Limited Supply
Limited Supply

S17 E4: Stop Building Landing Pages by Hand...Here's What's Next

7/29/202626:504,817 words
0:000:00

Nik breaks down what's actually new with GPT-5.6 Sol and why it's a bigger deal for marketers than most people realize. He walks through how he's using it inside his AI agent "Jet", why hyper-realisti...

AI marketing brief · Qwen 3.7

AI-Driven Landing Page Workflows: Balancing Automated Execution with In-House Strategy for DTC Brands

The short version

This episode explores modern workflows for building high-converting DTC landing pages using AI tools. It emphasizes keeping foundational strategy, research, and wireframing in-house to preserve brand context, while automating execution tasks. The discussion also covers utilizing AI for post-launch QA and the increasing need for rapid landing page and ad creative production velocity, including influencer-specific pages.

Synthesis of podcast conversations. Speaker claims are not independently verified.

01The takeaway

Keep Landing Page Strategy and Wireframing In-House

Outsourcing foundational strategy to agencies often lowers conversion due to a lack of deep brand context. Marketers should own the angle and initial wireframe internally, using AI to accelerate the process rather than relying on external vendors for core strategy.

Transcript evidence

What the speakers said

The host contrasts an agency design with an in-house AI-assisted design, noting the in-house version was superior in UX and copy because the client possessed deeper knowledge of the user mindset.

02The takeaway

Feed First and Third-Party Customer Data into AI Agents

Marketers should aggregate first-party data like reviews and CS messages, alongside third-party data from social platforms, into a comprehensive document. This context-rich manifesto provides the necessary background for AI agents to generate accurate, brand-aligned wireframes and copy.

Transcript evidence

What the speakers said

The host advises compiling a detailed document of customer sentiments, patterns, and objections from various channels to serve as a context-rich manifesto for AI agents.

03The takeaway

Use AI for Wireframing but Retain Human Designers for Final UI

AI tools are effective for rapidly generating wireframes and structuring layouts. However, final UI design and development should still be handled by human professionals to ensure a premium aesthetic that builds consumer trust, unless a fully modular design system is in place.

Transcript evidence

What the speakers said

The host states that while AI can efficiently execute wireframing tasks, human designers are still required for final UI to establish a premium aesthetic, noting that poor design signals a lack of brand quality.

04The takeaway

Leverage AI for Post-Launch QA and Heatmap Analysis

After a landing page is live, AI can analyze behavioral data from heatmaps and GA4 to identify friction points. This data can then be fed back into the AI agent to automatically inform and optimize the next iteration of the page wireframe.

Transcript evidence

What the speakers said

The host suggests using AI to analyze post-launch behavioral data, such as identifying if a poorly designed pop-up close button is causing abandonment, and feeding those insights back into the AI for the next iteration.

05The takeaway

Utilize Influencer-Specific Landing Pages for Social Traffic

Creating dedicated landing pages tailored to specific influencer traffic is a tactic utilized to optimize conversion rates from social and influencer campaigns, addressing the increased demand for landing page velocity.

Transcript evidence

What the speakers said

The speaker mentions personally building influencer landing pages alongside ad variations as part of a rapid testing workflow to meet the increased frequency of needing to build pages.

From listening to doing

Ideas to test

Suggested experiments, not proven results. Choose what fits your brand.

  1. 01

    A/B test a landing page where the strategy and wireframe are built entirely in-house using AI agents against a page where the wireframe and strategy are outsourced to an external agency, measuring conversion rate and add-to-cart rate.

  2. 02

    Test the impact of feeding a comprehensive customer manifesto compiled from first-party reviews and third-party social comments into an AI copywriter versus prompting the AI with only basic product features, measuring scroll depth and time-on-page.

  3. 03

    Implement an AI-driven iteration loop where heatmap and GA4 drop-off data from a live landing page is automatically fed into an AI agent to generate a revised wireframe, then A/B test the AI-revised page against the original human-optimized page to measure conversion lift.

  4. 04

    Implement a new landing page templating tool or workflow and measure the reduction in hours required to launch a new paid social ad creative landing page.

  5. 05

    A/B test a dedicated influencer landing page against the standard product detail page to measure differences in conversion rate and bounce rate for traffic sourced from influencer campaigns.

Context & limitations
  • The speaker references specific AI model versions and tools which may be proprietary, unreleased, or hallucinated names; their actual capabilities and availability should be independently verified.
  • The claim that in-house teams will always outperform agencies in landing page conversion relies on the assumption that the in-house team possesses superior brand context and AI proficiency, which may not hold true for all organizations.
  • The effectiveness of feeding large volumes of unstructured first- and third-party data into AI agents for copywriting depends heavily on the specific AI model's context window and reasoning capabilities.
  • The transcript includes sponsor pitches for CTV platforms which were excluded from this analysis, meaning the discussed strategies focus solely on landing page CRO and AI workflows.
  • The transcript contains a likely transcription error using the word adversarials instead of ad variations, making the exact nature of the produced items unclear.
  • The speaker's claim of producing multiple ad variations and landing pages in a single session is a self-reported personal anecdote and may not reflect typical team workflows or the quality of the output.
  • The specific tool, software, or method used to achieve the claimed production velocity is not detailed in the provided transcript snippet.

Transcript

EN

Welcome back to Limited Supply, the podcast where we get deep into the tactic...

side of e-commerce, digital marketing, and building consumer brands.

I'm your host, Nick Sharma.

I've spent the last nine years building, scaling, and investing in brands, and through

this show in my weekly newsletter at Nick.co/email, I'm here to share everything I've learned. The wins, the losses, the experiments, the tactics, and the insights, also you can unlock your next $100,000 in revenue. Today's episode is a good one, but before we dive in, let me tell you about our chosen

sponsor for this week's episode. Here's the problem with most TV platforms.

You'll only get access to about 20% of the inventory, which is the programmatic slice, and

you can't measure it the same way that you measure meta. Tatari fixed both, direct publisher buys across linear and streaming, and, next-day attribution against cack and row as, which is why 400 plus D to C brands like Manscape, Tacobas, and Chime,

all run to Tari as a TV performance channel.

You should see how they do it. Check out nik.co/ta-r-i. That's nick.co/tatari. Welcome back to another episode of "Limited Supply," I'm your host Nick Sharma. And today, today I just want to ramble a bit, and mainly kind of talk about something

that has come up a couple of times in the last two weeks. So, you know, I've kind of shared a little bit of what I've got going on, a little bit of it is consulting, you know, working with brands of different sizes on just a couple of different things that I'm mostly focused on, which is usually landing pages or growth or building a better system, sometimes it's putting a better process in place.

But all kind of in the world of D to C, the other thing is working on bringing a couple of products to market, and then the other thing I'm spending on my time in is the world of AI. So, being in that world of AI and also then consulting with brands on things like landing pages or growth or whatever it is, I find myself kind of figuring out a lot of things

on my own. And I think one of the fun parts about this podcast is then figuring out how do I bring those ideas to you guys and kind of just share what I'm working on. And a lot of times I find that when that happens, people go and test it out because, you know, if I'm able to do it, most people are able to do it.

So people test it out and they get some cool results out of it. So that's what we're going to do today. I kind of want to just actually talk through this whole landing page building process. So, you know, I've talked about AI landing pages in the past on the podcast. And every week that goes by, there's new updates that come out in the world of AI.

And you know, there's a joke that you have to basically be unemployed to who stay up

to date with the world of AI, which I guess I am a little bit to some degree. So that's why I'm here to tell you what to focus on out of everything that came out that I saw. So a couple weeks ago, or maybe it was maybe even a month or two ago, when we were talking about AI landing pages, you know, I think it was basically at the point of, okay, these

tools are now out and you can build individual pieces of this or multiple pieces and then kind of ship it, but it wasn't like, you know, tried and tested and true. So for the last couple weeks, I've actually just been going super deep in this. You know, there's been a bunch of other updates. So GPT 5.6 came out, which is the latest OpenAI model. That model has three different versions of the model, kind of like a light

medium and heavy, which that heavy one is 5.6 soul. That competes with Claude's Fable 5, which, you know, famously got banned for like two, three weeks after it first came out,

because it was so powerful. So that's like OpenAI's model that came out, and it's actually

very, very good. It's the first time I've connected a model into my Hermes agent, Jet, who I've talked about before, and felt like, okay, Jet is now thinking two steps ahead of me. When I say something, if I ask for something, or if I'm working on something, he's actually thinking two steps ahead of me and getting me more things or bringing things to me. Anyways, so those new models came out, which are also very good when it comes to design, and especially, very good

when it comes to imagery, very hyper realistic photos with your product or your bottle or your capsules or your clothing, whatever it is, right? So that's one big update. The other one is, oh, and by the way, what's cool is one of my sponsors recently was this company journey in my newsletter, and J.U.R. and I've talked about them probably six to eight months ago, before there was any

Kind of relationship with a newsletter.

where he was saying that he was creating, like, listicles and hero product bundle pages using journey, and then I checked it out and used it for a brand and thought it was pretty phenomenal. But, you know, what I actually, I guess, want to talk about, too, is like today how this whole process of what journey does. Well, I guess what it is more, more elongated out kind of like into the process that I use, but then also which parts of that should be done by you as the marketer

or the brand, and which parts of this should be done by, you know, external vendors, whether it's a freelancer or an agency. So the first time this whole idea came about of, you know, how to think about this new approach of building landing pages was a few weeks ago with the Shopify additions, but more recently, you know, last week I had two phone calls. One was with a supplement brand that,

you know, is doing, I think, 300 million online and basically looking to figure out how they kind of

crack the code again on meta like they used to. The second one was with a brand that is doing about half a million dollars online and, you know, they've got an agency that they hired to, to basically help them optimize some of their web pages and, you know, the differences, I guess,

like the 300 million dollar brand here, they're basically, you know, they're, they're spending

a bunch, they're playing this game of like you acquire subscribers through meta and, you know, make money on the LTV, which is a great game to play. The, you know, they have a fold like internal design team, they develop their stuff internally. So for them, their restrictions are not really like the people to execute or get the project going, but rather, you know, how do you actually,

like, what is it that we're testing and then why and how do we measure it and how do we do it?

And, and then the other brand is more like, you know, they're, they're much smaller. So they have hired a external agency to essentially help them figure out, like, how do they just build better higher converting pages? And so, you know, as we started talking through it, I started to realize that today, in today's day and age, like you could previously have done this whole thing with an external agency, but unless that agency is actually really good at the strategy and insights

and research piece, you're, you know, you're, you're very likely not going to have something that converts like it should as or or it would if, you know, that those pieces were done internally. And then it just got me to think like, well, with where, you know, where we are with these AI tools, it's actually a lot more reasonable to do most of this now in house. So, you know, when I think of

landing pages, I think of this process of, you've got the idea, right? So, what's our idea for the page?

Or why did he even become an idea? Then you've got the strategy. So, you know, when I think of strategy, I'm, I'm usually thinking about the angle, who is this page four? What's the page's purpose?

You know, how is this going to be different basically than other pages? And then you do the research,

now the research is where, or actually I'll, I'll go through these pieces more in depth, but, you know, then you've got research, and then you've got copywriting, and then you've got UX design, and then UI design, and then development, and then QA, and then, you know, iteration. So, starting with the research piece, you know, research is where you need to dive in to a bunch of first and third party data sources. Now, what are examples of this? First party data sources here,

would be like your customer reviews, your comments in any of your ad platforms, whether it's Facebook, or TikTok, or YouTube, your customer service inbound messages, whether it's questions, or complaints, or concerns, you know, all, all those three are actually very valuable to understand,

because, you know, you're basically trying to crack the code on what is the mindset

or what is going on in people's head as they're in this journey of buying with you. Instagram, or TikTok, or really any social platforms, DMs, great place to source. Basically, all the information of what is that your customers are actively thinking when they're about to make a purchase, or they have concerns about kind of very similar to customer service, but I think it's a bit more, you know, closer to the battleground in terms of like what are people actually thinking

in real time? If you do any events, you know, those are usually a great place to grab quotes, or understand how people perceive your brand or product, and then the other one I'll say is if you

Work with creators, like if you've got whether you've got your own kind of cr...

or you just work with large influencers, or even small ones, but, you know, a lot of times when

these people talk about your product or their posting, they're actually getting real-time feedback

as well from their community about what it is that you're selling. So, if you're able to grab

and hold on to and, you know, basically use any of that, all of that is, so those are actually

all just examples of first-party data. Third-party data sources, but sorry, I was going to say all of that is what you want to feed into your, to this like, AI brain, or, and for me, it's jet, for you, it might be your Hermes agent, it might be just cloud code or codex, whatever it is. Third-party data sources you want to add in, well, the ones I like to think about are, well, first I just go straight to TikTok, and I'll just search the brand name of the product,

and I'll just start watching videos. And, you know, you'll find ones that are really insightful, and you'll find ones that are just kind of silly, but what you want to do is basically grab patterns.

So, you know, either feed these into jet, or your Hermes agent, or whatever your AI system is,

I want to just start saying jet. So, if I say jet, just receive that as, you know, your AI system of choice. So, I usually feed that right into jet, and then have jet like monitor that, you can do the same for Twitter, same for Reddit, and then, you know, and YouTube, of course, how YouTube's a big one. And then, of course, just go to Google and just search for other places that people are talking about your product, it might be like bodybuilding.com,

you know, it might be these third-party forums, it might be reviews on e-retail sites. So,

you know, maybe a different website is selling your product, and you can grab the reviews from there. But, in either case, once you find these sources, you can again tell jet, hey, continue to monitor these sources, and see if there's new things that pop up. So, that's where I like to do the

research, or rather grab the data sources for research. And then, the second part of that research

is, you know, creating like a doc, I usually like to create a Google doc that has all of the notes of the things you want to grab. So, for example, if there's things about how, you know, if there's patterns of what people are saying in the reviews that they enjoy most, write that down in great detail in this Google doc. If you've got a pattern of certain traits, people recognize in customer reviews positively, or in customer complaints through, you know,

gorgeous or customer service, then note that down as well, because you want to make sure you have a counterpoint for that in this doc. This doc is basically kind of like your manifesto for this landing page, and you want to put more things than you think you would probably need than fewer, because you want to give, you know, we're giving us all the jet. So, we want to give jet as much information as possible, and as much context, because more context that has the more it'll start to

be able to think like us. And when I say think like us, I mean more like, you know, when you think about putting something on a page, it's because you have that additional context, right, from before. So, the more context you can give, the better chance it has to get to those conclusions on its own. Now, after the research phase, you've basically got this massive Google doc, and the job is to then turn this into, you know, a wire frame with copywriting on it. Now, this is usually the part

that, well, that's not usually, but, well, this is, yeah, this is usually the part that you would have to get a designer involved. And, you know, a lot of times, I would just try to build this

part and figure myself without a designer and my wire frames are always looking super messy,

because, you know, I'm just adding in basically rectangles at text. But what it did do is when I would work with the designer on the UI portion of it, everything was already snapped into place, how I wanted it. So, there wasn't this kind of back and forth of getting the wire frame in a place where it looks good and reasonable, like, from my point of view as a marketer, and then also doing the same for UI, or sorry, UX and then UI. So, now, what I've started to do is basically

put together, that you can do this two ways. I did one way where I drew it out on a piece of paper, use that to turn that into a brief with a codex or a cloud code, and then I'll give that to manists to actually build the wire frame. The other way to do it is you basically, you know, use whisper flow or willow, one of those apps that are text or speech to text with AI, and you basically just ideate this page and kind of speak it into codex or cloud code, turn that into a brief,

and then give it to manists. But in either case, I found manists is probably the best tool for helping to build wire frames, more so than cloud or codex. And I think it's just because the way that

Manist works is, you know, manist doesn't use its own, it's not its own model...

use open AI models, and cloud code, you use anthropic models. But manists sits on top of all the models, and will take your task of building a wire frame, and then divide that task into 70 subtasks, and use different models for different subtasks to create the best possible final product, and then it uses its own AI to put all those pieces together. There are a lot of CTV platforms claiming

they'll unlock growth for your brand, but here's what most marketers don't realize, they're not

all the same, and choosing the wrong one is expensive. Most of them are programmatic CTV only, which sounds good, until you realize that's only about 20% of what's actually on air. Tatari is different. They run convergent TV, which is direct streaming and linear together in one platform. So, you're reaching more customers across the full TV landscape on premium inventory, not just the programmatic slice. That's exactly how brands like man'scape, liquid IV, and

chime are driving both performance and real brand scale at the same time. See how they did it at nick.co/tatari. So, manists is phenomenal for apps and webpages, and, you know, wire frames. Where manists is not good, and I would say most AI is not there yet, although I think we're just weeks, or maybe a month or two away from it, is much better UI design. So, this is usually where

I would recommend. So, basically, this process up until now, right? Idea strategy, research, wire frame, and copywriting. That should all be done in-house by you, and that's kind of the punchline here of this episode, is you cannot outsource this piece, because no one knows your brand better than you. No one has as much context as you, and I'll give you a quick example. Second brand, I talked to you last week. They showed me a figment design of what their agency

put together, and then, you know, she showed me a design that she put together in clawed code herself based on what she knows about the user, and not only was, you know, from like a copywriting, or just a fundamental, you know, page, angle, and punchlines, like, from all those standpoints, it was way different, but even from a UX standpoint, like how people go about thinking through

their variant selection, or, you know, the step-by-step process of that, or what you need to know

as you're going into an add-to-car button, all those things were fundamentally different, like completely different, not even close to what the agency put together, and it was just another example of why this, it's so important to do this piece in-house. If you're spending, you know, 10,000 to $100,000 to $1,000 to $10 million a month on Facebook, and you're driving it all to a page that you are not taking the initiative yourself to go in perfect, and you're just putting it off

to an agency to do, this is all your fault. You cannot blame the agency for not building you a high-quality page, right? An agency is really good at the process of executing that page, but if you're, if you're bringing in an agency, and especially just bringing one in for the purpose

of building a landing page, these are people who have not worked with you before, maybe have never

tried your product or even hurt your brand, then there's no way you can expect this agency to out of the blue, you know, knock it out of the park. They might have been able to do that for other brands, you know, maybe the people involved have domain expertise on something similar, but again,

they don't know your brand, so that's why I can't stress enough how important it is for this whole

process up through the copy and the wireframe to be done internally, and especially now with madness or cod code, you can do all of this stuff, and you barely have to lift the finger. In fact, you can speak most of it, and it'll all get done. The next piece of it is where I'd recommend you bring designers in, so whether you use a freelancer, a contractor, you have an agency you're already work with for web work, you have internal designers, you I design, you want that to look

really nice, really premium, really luxurious, really bougie, and it also, you know, good UI design, it's like a good lifestyle image on your product page. It just sets the mood, it sets the tone,

and, you know, it's really a big trust signal, which is something I always like to say too.

That part you should do with a person for sure. Dev, highly recommend, also probably doing that with a

person. The only exception would be, you know, if you've already got a design system built out,

For example, we've talked about brands here who rip landing pages, you know, ...

pages, and get them going relatively quickly, it's because they've got a design system already of

place, in that case, they don't even need the UI designer because everything's already built out,

they're just swapping imagery and text usually. But unless you've got a modular system already built out, which I have seen some brands actually now doing with Framer, which is, you know, though they'll have Framer kind of set up where they've built all these bottles, and then the marketing team can go in and just move modules around or swap things out. And, or they can connect jet via API and a Framer to go ahead and rearrange those pages or build new pages, etc.

If you don't have that, do it with a real developer because, again, it sets the tone, it sets the mood for the page. When you have a page that's properly designed and properly developed, it signals

high trust, right, because good brands would never have a bad page. Only Janki brands would have

Janki pages. The last piece is QA, before we go to iteration. So QA is, you know, you can give

this one AI as well, because you can have it stress test the page on multiple devices and simulations.

But, also, I just recommend you go in from Wi-Fi, non- Wi-Fi, iPhone, Android, small iPhone, the regular Pro size iPhone, then the Pro Max as well. And then, of course, different laptop screen sizes. And then, you know, really, here you're looking at like pixels, are events firing as they should, are they being reported back? You know, if you have adversarial domains, you want to make sure you've also also whitelisted those domains in your ad platforms as well, because otherwise they're

going to get flagged. And then, you're looking for, you know, all your buttons working and all your callouts. You know, if you've got an ex-in-intent pop up on the page that works, if you've got an offer that pops up as you scroll the side, you know, make for that pops up. CTAs, floating CTAs, like, all the whole works, right? And then the last piece is iteration.

This is where you've now got the page live. It's your driving traffic to it. And you've got

something like a clarity or a heat map on it. You've got GA4, you know, you've got one, you know,

basically, all these tools that are tracking the page. And then you're feeding all this back into

jet as well. And jet will say, oh, you know, well, heat map absorbed that these people were spending most of the time here or that, you know, when people went this route, they were more likely to buy or, you know, your email pop up is causing people to, you know, not know what to do because the X is too small or it doesn't say no, thank you. So people just abandon the page. But all these insights that you would otherwise not see because you're not spending, you know, time looking

at the recordings or just the fact that jet can do it in real time, right, instantly, and then make these assumptions. All these things can be done by AI and then it feeds right back into when you wireframe the second time or wireframe the third time. And it just makes it all smarter. So that's basically the system that I would use to start cranking out these pages. Now the pages, you can actually get them deployed and Shopify. You can also, like, if you don't, don't have a

developer, but you want to get these pages live and you're building listicles or much simpler designed pages, then I would recommend you actually one build out your design system, you know, in GitHub, whether it's in a Claude, Claude or Codex, you can use either actually to build the design system. You can also use Manus to build yourself a brand book for scratch, like a true

80 to 100 page brand book, which is pretty amazing. You want to do this in general for any project

you work on because then this becomes, you know, where you pull from. So for example, whenever I build a project, I tell any agent to pull the branding from identity.nick.co. And if you go there, you'll just see it's like a brand design book for myself, but it makes sure that any of the projects I work on have design continuity across the board. So I would recommend you do that and then you can actually use Manus to get, you know, it's not going to be a perfect UI copy of your site, unless you're

site super basic, but you can get, you know, you can create a tutorial pages or influence or specific landing pages that might be slightly off from from your how your main website works, and you can get these pages live pretty quickly. So yeah, so if you want by the time this episode is out, it'll be Wednesday. Hopefully some of you can listen to it and maybe build a couple of these landing pages and DM them to me. I'd love to look at them and even give some feedback or

pointers or tips. I've been doing a bunch of help just over Twitter DMs, which has been a bunch of fun getting back on Twitter around this kind of stuff. So if you happen to build a page,

I'd love to see it DM it to me.

What's stopping you from building out some of these pages? Is it like a political process thing

at your company? Is it that your unclear how to go about it? Because this is becoming a very,

this is something that is coming up now, almost every other day or every day. And it's something

that's going to be resolved in the way that I just mentioned it, or even more consolidated and simple.

So if you're not getting what the time's now, then, you know, you might be in a bit of an uncomfortable

spot later. So I'd love to help you get across that hump because I'm also going through it now. So we might as well all learn this together. All right. That was one big ramble for today's episode. I hope that it inspires you to build out a couple of these pages and test them. I spent a night

at the office last week. Basically, literally the whole night, till about four in the morning,

I was able to crank out nine adversarials and two influencer landing pages in about five or six hours. And that was kind of learning the whole process from scratch there. So I think I'll be able to do it faster and once I perfect this process, I'm happy to talk about it here too. All right.

Well, DM me on Twitter. If you've got any questions, if you want to show me what you're building,

and I'll see you next week. Thanks for listening. We'll be back next time to cut through the noise on C.P.G. Retail and Ecommerce. If you enjoyed this episode, why not share it with a friend and be sure to subscribe wherever you listen so you don't miss the next one.

Compare and Explore