Bolting On AI Isn't a Strategy

Full transcript

Angus Robertson 00:05
Welcome to Outcome Marketing Live. I'm your host, Angus Robertson, and with me is Roark Pollock, a fractional CMO in the Outcome Marketing network. Today our topic is that you cannot bolt AI onto your go-to-market, and Roark has practical experience working through this with his clients. Welcome, Roark. How are things in Austin these days?

Roark Pollock 00:30
Thanks for having me, Angus. Things are great. I wish we could start to see a little bit of fall weather.

Angus Robertson 00:35
That just happened for us in LA. It was over 90 last week, and then it plunged. We're seeing the 60s during the day, which is unusual for us. I like the cooler weather, but it was a bit of a shock.

Roark Pollock 00:53
I can imagine. It's supposed to cool off here this weekend. I'm looking forward to it.

Angus Robertson 00:58
Roark, you have a lot of experience running marketing in software companies, and with AI. You have a technical background as well, so you've been building with AI yourself. Our conversation today is about software companies and their go-to-market. You have to have AI in there. If you don't, you're pretty much irrelevant. But there are significant differences in the type of AI that's valuable. What are you seeing?

Roark Pollock 01:36
There's a lot to unpack there, but I see companies treating AI as a messaging problem, and a lot of times it's a business problem. The first instinct is to add AI to the headline somewhere. I understand that. But to me, that's table stakes now. Everybody's saying it. As a proof point, walk through the expo at any cybersecurity conference and you're going to see AI plastered all over every booth. It doesn't set you apart anymore.

Roark Pollock 02:17
The question I've been pushing clients on is: what change are you seeing from your AI in how you build, how you deliver or how you sell? If the answer is not much has changed, then no headline you put out there is going to fix that. If the answer is a lot has changed, then the story you need to tell is about that change, not about the technology.

Roark Pollock 02:59
And it has to be a single story. I've run across clients recently with one story for investors, another for prospects, and then they talk about it differently internally. The real messaging and positioning work is drilling down to a single narrative that works for everybody.

Angus Robertson 03:25
It's challenging because AI is moving so quickly. Six months or a year ago, you needed AI somewhere on your website. Now everybody has AI on their website, and it feels meaningless. So how do you describe what you're doing with AI that provides real benefit to customers? Does it come down to people, process and technology, or do you look at it another way?

Roark Pollock 04:06
I simplify it a little more than that. Whether it's investors or your audience, they want to know how you're different. That's where your moat exists. It's not the AI. If you use AI to build it, your competitors can do the exact same thing. AI can't be your moat any longer, because your competitors have access to the same tools, and software is getting hard to defend as a moat when anybody can take it and run with it.

Roark Pollock 04:49
So it has to come from somewhere else: your distribution, real relationships in the community you serve, access to proprietary data. Whatever it is, your AI story should point at that, not at the tools. Once you establish that, your AI story stops being a generic tool. Your differentiation becomes: we've built context for our AI that nobody else can replicate. We've put guardrails on our AI that nobody else would think of. We've built ongoing evaluations into our AI, because we know what good looks like better than anyone coming from outside the market.

Angus Robertson 06:03
You're picking up on something powerful but hard to shift to. Distribution, brand, data no one else has, or specific processes and workflows you can do better than others, with the right templates and guardrails. Is it that sort of thing?

Roark Pollock 06:26
Coding at this point has become cheap. I'm doing this for myself, honestly. I've been diving into AI, and I have a side project I've been building to see what I can do with AI outside of client work, using AI to code and build software. The coding has become cheap. It's the skill to write the specifications for your software that becomes the point of differentiation. That's what I meant by having expertise in the community you serve.

Roark Pollock 07:20
I was describing this to a friend who has his own business here in Austin. I told him what I'm building and what it looks like. He thought for a moment and said he could use something almost exactly like that in his market. He's in the home remodeling business. He said, I could use almost that exact same software for a different purpose in my market. I said, that's great. I could do the coding because I'm using AI. But I have no idea what that market looks like. His expertise in that market would be the moat. You've got to have the expertise, and it has to be reflected in the way you implement AI and the specs you use to build software.

Angus Robertson 08:27
I'm a fan of services companies that turn into software companies, because they have the subject matter expertise. They understand the market: the people they serve, the problems those people have, and how best to solve them with technology. You and I have talked about how, when marketplaces like Amazon took hold, long-tail products and long-tail keywords became something that could work. You could have a very specific, niche business.

Angus Robertson 09:18
With AI, the barrier to entry has dropped for people with subject matter expertise, a bit of curiosity and some capability with systems. They can build impressive software. What I'll say is it's easy to get to a proof of concept. Maintaining it over time, with the support, the security and the user experience, is not as easy as people make it sound. Any thoughts on services companies turning their knowledge into software?

Roark Pollock 10:02
I've been seeing this firsthand with clients. Service companies are figuring out how to build with AI internally to better serve their customers, and they're turning themselves into product companies, which comes with all the needs around that. Now I need a roadmap. I need a release cycle. I need launch plans and release notes. And how am I going to talk about these tools externally?

Roark Pollock 10:42
Here's the advice I've been giving the services companies I talk to. They're building an AI tool over here, and another AI tool over there for a different need, and their instinct is to launch this tool and then launch that tool. I think it's fantastic that they're taking advantage of the tooling and the infrastructure they're building. But my advice has been: don't launch a tool, or a series of tools. Launch your operating system instead. You don't want it to look like a Claude Code agent running here and another one running there. You want one operating system or platform that says, we've rethought the whole business and how we deliver. That signals to customers, and just as importantly to investors, that you've made a real investment in your infrastructure, not some add-on tools.

Angus Robertson 12:06
That makes sense. Of course you can code faster and do things better with AI, and AI is spreading to every discipline. It's not only coding. There's a bit of a crisis in the math community, there's been one in software development, and creatives in entertainment have the same concern with video and movies. Beyond making things faster and better, I think you can connect things you couldn't connect before, across functions or across process steps. In marketing, you can connect strategy with websites with campaigns. In software, you can connect the specification with the build, with QA, with how customers use the product. Those used to be hard to connect. Are you seeing the same thing?

Roark Pollock 13:21
Absolutely. A lot of the a-ha moments for the clients I work with are about bringing disparate information silos together, and that opens up a new way to serve their customers. But you hit on something, Angus. At the end of the day, you've got to know what problem you're solving with AI. The tendency is, I've got this great tool, let's jump in and start doing stuff. It's really about what problem you're addressing for the customer, and whether you understand the customer's needs well enough. Then, how are you going to take it to market? The go-to-market motion can vary completely depending on what type of product you're developing.

Angus Robertson 14:34
You talked earlier about consistency in messaging across channels and audiences. Do you have approaches or examples where that has worked with AI?

Roark Pollock 14:55
One of the things we've done is build AI tools that help sales teams surface the right messaging, or review messaging they're developing on their own. Anything we build from an AI perspective sits on a foundation of our golden messaging, plus all the approved deliverables that sales or customer support rely on, so they can find or generate content that fits what we want to say to customers very quickly, and it comes out of the tools the way we want it to.

Angus Robertson 15:43
That's cool. When go-to-market teams, whether sales, customer success or marketing, have their AI reference the same golden set of messaging, it solves something marketers and product marketers have struggled with for years. The people we're trying to help don't always have the time or the interest to read all the content. That's the benefit of AI. AI agents and tools will read everything, every time.

Roark Pollock 16:29
Right. It can be your brand police or your messaging police for you.

Angus Robertson 16:34
Something else we talked about that I thought was interesting: what kind of hunter are you? A gatherer, a rabbit hunter or an elephant hunter? What do you mean by that, and how does it relate to go-to-market and AI?

Roark Pollock 16:56
I'll be honest, I stole that from a book I read a while back called T2D3. You're probably familiar with the model. If you're selling a very low-cost product and you need to close 10,000 of them to hit your revenue number, your go-to-market is going to be very different than if your ASP is over $100,000 a year and you only need to close 50 deals. So are you gathering, picking nuts off the ground? Are you farming or rabbit hunting, after small game? Or are you a big game hunter, after deer or elephants? Those are the three camps, and your go-to-market changes with each. You need to pick one to start and focus on it, because you're not going to gather nuts with a bow and arrow, and you probably want more than a bow and arrow if you're hunting elephants.

Roark Pollock 18:26
If you're a gatherer, with small deals, you want a PLG motion that's very low-touch, and your cost of acquisition is much lower. If you're farming, with mid-sized deals in the mid-market and a higher ARR, maybe it's an inside sales and marketing motion. If you're serving big enterprise with big ARR deals, it's built on an enterprise field sales team, and your CAC can be higher.

Roark Pollock 19:08
The big mistake: AI makes the first two motions a lot easier and cheaper to run. So there's a temptation to say, we're hunting elephants, but we could also do PLG, and we could also do inside sales. Pick one, stick with it, and prove it works. A hybrid might be fine. A lot of the clients I've been working with lately have a land-and-expand motion, and that works. You get in with a smaller ARR, then expand within those clients. That's a great motion, but it doesn't mean you're picking multiple motions on the front end.

Angus Robertson 19:53
I agree with you. Another way I've heard it, from a CRO I used to work with, is minnows, fish and whales. Companies are generally okay at products and markets, though they can get better. But the motions for those products and markets are equally important, and there are two obvious pairs. Direct or indirect and digital. And land versus expand. For a lot of companies, getting clarity on land and expand first is valuable, because existing customers who are happy with you have more potential to grow with you. Then: are you going after smaller deals in a large market, or bigger deals in more targeted markets?

Angus Robertson 21:20
Understanding those dynamics goes to your CAC, how you structure your sales team and your marketing, and what your messaging and campaigns look like. Too often executives get frustrated with sales and marketing because they're not moving fast enough. But if you're trying to land and expand, do digital and channel, and chase big deals and small deals all at once, you're setting your sales and marketing team up to fail. Making sure that alignment is there at the management level, the board, the CEO, the CFO, is so important.

Roark Pollock 22:07
100%. You need to make your strategy clear, and the metrics need to align to it. And we haven't even opened the can of worms that is channel partners. That's a whole other conversation.

Angus Robertson 22:25
I think it goes back to the AI moats, too. If you have really good channel, ecosystem or partnership relationships, on the technology side and the distribution side, that can give you a significant leg up.

Roark Pollock 22:40
Yes, it can. Absolutely.

Angus Robertson 22:45
I brought up that you have a technical background. How have you been looking at AI? Are you adding it where it makes sense, or diving in? Has it changed how you think and how you work?

Roark Pollock 23:20
It's 100% changed how I work, and this is where I'm trying to eat my own dog food. If I'm telling people not to build individual tools and to think about their own operating system, that's exactly what I've been doing for myself. I've built my own digital twin, I guess you would call it, built on my experience, my background, the frameworks I've developed or curated over the years, and a number of deep research projects on the buyers in the markets I typically serve. I use that as the foundation for my own operating system, with a variety of other agents for different purposes tied into it.

Roark Pollock 24:16
Long term, especially as a small business or a solo entrepreneur, the question is how you start building your own agent workforce. Your marketing department does one thing, your record-keeping department another. Everything you do with AI these days is about the context you provide and the guardrails you put in place. So how are you building that structure of knowledge? A lot of that is the infrastructure. So yes, I'm building that infrastructure and that operating system for myself.

Angus Robertson 25:03
That future state is hard, because we're going to be working with humans and agents for a long time now. What does the human do? What does the agent do? How do you mix them across problems, areas of expertise and functions, and where do they each come into a workflow? I've got about 15 agents helping me now, and I'm the limitation. I'm absolutely the limitation. It's crazy how smart these agents are, and we're at the tip of the iceberg. More is coming. I haven't figured out how to balance that yet.

Roark Pollock 26:03
I like the idea of thinking about the org chart. The org chart is built on the work that needs to get done. If agents are doing a certain task or series of tasks, where do they fit on the org chart? You might have boxes that are agents, or workflows, and then people within the same org chart. That combination works because it's all built around the work that needs to get done for the business. That's the way I think about it.

Angus Robertson 26:51
I think you're right, and I don't think we've got it figured out yet. It's going to be interesting to see what the org charts of the future look like, and what role we as humans have in them. Well, this has been great. I love talking about AI and go-to-market, and I appreciate your insight and experience. We even got to talk about acorns and elephants. Anything you want to leave our audience with on messaging, AI and go-to-market? And how can people reach you?

Roark Pollock 27:33
Thanks, Angus. If I were to boil it down to one thing: AI can't be your moat, so your story can't be about AI. In practice that means a few things. Develop one story that rings true for all your audiences. Launch the operating system you're building, not a single tool or a series of tools. And think hard about the specifications you're writing for your AI infrastructure, because that is the differentiation you bring. The AI tool is helping you execute.

Roark Pollock 28:18
People can find me at p1growthmarketing.com, or I'm easy to find on LinkedIn under Roark Pollock. I love talking to founders and marketing leaders who are rewriting their story for AI, or trying to figure out what kind of hunter they are. If that's you, send me a note on LinkedIn. I'd love to chat. Thanks, Angus.

Angus Robertson 28:49
Thank you, Roark, and thanks everybody for joining. See you on the next Outcome Marketing Live. Take care.