-rewards things that haven’t been able to be done before because you didn’t have technical people in those roles.

MIKE DUBOE

Adam, this whole conversation has been a great reminder of why I was and continue to be so bullish on everything you’re cooking up at Postscript. And so yeah, we could have kept going for a lot longer, but I know we’re approaching time. So thank you for jumping on today and hopefully this gives more people a flavor of why this channel has such potential, why Postscript is taking a very differentiated and long-term oriented approach to this and why AI is just really rocket fuel to everything you’ve been doing to date. Thanks for joining us. Yeah, appreciate it.

ADAM TURNER

Awesome. Thanks guys.

RISHABH JAIN

Awesome. Thanks.