Humans of AI: Presented by WRITER
Humans of AI: Presented by WRITER
When the Funnel Collapses: Rebuilding Marketing with AI Agents | Christian Westcott
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
For decades, the inbound marketing playbook was clear: create content, rank on search engines, and capture leads. But what happens when that machine breaks?
In this episode of Humans of AI, hosts Alaura Weaver and WRITER CMO Diego Lomanto sit down with Christian Westcott, a search industry veteran who watched the traditional top-of-funnel collapse as AI search engines changed how buyers find information. Instead of denying the shift, Christian reframed his entire role.
We explore the messy, uncomfortable, and necessary process of tearing down your own expertise to build something new. Christian shares how he moved past the fear of obsolescence to build AI agents that eliminated 95% of his team's operational friction, saving 80 hours a month on QA alone. More importantly, he reveals how he's pioneering new ways to measure and capture "AI visibility" in a world where traditional attribution is getting cloudier.
If you're a marketing leader trying to figure out how to adapt your team's workflows for the AI era, this episode offers a pragmatic, optimistic look at what's possible when you stop improving the past and start inventing the future.
Listen to discover:
- Why the shift from search to AI is a move from productivity to differentiation.
- How to identify the "cluttered pantry" workflows in your team that are ripe for AI automation.
- The reality of building AI agents: why false positives and iteration are part of the process.
- How to measure success when traditional attribution models no longer tell the whole story.
- The "Monday Morning Action" you can take to start transforming your team's work today.
Watch the full unedited conversation on YouTube
Subscribe to Humans of AI for more stories from people navigating the intersection of business and artificial intelligence.
Watch the full video interview on the WRITER YouTube channel for bonus content and deeper insights.
Learn more about WRITER at writer.com.
For decades, we knew exactly how the search game was played. You create content, you rank on Google, people click, they read, they fill out a form, and eventually they buy. It was a machine. And if you knew how to turn the dials, you were invaluable. But what happens when the machine breaks? What happens when the very thing you spent your entire career mastering, the thing that defined your value to the company, suddenly stops working?
SPEAKER_00Sites, including ours and our customers, and pretty much every site out there that's reliant on, at least informational queries, started to see their search traffic decline.
SPEAKER_02The top of the funnel hasn't just shifted, it has collapsed. And for those of us who built our careers on that funnel, it forces a deeply uncomfortable question. What defines our value when AI does or completely changes what we used to do? I'm Alora Weaver, and this is Humans of AI. Today, we're looking at the messy, uncomfortable, and ultimately necessary process of tearing down your own expertise to build something new. We're also going to hear from my boss, Diego Lomanto, our chief marketing officer. He sat down with Christian, and if you want to see the full unedited version, check out the Writer YouTube channel. The link is in the episode description. To understand where we are, we have to understand where we came from. Christian Westcott has been in the search industry since the 90s. He was the first marketer at writer, and he built our inbound funnel from the ground up. He knows this world better than anyone. But about two years ago, the ground started shifting beneath his feet.
SPEAKER_00In the old world, you put out some content that would attract people who are carrying out an informational query in search engines. They click through to your website, you'd nurture them. That whole top of funnel has really collapsed now.
SPEAKER_02Collapsed. They aren't clicking the way they used to.
SPEAKER_00People are getting everything they need from interactions with LLMs. They are using Google for the validation stage, let's say, just to check that everything they've learned is correct. And then by the time they come to your website, they already know your product you want to they want to evaluate.
SPEAKER_02This is the shift from productivity to differentiation. It's no longer about producing more content to capture more clicks. It's about how you show up when the AI is doing the synthesizing for the buyer. But recognizing that shift is one thing. Actually changing your day-to-day work, the work that made you successful, is entirely different. It requires a level of vulnerability that most of us are afraid to show. This is just a fad. People will always click through to websites. The denial was loud. Some practitioners doubled down, saying things like, very little has actually changed in SEO for years, and advising teams to just keep doing SEO as you were. But by September 2024, brands started noticing the shift. Traffic to home pages dropped, but product pages still saw activity. Then holiday sales pattern shifted. And by 2026, AI visibility became the number one priority for CMOs. 33% are now reporting share of AI voice as their top metric, overtaking traditional search at just 21%. Christian saw what I saw. He looked at the collapse of his domain, and instead of dismissing it, he reframed it.
SPEAKER_00And it it didn't really mean that I had to completely upend my entire role. A lot of things I was doing already, but it was reframing the importance of those things and how they fit into this new world.
SPEAKER_02Reinventing your role isn't a clean process. It means taking apart the things that already exist, pulling everything out, deciding what's worth keeping, and building something better from scratch. Think about it like a cluttered pantry. You can't just rearrange items on the shelf. You have to empty it completely, toss what's expired, and create a new system that actually works. Christian's cluttered pantry was the website management process.
SPEAKER_00Looking at my team and the work they were doing, managing the website is a time-consuming process, keeping a website fresh and relevant and on-brand, you know, that that takes a lot of time. So I wanted to see how we could utilize AI to basically streamline the work that they were doing every day.
SPEAKER_02Christian started looking for the friction. So he did what any good experimenter does. He gathered data.
SPEAKER_00So I took 10,000 previously completed tasks that we'd worked on, and then I used Writer Agent to run an analysis of all those tasks with the goal of designing an Asana intake form that would ask all those necessary questions up front so we didn't have to have the back and forth. So it did the analysis of the 10,000 tasks, then it prototyped an Asana intake form for me based on that. I then worked back and forth with the agent to refine it, add in some different use cases.
SPEAKER_02We talk about AI transforming business, but transformation doesn't happen in the abstract. It happens in Slack channels, it happens in Asana tickets, and it happens through iteration. Christian refined the intake form and in doing so discovered something bigger.
SPEAKER_00Well, with that intake form and the analysis it did, it actually revealed that the team was spending a huge amount of time on QA, which then led me to develop another workflow around our QA process.
SPEAKER_02So he built a QA agent, and then he refined it and refined it again.
SPEAKER_00Obviously, there's a lot of back and forth refining the agent to make sure, you know, eliminating false positives.
SPEAKER_02False positives, edge cases, unexpected outputs. This is the messy reality of building AI workflows. It's not a one-and-done process, it's continuous refinement until you get it right. And then, one day, it just works.
SPEAKER_00What really blows my mind is some days I will log on first thing in the morning. I've assigned a ticket to my team while I've been asleep, and they've already addressed it and fixed the problem before I even knew the problem existed. It just blows my mind.
SPEAKER_02When you realize the system you built is actually working. The intake form now eliminates 95% of the back and forth. The QA agent is catching issues before they go live. And the impact isn't just theoretical, it's measurable time given back to his team.
SPEAKER_00I'm estimating that we've saved about 80 hours a month just from running that one playbook for the QA agent. That's essentially half a full-time employee who we can now redeploy onto more impactful work as a result of that.
SPEAKER_02But Christian didn't stop at optimizing existing workflows. He started building entirely new ones. AI native workflows for work that almost no one was doing yet.
SPEAKER_00I'm running playbooks to check our visibility across the AI engines. So I have that running on a regular basis, updating a dashboard, and we can see how the sentiment is for various prompts, we can see citations that we're getting, and that allows me to sort of monitor how we're performing from an AI visibility standpoint.
SPEAKER_02Think about what that means. He's not just using AI to speed up existing marketing work, he's building systems to monitor and respond to a landscape that's emerging in real time. Sure, platforms like HubSpot and SEM Rush are starting to surface AI visibility data, but Christian isn't just consuming reports. He's automating the monitoring, connecting it to writers' specific context, and building playbooks that let him act in real time as new data surfaces. That's the difference between having access to data and having a system. And building those systems in uncharted territory means experimentation doesn't always work the first time.
SPEAKER_00So an example of this is around creating natural language queries.
SPEAKER_02Let's pause here for a second. If you're not deep in AI visibility, natural language queries might sound like jargon, but it's actually pretty straightforward. These are the real questions people ask AI tools like ChatGPT or Perplexity when they're looking for solutions, not keyword strings like enterprise AI platform, real questions like how do marketing teams at Fortune 500 companies scale personalized content without adding headcount? Or what AI platforms help enterprises govern brand compliance across agents? Christian needed to simulate how real decision makers actually talk to AI assistants. Because if writer isn't showing up in those responses, if the platform isn't being cited when people ask about problems writer solves, then all that organic visibility is going to competitors.
SPEAKER_00I had this really advanced playbook that I put together and I'd put all these very specific instructions about how to formulate these queries. And I just kept getting stuck on why do they sound robotic when they're coming out the other side? It just doesn't make any sense. I've given all these very detailed instructions.
SPEAKER_02The queries weren't passing the human test. They didn't sound like how someone would actually ask an AI for help. And if they don't sound real, the monitoring data isn't useful. The fix was surprisingly simple: a single line of instruction to use the LLM instead of a script. But that's what pioneering looks like. You're not following best practices, you're discovering them. And that shift from optimizing what already exists to inventing what doesn't, that's where the real transformation happens. So Christian built the monitoring. But monitoring alone doesn't improve visibility. You have to act on it. And that's where the work gets strategic.
SPEAKER_00Your website's your primary vehicle for conveying everything you want to say about your business, about what it offers, what features you have. You know, if your website isn't fresh, if it isn't up to date, if it doesn't contain all of that information, then nobody else is going to represent that for you out there on the web. It's your job to do that with your website. So I've been thinking about how do we accelerate refreshing our website, getting the most up-to-date information on there that can feed the LLMs to help us with that new sort of funnel that's out there.
SPEAKER_02Think about what he's saying here. Your website is the source of truth. If it's stale, if it doesn't reflect your current positioning or capabilities, LLMs can't represent you accurately. They're synthesizing from what's available. So Christian's focus became how do we keep our website fresh enough to feed AI engines the right information? But here's the challenge for CMOs. How do you prove this is working? Traditional attribution is broken, but you still need some signal.
SPEAKER_00And then coupling that with the data that we're getting from the demo form, for example, and we can start to build a picture as to how the program's starting to perform.
SPEAKER_02Christian added a how did you find us question back to writer's demo form. He added options like LLMs, Slack channels, WhatsApp groups, the new discovery paths. It's not perfect attribution, but it's directional signal.
SPEAKER_00I think we just have to sort of correlate visibility increases from the monitoring that we do with direct traffic increases as well, just as maybe a proxy to what's happening there.
SPEAKER_02That's the pragmatic approach for CMOs right now. You're looking for correlation, not causation. Visibility going up, direct traffic going up, demo form showing more LLM sourced leads, that's your signal. It's fuzzy, but it's enough to show momentum. When I debriefed our boss Diego after the interview, he immediately picked up on this pattern.
SPEAKER_01When I asked him, you know, what are some of the things you're excited about that you've done? He went to internal workflows before he went to visibility workflows, right? He talked about intake, he talked about QA, he talked about things that made his team function better thanks to AI. And I think there's going to be a healthy balance of those two categories of transformation that you're going to find.
SPEAKER_02That's the strategic insight from a CMO perspective. Christian didn't start with external transformation. He started with his team. He made them function better first. And that internal proof became the foundation for everything else. But for practitioners, the question remains: how do you get buy-in to actually build these systems? How do you convince leadership to let you experiment when the ROI isn't immediately obvious? Christian's advice here is crucial because it acknowledges the reality of corporate dynamics. You can't just ask for blind trust. You have to prove it.
SPEAKER_00And I think to leadership, I think you've got to get something, you know, you've got to get a win under your belt in terms of showcasing what's possible here. Because we talk about AI a lot, and you know, everybody knows AI can transform processes and workflows, but I think you have to see it with your own eyes to really believe that and to see what potential there is and spark the imagination.
SPEAKER_02Get a win under your belt. Start small, find the friction, automate it, and show the results. But even with those wins, there is a larger, more uncomfortable truth we have to accept. The old ways of measuring success, the clear, linear attribution models we relied on, are fading.
SPEAKER_00I do think, you know, as as we look forward, I think everybody's just gonna have to start getting more and more comfortable with attribution getting cloudier, unfortunately. I just think that's the nature of where we're heading.
SPEAKER_02Attribution is getting cloudier. That is terrifying for a marketer, but it's also liberating. It forces us to return to the fundamentals. And sometimes it even challenges the fundamentals themselves.
SPEAKER_00This might be controversial, but um, the advice previously was you know, right for humans, not for algorithms. I swore by that for decades. But you know, I think now it's right for humans with taking consideration to AI crawlers as well, because there is a slight way you have to adapt what you're writing now to make sure it resonates with AI search.
SPEAKER_02That's a subtle but important shift. It's not about gaming the system, it's about recognizing that there's another party at the table now. Humans will read your content, but AI will consume it too, synthesize it, and use it to answer questions. Here's the key difference. Traditional web crawlers looked for keywords and metadata. They scanned for signals. LLMs actually read your content. They understand context, extract relationships, and synthesize meaning. That changes what optimizing even means. It's not about stuffing keywords or manipulating page structure. It's about writing clearly, structuring your arguments logically, and making your expertise genuinely understandable, not just to humans, but to the systems that will represent your brand in AI-generated responses. Christian shows us that history doesn't quite repeat itself, but it does rhyme. The shift to AI visibility is just as profound as the early days of SEO. But the lesson isn't to learn a new way to game the system. The lesson is about how you lead through the fuzziness. When I talked with Diego after the interview, he framed it this way.
SPEAKER_01As a CMO, I'm on the hook for overall performance, and you need to measure me on that as opposed to looking at my channels and telling me one channel is better than the other, so go do that. The channels don't tell the whole story anymore.
SPEAKER_02That's the strategic shift. Don't let imperfect channel attribution paralyze you. Reset expectations with your stakeholders. Focus on the overall numbers, and realize that transforming your own internal workflows is the key to company-wide change. So, what defines your value when AI changes what you used to do? Your value is your ability to adapt, your willingness to look at the messy reality of your own workflows and say, there has to be a better way. Here's your Monday morning action. Look at the work happening in your organization right now. Identify one high-impact, highly repetitive task that is slowing your team down. Don't try to boil the ocean, just find that one point of friction and experiment with an AI workflow to solve it. Get that one win under your belt. Because the future belongs to the experimenters. I'm Alora Weaver. Thanks to Christian Westcott for sharing his story. Thanks for listening to Humans of AI.
