Zero to Campaign With Everyday AI, in Four Steps

GTM
Written By Christina Ellwood

Episode Summary

Doug Bell spent twenty years as a CMO, CPO and CRO in SaaS before a short retirement, and came back in late 2022. What he found was that the constraint was not really his own capacity, it was the playbook, and the same three problems he had faced for years: bad lists, bad messaging, and speed. The method he built with Jordan Crawford, called Cannonball after the coast-to-coast race, is his answer to all three at once, and it runs in four steps. Focus means understanding the market rather than trusting the ICP you were handed, and finding what he calls the existential data point, the metric that decides whether a company there lives or dies. Investigate turns that into scored segments. Narrate writes messages he says a prospect would pay to receive. Deploy gets it into market on Clay. The research and messaging tooling is deliberately ordinary: ChatGPT, Claude, and dictation rather than typing.

Key takeaways

  • Three problems, one method. Bad lists, bad messaging and not moving fast enough are the three challenges he says he faced for the longest time, and the point of the method is that it deals with all three together rather than one at a time

  • Four steps: focus, investigate, narrate, deploy. Focus is understanding the marketplace, investigate scores the segments, narrate writes the messaging, deploy gets it into market, and he says the whole thing can run in an hour on everyday AI

  • Do not trust the ICP you were handed. His starting move is to find the existential data point instead, the metric that decides whether a company in that market is about to live or die, and to build segments from that

  • Write messages a prospect would pay to receive. He calls them permissionless value propositions, and the test is that the value comes from the email itself. Sometimes the brand is not mentioned in it at all

  • Most messaging is about the sender. His rendering of the category runs I am great, it is fantastic, buy my stuff, you have money, I have money, I like money, give me your money, and the method exists to write the opposite of that

  • Two models, two jobs. ChatGPT for research and deep research, which is the focus and investigate half, and Claude for analysis, thought partnership and getting to the messaging. He is emphatic that it is always Claude for that second job

  • He has moved past prompt engineering. Partly because Claude is a strong enough thought partner, and partly because he stopped typing and started dictating. His approach is to say what you want, how it should act, what result you expect, and then ask it whether you are thinking correctly

  • The person to give this to is in RevOps. His answer is to start with the people already struggling with the BI dashboards and creating the reports: someone obsessed with data because of what it teaches about the market, who can combine market-reactive data with pain segmentation and orchestrate between the two, and who is not siloed away from the leadership meeting

  • At smaller companies it is the programmatic marketer. At companies under 100 million he points to the digital marketer, the one used to a mashup of first and third party data who is on with the Google rep every week and good at managing obscure data sets to results. When you offer product-led growth marketers he agrees they are gold

  • The method is vertical, and he says so. Cannonball is built for vertical SaaS and he calls it horrible for horizontal SaaS, pointing horizontal companies at The Deal Lab or Eric Nowoslawski instead, and adding that building your own is fine because this is not the world of IP

About Doug Bell

Doug Bell is a fractional CMO with Chief Outsiders and co-founder of Cannonball GTM, the go-to-market methodology and Substack he built with Jordan Crawford. He describes roughly twenty years as an in-house CMO, CPO and CRO in SaaS before a short retirement, and a numbers background before that doing M&A work at GE Capital and at Cisco, which he says left him approaching things from both a business structural standpoint and a data standpoint. He came back to the work in late 2022 expecting the technology to be at least as impactful as he had thought, took on more clients than he could serve, and built the playbook out of that pressure. Cannonball runs in four steps, focus, investigate, narrate and deploy, and he is unusually direct about its limits: it is built for vertical SaaS, he calls it horrible for horizontal SaaS, and he names two other practitioners he would send a horizontal company to instead.

 

In this episode

00:42 Welcome and guest introduction
01:07 Chief Outsiders, and co-hosting Cannonball GTM
02:13 Twenty years as a CMO, CPO and CRO in SaaS
02:44 Late 2022, and what brought him back
03:30 Not a scale problem, a playbook problem
03:53 Bad lists, bad messaging, and speed
04:20 Finding people in pain the brand can solve for
06:47 Eight hours, and the first Cannonball
06:56 Shovels, and a campaign in four and a half hours
09:45 What clients and boards are actually doing
10:05 A campaign is seven or eight or nine or ten things
10:44 The verticals: construction, logistics, transportation, financial, government
11:20 Selling labor into construction firms that cannot hire fast enough
11:54 What Cannonball actually refers to
12:24 Four steps: focus, investigate, narrate, deploy
12:44 Not believing your ICP, and the existential data point
13:02 Equipment on lots, and three months from a cash flow problem
13:22 The pivot point for a whole market, and scoring segments
13:39 Narrate, and permissionless value propositions
13:55 Messages a prospect would pay to receive
14:10 Deploy, and building the Clay tables on the spot
14:43 Which generative AI tools he actually uses
14:57 Everyday AI, and not knowing where to start
15:12 ChatGPT for research, focus and investigate
15:32 Claude for analysis, thought partnership and messaging
15:49 Past prompt engineering, and dictating instead of typing
16:25 Research, chat. Thought partnership and creation, Claude
16:32 Any other tools
16:58 The context window complaint
17:28 Who in marketing benefits most
17:53 Start with RevOps
18:13 Someone obsessed with data, for what it teaches
18:53 Pain segmentation, the second
19:13 Orchestration, and why they cannot be siloed
19:29 Vertical SaaS, horizontal SaaS, and the limits of the method
19:47 Building your own, because this is not the world of IP
20:07 Who does this at companies under 100 million
20:11 The digital marketer, and first and third party data
20:51 Product-led growth marketers are gold
21:08 What the marketing organization looks like next
21:43 A John Miller post on fractional CMOs ruining the market
22:11 Why brand building takes a back seat for a little while
22:33 First and third party data, not intent data
23:16 Non-siloed organizations built on a data org
23:58 Systems of action, and titles disappearing
24:30 Judgment, and what the C-suite is for
25:30 Competitors adopting everyday AI on a low burn rate
25:47 Your moat is your brand, and your data
26:19 Resources for listeners
26:27 Cannonball GTM, and who it is for today
26:37 Go-to-market engineers, and a version for founders
26:55 Kellen Casebeer and The Deal Lab
27:11 Eric Nowoslawski on rapid message iteration
27:51 Outbound Kitchen, and rethinking outbound
28:04 The roundtable and the summit
28:53 Judgment times ten
29:28 Not about doing, about listening
29:33 Wrap-up

In Doug’s words

“It wasn’t about my ability to scale personally, it was definitely helpful, but it was about my ability to build a new playbook.”

— Doug Bell   (03:30)

“It is going from zero to a campaign. We can do it in an hour using everyday AI. There are four steps: focus, investigate, narrate, deploy.”

— Doug Bell   (12:24)

“They’re so good that a prospect would pay to receive the message, meaning the value is derived from the email.”

— Doug Bell   (13:55)

“So research, chat. Thought partnership and creation, Claude.”

— Doug Bell   (16:25)

“You want somebody who is obsessed with data, not for the sake of data, but because the data teaches you something about your market and how your market is reacting to your message.”

— Doug Bell   (18:13)

“This is not the world of IP. This is the world of test and experimentation.”

— Doug Bell   (19:47)

“It’s not about doing, it’s about listening.”

— Doug Bell   (29:28)


 

Resources

Doug Bell, Cannonball GTM and Chief Outsiders

  • Doug Bell on LinkedIn: linkedin.com/in/dougbell1

  • Cannonball GTM: The go-to-market methodology and Substack he co-founded with Jordan Crawford, with a live stream. He says it is aimed at go-to-market engineers and advanced SDRs today, with a version for founders planned

  • Chief Outsiders: chiefoutsiders.com. The fractional executive firm he works through

The Cannonball method

  • Focus: Understanding the marketplace rather than accepting the ICP as given, and finding the existential data point for that market

  • Investigate: Turning that into segments and scoring them, so there is an ordered view of what is worth going after and what is not

  • Narrate: Writing the messaging for the resulting list, in the form he calls permissionless value propositions

  • Deploy: Getting a campaign into market, usually outbound first, building the Clay tables on the spot

  • The existential data point: His term, and he says he made it up. The metric on which a company in that market lives or dies. His worked example is construction leasing, where equipment sitting on lots rather than deployed in the field predicts a cash flow problem within months

  • Permissionless value propositions: Messages good enough that a prospect would pay to receive them, where the value is in the email itself. He says they sometimes do not let the brand mention itself at all

Tools he names

  • ChatGPT: chatgpt.com. His research tool, for the focus and investigate half. He says he reaches for 4o, is not a huge fan of 5, and adds that the more they use 5 the better it gets

  • Claude: claude.ai. His tool for analysis, thought partnership and messaging, and he is emphatic that for that job it is always Claude

  • Superwhisper: Dictation rather than typing, which he credits alongside Claude for moving him past prompt engineering. He gives no URL for it

  • Clay: clay.com. Named for building the tables during the deploy step

  • Gemini: Discussed rather than recommended. He calls Gemini 2.5 Advance a good model and says Google could have won on context window size, then had the biggest of anybody and shrank it, which leaves it in his view a bad combination of ChatGPT and Claude that is trying to be both. He expects Google to fix it

People and publications he names

  • Jordan Crawford: His partner, and co-founder of the method

  • Kellen Casebeer, The Deal Lab: deallab.com. He calls it brilliant for horizontal SaaS, which is where he says his own method is weakest

  • Eric Nowoslawski: Named for a methodology built on rapid message iteration turning into fast campaigns, and also recommended for horizontal SaaS

  • Ken Norton: Named in the same breath as the other two as someone he likes what they are talking about

  • Outbound Kitchen: The Substack he recommends for anyone ramping or rethinking outbound

  • John Miller: Named, not recommended. Doug cites a post of his arguing that fractional CMOs are ruining the market and taking a long-term view on brand, says he has a lot of respect for him, and then says he is wrong and spends the next several minutes explaining why

Related AI Realized episodes and events

 

Frequently Asked Questions

  • You build a marketing campaign with AI in four steps, which Doug Bell of Chief Outsiders calls Cannonball. The steps are focus on understanding the market, investigate to build and score segments, narrate to write the messaging, and deploy to get it into market. He says the sequence can run in an hour using what he calls everyday AI, and the reason it holds together is that it attacks his three long-standing problems at once, which are bad lists, bad messaging and not moving fast enough.

  • An existential data point is the metric that decides whether a company in a given market lives or dies, and it is the thing Doug Bell of Chief Outsiders looks for instead of accepting an ICP as given. His worked example is construction leasing: large fleets of equipment sitting on lots rather than deployed in the field, where falling below a threshold predicts a cash flow problem within about three months and going out of business within six. Once that pivot point is known for a market, segments can be built from it and scored.

  • A good cold outbound message is one the recipient would pay to receive. Doug Bell of Chief Outsiders calls these permissionless value propositions, and the test is that the value comes from the email itself rather than from what the sender is offering. He contrasts it with most messaging, which he characterizes as the sender talking about themselves, and says they sometimes do not let the brand mention itself in the message at all.

  • The best AI tools for marketing are different ones for different jobs, rather than one tool for everything. Doug Bell of Chief Outsiders uses ChatGPT for research and deep research, which maps to the focus and investigate steps, and Claude for analysis, thought partnership and getting to the messaging, where he says it is always Claude. He notes he reaches for 4o over 5 while acknowledging 5 is improving, and he names Superwhisper for dictating rather than typing.

  • Doug Bell of Chief Outsiders says he has evolved past prompt engineering, for two reasons that work together. He credits having a strong enough thought partner in the model, and credits stopping typing in favor of dictation. What replaced the prompt is closer to a briefing: tell it what you are looking for, how you want it to act, and what result you expect, then ask it whether you are thinking correctly and ask it explicitly to act as a thought partner.

  • At a company above 100 million in ARR, this starts in RevOps, with the people already struggling with the BI dashboards and creating the reports. Doug Bell of Chief Outsiders wants a person obsessed with data, not for its own sake but for what it teaches about how the market reacts to the message, and describes the job as combining three things: market-reactive data, pain segmentation showing which accounts are hurting and when, and the orchestration between the two. He adds that this person cannot be siloed and should be in the leadership meeting.

  • At a smaller company, this work falls to the digital or programmatic marketer. Doug Bell of Chief Outsiders describes the person at companies under 100 million who is used to the mashup of first and third party data, is on a call with the Google rep every week, comes in after the weekend having found a new tool, and is good at managing obscure data sets to results. When product-led growth marketers are put to him as the label, he agrees they are gold.

  • No: the approach does not fit every kind of software company. Doug Bell of Chief Outsiders says the method he founded with Jordan Crawford is very good for vertical SaaS and horrible for horizontal SaaS, and sends horizontal companies to The Deal Lab or to Eric Nowoslawski instead. He also says building your own methodology is legitimate, on the grounds that this is not the world of IP but the world of test and experimentation.

 

  • [00:42] Christina Ellwood: Welcome to AI Realized, the podcast for enterprise executives leading AI deployments. From tackling security, data, and operational challenges to navigating organizational transformation, AI deployment offers a unique opportunity to redesign organizations from the inside out. I'm Christina Ellwood, your host for today's episode, and we're talking today to Doug Bell. He's part of the Chief Outsiders team, and he's also the podcast co-host of Cannonball GTM. Welcome, Doug.

    [01:17] Doug Bell: Thanks. It's good to be here.

    [01:18] Christina Ellwood: It's nice to have you here. I am very excited by the work that you and your compatriot on Cannonball are doing to use generative AI in doing sophisticated marketing work, strategy work, segmentation, and so forth. So I'm excited to have you tell our listeners how you are applying AI in that world, and what advice you m- have for them in deploying such technology into their workflows. So let's start with, I know that you're a fractional CMO, so you are in the seat. You know what it is to be running a marketing organization, working with a sales team, trying to keep that funnel full and moving and, and squeezing more and more value out of the resources that the sales and marketing team have. So I know you speak from that lens. So tell me what inspired you to go so deep in the AI world, and what you've learned so far.

    [02:13] Doug Bell: Okay. So to do so I've gotta take you back to fall of 2022. And at that point in time I had spent probably 20 years as a CMO, CPO, and CRO in SaaS, and was, I thought, at the end of my journey. Prior to that, Christina, I was always a numbers person. I was doing M&A work for GE Capital and then for Cisco for a while. So I always approached things from both a business structural standpoint and a data standpoint. And I really thought my sort of ability to help was over in the fall of 2022. That was October. November, I believe, 9th, 2022, ChatGPT 1 came out, and that brought me back into the workforce, by and large because I felt like it was gonna be as impactful, maybe even more so, than I expected at that point in time. And what that meant for me was getting back in as a fractional CMO for SaaS yet again. And what really got me in the weeds was that I forgot about the AI thing for a little bit. I was just excited to, to be back and helping companies grow faster. And what I found was that, like a dumb, I added too many clients too fast, and I was having trouble scaling. So my friend AI came to the rescue, and what I realized really quickly was that although it, in this case, it was ChatGPT plus Claude, Gemini came later. What I realized pretty quickly was that it wasn't about my ability to scale personally, it was definitely helpful, but it was about my ability to build a new playbook, if I could be so bold. And that new playbook was based on AI's ability to get to better information faster. And Christina, I have to pause and just say, what good is that? To answer my own question, and that is to say this. My experience was, for the longest time, facing these three toy- these three challenges, and they were bad lists, in other words, who to target, bad messaging, I'd written a lot of it, and moving at the speed I wanted to move. And that's where AI came in. It allowed me, and therefore my accounts, to conquer those three challenges. How could I find people experiencing pain that the brand I was representing could solve for? One, that's the list. Two was how to create a message that's not about the brand. All the messaging right now, with the exception of what we'll talk about today, is me. "I'm great. It's fantastic. Buy my stuff. You have money. I have money. I like money. Give me your money." That's most of messaging today. And the third thing I think is equally as important was moving with speed, and all that means is, okay, once I have those two first things, how do I test that and get it into market quickly? So that's how I got so deep in the weeds, because I went from concept to, "Oh my God, I just goofed. I have too many clients. How do I deliver better value?" And then really I realized, guess what? I get to deal with each of these three problems in kind, and then have a solution that deals with all three things collectively.

    [05:05] Christina Ellwood: Now, somehow you went from having the problem, solving it yourself for yourself or your clients in your own situation, to partnering with, with Jordan to do the Cannonball. So how did that happen?

    [05:17] Doug Bell: That happened, it's funny 'cause I bumped into Jordan right before making the decision to retire for six weeks, and he was a guest on a podcast I had at that time. And he just- he's a brilliant guy. Hopefully people listening to this podcast go check him out. He's just a brilliant guy, and he and I instantly bonded, and we would get together and, pardon the word, we would bitch, we would kvetch, we would complain. And in every case, that sort of complaining ultimately was about how we were struggling to help clients at speed. And we were both in parallel working on different aspects of everyday AI, helping clients. Jordan is a go-to-market engineer, growth hacker. He could have been a neurosurgeon. He could have been a rocket scientist. He's a data guy. That's what he does. And so he was solving the problem for the first thing I talked about, which was bad lists. He was solving for that problem, and I was solving for the bad messaging piece. And that sort of came together fall of 2024 when Jordan called me on a Friday night, 'cause Jordan, I think, is a boomer. He's actually a millennial, but he's a boomer, and he just calls people. He calls me on a Friday night. I'm like, "Oh, God." And I pick up the phone, and he goes, "Look, man, you're the only guy I know that can host a live podcast live stream who understands the nuances of AI. I'm doing this thing called a Cannonball. Remember the movie Cannonball With Bum Bum Bum?" "Yeah, I got it." "I'm doing it tomorrow, and I'd like to you to come over to my house and host it." I'm like, "Dude, Saturday? Really? When I-- What am I gonna-- I got family, I got kids, I got things." He goes, "I had 800 people signed up."

    [06:46] Christina Ellwood: Whoops

    [06:47] Doug Bell: Okay, I'll be there. And so what we did in those eight hours, Christina, is we, we did the first Cannonball. We didn't realize we were doing the first Cannonball. And what we did is we went from not knowing a brand at all, at that time it was Shovels. We went from not knowing a brand at all to shipping a campaign using clay.com, and we did it in four and a half hours. Wow. And we had 800 people signed up, 600 people showed up, and we had at all times about 300 people on the call. Fast-forward a couple months and I reach out to Jordan, I'm like, "I think there was something there. By the way, this is what I'm working on. What are you working on?" And we like to talk, and we learn through talking, and we learn through doing, and we decided to start a live stream. People are like, "Okay, where can I get your content?" I'm like, "Maybe I should do something about that." We created a Substack, and we are now top 50-- we are a top 50 business Substacks, and the live stream is growing in popularity every day. And I have to say, at the end of the day, it's still a hobby for us. We're taking that work, we're applying it to our clients, but we're still having fun. Y- if you show up for the live stream, you will see we don't take ourselves seriously, and it's about our learnings and what we're sharing in terms of what we just learned that week with our clients.

    [07:58] Christina Ellwood: It is a dream podcast because you guys are having so much fun. It's entertaining all by itself. It's a little like Twitch for, for marketing geeks, which is pretty awesome. But more importantly, I think the seriousness with which you take the task at hand and the practicality that you apply to what's the problem we're solving, how are we gonna use this to solve it, and why are we using this to solve it? You don't ever deviate from the appropriate application, and I really, uh, respect and appreciate that. So let's go back to for the, the growth hackers, the CMOs, the heads of fill in the blank on the commercialization side of the business, people who are wrangling content and wrangling communications What are some of the top lessons you've learned about the best way to use AI for the three challenges that you laid out?

    [08:52] Doug Bell: Okay. And look, I, I have to say, I've sat in the shoes of a lot of people that are listening in or watching the podcast today. I was an in-house CMO, CPO, CRO for 20 years, and I know how hard it is to pivot from theory to practice. So let me give you some just really solid anchors, and I'll start with what not to do, and then I hope we can talk about what to do and to really where to dig in, 'cause I want this to be as demonstrative as it can be. And Christina, you know that from the live stream. It's like we quite often are like, "Practice and what's the theory?" Like practice first, what's the theory later on? So here's the thing. So first thing is that, sorry to bring this news to you, but your playbook is broken and it sucks, and that's not your fault. I sat on a stage yesterday with the former CMO of Zendesk and of Slack, brilliant guy, and at one point we started talking about how the playbook is broken, and we turned to each other, we're like, "Dude, we created the playbook," right? So I say this from a place of love, and I say this from a place of somebody that helped create the playbook that you're experiencing right now. And what I think I'm seeing a lot with my clients and with my board seats and advisory seats, what I'm seeing is people are taking the existing playbook and they're amplifying it. It's how do I send more messages more quickly? This is the SDR, AI SDRs you're seeing out there, right? Or how do I make my messaging even more personalized from my bad list? And then ultimately, a campaign equals seven or eight or nine or 10 things, 'cause guess what? You're running a corporate marketing organization that has marcom all the way through digital marketing, and a campaign is this thing, it's a giant thing, and there are people that have to get involved. Okay. So that's what we're swimming upstream against. And then here we are. I've had the CMO's job. It is the hardest job. It is political, it is numbers driven, it is creative, it is analytical. It's all those things. So places to start. The first thing is can you tell yourself with deep honestly, honesty that the target list that you have needs your product right now? And if you cannot answer that, then you need to go back. And so again, I get it's, "Oh God, Sales is gonna say this and I'm gonna, ah." Let's go to the speed thing. Here's the speed thing, and especially true, Christina, I know you're true. Markets into the following verticals, I'm gonna hound you until you do this right. If you are selling to construction, logistics, transportation, financial, anything to do with government work, government sector, you can know with a little bit of work, and in this case, by the way, there are other methodologies. I'm gonna mention them. I just happen to be the author of one. And with some... Of one. And with some construction industry, all of that data is public. And I'll give you a quick example, which is to say, one of my clients right now is selling labor into construction firms that are scaling 300% or more a month, and they can't hire fast enough. Okay? That's a pr- example of a profound pain, and guess what they offer? They offer people that help scale. Bid managers. Con- There's my methodology. There's another organization called The Deal Lab, which approaches it differently. They have a different approach. It's called, they call it messaging arbitrage. They get the same result, but the point is they're gonna make sure you get to a great list first, just like we do

    [11:54] Christina Ellwood: Great. No, that's very helpful. It strikes me that we didn't explain Cannonball. So for those who did not grow up with the Cannonball Run, give them the short story of the Cannonball Run.

    [12:05] Doug Bell: Yeah. Okay, okay. It's an underrated movie from the '70s. Cannonball Run is actually a real race. It starts in New York and ends in LA. You can only use like stock cars, like cars that you would pull out of your driveway, and those-- You have to make it to California. The first person, I think it's New York to LA, first person to make it there wins, so to speak. And you're dealing with all the stuff, the cops, getting lunch, going to pee, all those things. Cannonball is the same idea. It is going from zero to a campaign. We can do it in an hour using everyday AI. There are four steps: focus, investigate, narrate, deploy. Focus, as you can imagine, fellow CMOs, it's about understanding your marketplace. And guess what? We don't always believe that your ICP is correct, right? We go, "That's nice. Here's what we're gonna do. We're gonna go understand what the tipping point is between somebody that's in business and out of business." We call that an existential data point. I made that up. Thanks to me. Good job, Doug. And that always confuses people. And let me just say existential meaning I'm gonna live or die whether or not this is, this metric represents that. An example is in construction leasing. Think about large fleets of equipment sitting on lots. If less than 60% of that equipment is not deployed in the field, you're three months away from a cash flow problem, six months away from being out of business. That's what that is, and what the Cannonball does is says, "Okay, for that market, what is that pivot point for your entire market?" And then it says, "Great, let's create segments that we can go after and score those segments." That's the investigate piece. Okay, we're gonna line these things up and say what's good and what's not. Narra-narrate is when we go, "Great. We-- To get to that investigate piece," understanding what the segments are, folks, believe it or not, this does happen in an hour, I promise you. Then we go and we say, "What's the messaging for that list?" And we call these permissionless value propositions. These are messages... Christina, they're gonna think I'm full of crud, but I'm just gonna say it. They're so good that a prospect would pay to receive the message, meaning the value is derived from the email. Sometimes we don't even let brands mention themselves in that email. And then finally deploy, which is how do we get a campaign? Typically, it's easier to get an outbound campaign going. How do we actually create a campaign? And we'll build the Clay tables on the spot and start firing off proverbial emails, 'cause guess what? We don't-- We just pick brands. We're, we're not always partnered with those brands. We're like, "This brand looks good," and we throw them into the Cannonball. That's the whole thing. That's how it works.

    [14:28] Christina Ellwood: It's a great high-pressure system for creating output quickly, and it's unambiguous what the output is and what the value of the output is, because having been through the process myself, I can tell you it's obvious when you've got the right, when you've got the right output. We all know when the response rates are good. This is not a question that we have to answer. So- I know you're using generative AI. Is there a particular model that you're finding most valuable?

    [14:57] Doug Bell: That's such a great question. Christina, you and I are in the biz. We're in the biz. We're thinking, we're-- You and I debate agentic platforms and deployment methodologies and, uh, and sometimes we forget that everybody is using everyday AI every day, and we don't know where to start sometimes, or we don't know which the best model is. Thank you for that question. So I'm gonna say the following. I'm gonna oversimplify the world. For research purposes, for generating those really amazing sort of deep research, think about the F part of the Cannonball metho- methodology. So focus and investigate. Those two, ChatGPT is lovely. Not a huge fan of 5. I'm gonna use 4o, but I would tell you I should scold myself and say the more we use 5, the better 5 gets. But I will, for important projects, start with 4. Okay. And then in terms of analysis and thought partnership and actually getting to really great messaging, I'm just gonna anchor myself again on the Cannonball, it's always Claude. It's always Claude. And we could spend a lot of time as getting into prompt and prompt engineering, and I have to say I've evolved past prompt engineering because I have such a strong thought partner in Claude, but also because I stopped typing. I've started using something called Superwhisper. And Claude will train you to use Superwhisper. I don't know how else to put it, but you just really need to start talking to Claude. And my tips here are tell it what you're looking for, tell it how you want it to act, tell it what your results you're expecting, and then ask it to tell you whether or not you're thinking correctly, and say, "Ultimately, I need you to be a thought partner." So research, chat. Thought partnership and creation, Claude.

    [16:32] Christina Ellwood: Gotcha. No other tools.

    [16:37] Doug Bell: Okay, I can talk about Gemini for a second. I just will mention that Google, it's, it's interesting to see Google struggle. I'll just say that. Gemini 2.5 Advance is a good model. They could have won by, based on the size of their context window. So context window, we've all had an experience, we're jamming with Chat, we're jamming with Gemini, sorry, we're jamming it with Claude, and it goes, "Sorry, you've run out." You're just like, "I just spent 45 minutes understanding this thing, research project, whatever." Gemini had the biggest context window of anybody, and they shrank it. And so to me, it's a bad combination of ChatGPT and Claude. It's trying to be both things. Context window, it's just as the same size as everybody else. Now, smart people at Google, I expect them to figure out that they've screwed up, and I expect that context window to get bigger in the future, but right now it's not providing an incremental value compared to the other two models.

    [17:28] Christina Ellwood: Who in the marketing organization is the most likely beneficiary of your method today?

    [17:39] Doug Bell: Okay, I'm gonna presuppose some things, and then maybe I'm gonna say some things, and then Ch- Christina, you tell me if I have this right, 'cause the org is gonna change depending on the, how you answer these questions. Are these organizations above 100 million or below? Yeah. And, and ARR? Okay, good. Thank you for that one. Um, RevOps, please start there. Start with the folks that are right now struggling with your BI dashboards and the people creating reports. They are pattern matchers and natural problem solvers, and that's where your skill sets lie. And I just, I'm gonna quickly describe what I'm talking about, Christina, 'cause again, we wanna be demonstrative, right? We're practice then theory, so let me get right to practice, and then I'll talk about theory. In practice, you want somebody who is obsessed with data, not for the sake of data, but because the data teaches you something about your market and how your market is reacting to your message. That is your core skill set, and that is your core desire with that person. I'll start there. Now let's talk about theory. What that person needs to be doing is combining three things really well. The first is they need to be very good at taking data that is market reactive, right? So we talked about before, we want data that is market reactive, meaning I'm pushing stuff out there, whether it's good or bad, I'm getting a reaction to it. But that is useless if your second data set doesn't exist, which is the pain segmentation. My accounts are experiencing this pain in this moment. That's another data set. And then the third skill set is orchestration between those two And I would also empower them. If you have a leadership meeting, they're on that meeting, and they're hearing and listening 'cause guess what? They can't be siloed. So if you think about this person, they're good at gathering data from these two places, and they're very good at orchestration reaction to that data. Cannonball can be in the middle of that. Deal Lab can be in the middle of that. There's a guy named Eric Nowoslawski who also has a methodology he's creating. My methodology, the methodology I founded with my partner Jordan Crawford, is very good for vertical SaaS. It is horrible for horizontal SaaS. It just is. The Deal Lab, brilliant for horizontal SaaS. Eric Nowoslawski methodology, very good for horizontal SaaS. So if you're a horizontal SaaS or horizontal tech company, those are your guys. If you're a vertical SaaS, you're looking at a methodology like mine. And by the way, I do believe you can come up with your own methodology. This is not the world of IP. This is the world of test and experimentation. But if you're looking for the person that's gonna do this for you, that data analyst in your RevOps org, they love data, and they're gonna manage those three, three points on that triangle, if you will.

    [20:07] Christina Ellwood: Who do you see in the smaller companies under 100 million?

    [20:11] Doug Bell: Yeah, it's the-- So that's such a great question. It is, Christina, my experience, it is the, it is the guy that is the digital marketer or the girl that's the digital marketer, and they're so used to dealing with the mashup of first and third party data. They're just such a natural fit for this. They're the person that's on with your Google rep once a week. They're the person that, like, comes home from, sorry, comes to work from the weekend and they go, "Look at this tool I discovered." You know that person. We've all worked with that person, right? And so they're really good at managing that obscu- those obscure data sets to results. So let's call these programmatic marketers, if you will. Sometimes digital marketers. Those are the folks that are really good at understanding those- Those two

    [20:51] Christina Ellwood: PLG marketers.

    [20:52] Doug Bell: PLG marketers are gold. Mm-hmm. Yes. Agree.

    [20:55] Christina Ellwood: By the way, I hope they're not girls. I hope they're women.

    [20:58] Doug Bell: Sorry. Yeah, and you know what to say, Christina, I deserve to be scolded 'cause I am a feminist and I think my wife just heard me from UCSF. Well- You were correct. Female. Thank

    [21:06] Christina Ellwood: you. We'll do a counseling on you. Yeah, for sure.

    [21:08] Doug Bell: Yeah.

    [21:08] Christina Ellwood: Yeah. Okay. So, so- I'm with you. Yeah. Yeah. That's, that sounds... I know you meant it with love. What do you envision the marketing organization looking like two or three years from now?

    [21:22] Doug Bell: I don't know if your audience wants to hear this

    [21:25] Christina Ellwood: Be honest. What do you really think it's gonna look like? You can take a shorter horizon if you want, but I think it helps to be a little further out because then we get rid of all the assumptions about it depends, it depends. Do they learn? Do they, do we train? Do we inv- do we invest? Do we whatever? So take it out to a window where we've, we've, we're past all the learning side of it. What does it look like?

    [21:43] Doug Bell: Yeah. John Miller put a post out yesterday, by the way, I'm gonna refer to it to give you an answer. What, but John Miller basically said in this post was that fractional CMOs were ruining the market. I don't know if you saw this post, but what he was arguing is that there's a long-term window, a long-term view on brand. Well, a lot of respect for John, by the way, and I get what he's saying. He's wrong, and I'm gonna explain why, and I'm gonna explain what the organization of the future looks like. Right now, things are so hypercompetitive, especially in SaaS, that some of these big brands are about ready to topple over. I won't mention who, but they're wobbly right now. And so this idea of building brand or awareness over time, this has got to take a backseat for a little while. So this is why you're not gonna hear one of the more sacred parts of marketing organizations in what I'm gonna mention right now, which is to say this: It'll be a data org, and that data org is gonna be entirely about the science of reacting to signals from the market. I don't mean intent data. Data you're actually gathering from your CRM, from your MAP, from your first and third-party data sources, and from the amazing coding you're gonna do to go acquire and create data sets in something called an MCP server. And that, your-- literally, your marketing team's gonna be a data team. Now, I'm not gonna give you the depends piece, but the efficacy and quality and sort of data science-like quality of that team will depend. If your ACV, average transaction, is big, it's gonna be less of a data science org and more of a sort of programmatic and much more of a maybe a brand-centered org. The closer your ACV is, sorry, average contract value, is to, say, 10K, the more like a B2C data science org that's gonna be. So in the future, I fully believe that you're gonna have non-siloed organizations that are based on a data org that is about data science and about reacting to market signals, and that's gonna be led by what I'll call pod leaders. So instead of thinking about a hierarchy, think about the data organization at the center, then each pod leader is gonna be responsible for something: a product, a market, a geography. And they're always gonna plug in, and so the core competency is there. If you remember, it was really, if you will, at the end of the day, their core duty is managing and reacting to data, and then orchestrating. So think about orchestration like Clay, but you could see other brands like LeanData having a really big piece in this, the big orchestration brands. They'll be actually systems of action, if you will. So you're not gonna have a digital marketer- It's not happening. Product marketing? Nope. None of those things will exist. You will have a data organization, and I don't think that's three years, Christina. I think, frankly, I'm seeing it already in my portfolio, some version of that. Usually it's, "Hey, I'm responsible for recycling. I'm responsible for top of funnel." Like, they don't have titles, and it's all about the data all day

    [24:26] Christina Ellwood: Where do you see strategy, where do you see strategy fitting into that picture?

    [24:30] Doug Bell: That's you and me. Not you and me, sorry, this is the avatars that you or I, people that are in the seats, the C-suite, that level of folks. Those are the folks that are saying... 'Cause guess what? Directionality does not come from data. Directionality comes from judgment. And that, that judgment, the fact that you're in that C-suite, you've got amazing judgment, and you're probably an outstanding pattern matcher, and you're gonna lean into that stuff, right? Because what's gonna happen is, 'cause again, sorry John, you're wrong. What's gonna happen is you're gonna have to create that sort of long-term layer for the brand. You are gonna have to be the brand thinker who's reacting to the immediacy of that data. But your competition, and a- again, I'm a little SaaS-centric right now, so I, I would say to other businesses maybe a difference. But what's happening right now, um, again, I spoke at an event yesterday. I had a conversation with I swear to God 40 founders, all of whom are AI-centric SaaS platforms, all of whom are ready to eat your lunch, 'cause guess what they're doing? They're adapting methodologies and everyday AI 'cause they don't wanna get money from VCs, and they want to have a really low burn rate. And I'm not kidding, there was 150 people in the audience, 40 of them and I had serious conversations, and they're ready to move and they're gonna move on you hard. So it's about your ability to react now. But here's the beauty. Your moat is your brand. They don't have that, and your moat is data. And the quicker you get your data organized and the more you lean into your brand, the better. But guess what? They're coming for you Well, that- Yeah ... that was a sour note. I'm so sorry. Yeah.

    [26:00] Christina Ellwood: It's that you spoke your truth, and I appreciate- Yeah your candid representation of the truth. Obviously, it is not the only version of the truth out there, as you- Absolutely ... have already mentioned. But it is good to hear your perspective, which is informed in an unusual way because both you do fractional work and you're doing your Cannonball work, and you're also a public figure. So all of those feed into it. I have two final questions for you. One, what resources do you recommend for our listeners?

    [26:27] Doug Bell: Yeah. I think I've mentioned Cannonball way too much, so I'll just put a button on that and say it's-- We're called Cannonball GTM. We're on Substack. We have a website. It's not for everybody, as Christina mentioned. We're in this place right now where it's very much about the go-to-market engineers. It's about very innovative and advanced SDRs, and that's just where we are currently. We are trying to make it for founders. There's gonna be a version for founders we're gonna launch here in the next couple months. I really love what Eric Nowoslawski, what Ken Norton, and what Kellen Casebeer are talking about. It's Kellen Casebeer. It's called The Deal Lab, deallab.com. Just a brilliant thinker. He could have gone to any university he wanted to. He decided not to, and he's just been obsessed with data since he was 18 years old. He's just off the charts brilliant. Big Merle Haggard fan, by the way. If you're trying to get business from him, mention that. He'll let you talk to him more. And the other is Eric Nowoslawski. Eric Nowoslawski is looking at the world through the lens of how do I rapidly use AI for rapid message iteration that turns into fast campaigns that produce results. It's another way to look at it. I would highly recommend there. And then after that, I think you're just-

    [27:29] Christina Ellwood: Does he have a website? Or, or does there, is there a domain or, or-

    [27:32] Doug Bell: Just Google Eric Nowoslawski, N-O-W-O-L-S-L-S-W-A-S-K-I, Eric Nowoslawski. And I'm so sorry, I forget Eric's brand. I would lean into those folks, but I also have to say get on LinkedIn. Follow those guys, right? Y- Suddenly your feed will populate with others. But, and Christina, I read a bunch of other Substacks. The, the other Substack I would recommend is called Outbound Kitchen. Um, brilliant guy named Elric. He is a Frenchman living in Mexico City of all places. Highly recommend you look at that, especially if you're ramping or rethinking your outbound.

    [28:04] Christina Ellwood: Okay. We are looking forward to having you at the AI Realized Executive Summit, or excuse me, Executive Roundtable on September the 17th. We also hope to have you at AI Realized Summit on November the 5th. Thank you so much for your time today, Doug. I have my final question for you is: In this world of AI and the new role that you are playing, what is the single most valuable leadership skill you're using today?

    [28:33] Doug Bell: I give two? Am I allowed to give two?

    [28:35] Christina Ellwood: You're allowed to give two, but not three. Rationing you.

    [28:40] Doug Bell: I love you, Christina. Listening. And judging. The listening piece is this: I have to tell you, everything I thought I knew, I- it just, it's all breaking down, so I am, I'm a, just a sponge. Tell me more, tell me more, tell me more, tell me more. And then it is judgment. It is judgment times 10, because again, you're-- the thing we like to say is you're the guide, not the river. So in other words, don't try and be the river. Don't try and paddle upstream. Don't try and fight the river. Stay in the, stay in the course. So that's your job. You're the guides, the boat guide, whether it's like wa- white water raft guide. You're that person. You got a whole team on that thing, and guess what? You hit a rock, you drown the group. So it's about your really good judgment as we go through. But you know what? You don't see the river without listening to your team. There's a rock, there's a tree, there's a rapid. Those are the two things I'm leaning into most, and this is me practically every single day. It's taking me a lot more time to get through meetings, people. It's not about doing, it's about listening.

    [29:33] Christina Ellwood: I got it. Sounds really good. Doug Bell, founder, co-founder of Cannonball GTM and fractional CMO at Chief Outsiders. Thank you so much for joining us today on AI Realized podcast, and we will see you soon.

    [29:50] Doug Bell: Bye everybody. Thanks so much, Christina. Bye.

 
Listen to other episodes
 
 
 
Join AI Realized
 
 
 
Subscribe to the newsletter
 
 
 
Christina Ellwood
Previous

The Technology Works. The Deployment Is What Fails
Next

More Podcast Episodes