How should small professional services businesses split their marketing? I mean businesses where reputation and competence decide the sale: healthcare and allied health practices, lawyers, accountants, and people like us who sell SEO, AEO or AI SEO. My current thinking: Being found: - Good AI SEO does the job of being found on Google, Maps and AI search engines (ChatGPT, Gemini, Copilot). I do not think Google Ads are needed for that. - LinkedIn, Instagram and YouTube, in that order, also play a major role, because people check a business there before they contact it. - LinkedIn in particular works mainly as an online CV and a credibility check. Reaching people who are not looking yet: - Meta ads (Facebook and Instagram) are where owners and clients can be shown a problem they do not yet know they have. Advertising: - One marketer I have learned from ranks paid advertising in this order: Meta first, then Google (including YouTube), then TikTok, then LinkedIn last. His view is that LinkedIn advertising only pays for high-ticket sales (around USD $10,000 or more) to companies with 50 or more staff, or aimed at executives. Another says that even business owners tend to make their buying decisions through Facebook or Instagram. For trades such as plumbers or HVAC specialists, my view is that LinkedIn can be ignored completely, and that AI SEO plus Meta ads covers what they need. Is that right? For a professional services business with a small budget and limited time, where would you start, and how much weight would you give Google Ads, TikTok ads and LinkedIn ads?

I saw a strategy from an Australian agency that markets to allied health practices, and I would value your views on whether it holds up. They call it a three-rung "growth ladder" for deciding what a practice should pay for, and when: Rung 1: one or two practitioners at one location. A website and a Google Business Profile, built properly and matched to each other. Every service listed on the profile (up to 30) has its own page on the website, the details are the same everywhere, and booking is made simple. AI SEO is built into the website at the start. Their claim is that this is a one-off cost, and that most practices of this size never need monthly SEO or Google Ads. Rung 2: adding practitioners or a second location. Add Google Ads, because ads are the fastest and most measurable way to fill a new practitioner's diary. Their advice is that a practice at this stage should put money into Google Ads, not SEO. Rung 3: several locations and several disciplines. Add ongoing SEO on top of the ads, because keeping several profiles, locations and services in order is ongoing work. They say SEO usually takes more than a year to pay back, so only a larger practice can wait for that return. Their proof is one chiropractic clinic: online bookings rose from 5.9 to 9.3 a day in the first 10 days after the rebuild, with the same ads running. My questions: 1. Is this a workable strategy for a one or two practitioner practice? In particular, is rung 1 really enough on its own, or does AI visibility need ongoing work once the website and profile are right? 2. At rung 2, their choice is Google Ads. Would Meta ads (Facebook and Instagram) be a better or cheaper way to fill a new practitioner's diary? Or does Google win because people are already searching when they need a physio? 3. They also say that when someone asks ChatGPT for a good physio nearby, ChatGPT reads the Google Business Profile first. Is there any evidence for that? 4. How would you position and price ongoing AI SEO for very small practices, when competitors tell them a monthly AI SEO service is unnecessary?

So in case you guys missed it I posted about how Lovable and AI website builders are a scam on X and it ended up going viral lol - https://x.com/ryan_doser13/status/2110471378189312039 As you guys know I will NEVER recommend AI SaaS products like Lovable, Replit, Base44, Framer, and the 100s of other AI website builders that exist. These companies spend millions of VC backed dollars on marketing to convince you it's a magic solution. But in reality you are paying a premium for less usage, shitty models, shitty websites, and hand cuffing yourself to a SaaS ecosystem you don’t control. Here's my stack for building websites with AI that I talk about all the time: - Claude Code or Codex ($20/month minimum but I suggest $100/month) - GitHub (free) - Cloudflare or Vercel (free) - Astro web framework (free)

Wow!! Just listened to Friday’s replay and it was SO FREAKING GREAT! Be there Wed for the SEO and dangggg @Ryan Doser I don’t know how these calls keep getting better and better but they do and appreciate you BIG TIME!!!!! Thank you for keeping it real and soooooo much value in every call - not to mention amazing ppl! Let’s goooooo! If you’re not in the VIP you’re missing out!

I'm not that good at prompting. I've got an energy limiting condition, and remembering a clever prompt while I'm thinking about the actual work is too much most days. So I screenshot good prompts and save them. Then they sit in a folder and I never use them, because using them means remembering they're there. I read Felix Rieseberg's article on being more ambitious with AI. He leads engineering on Claude's apps at Anthropic, and it's one of the more practical ones I've read: https://felixrieseberg.com/how-to-be-more-ambitious-a-tutorial-for-humans-with-models/ Instead of trying to learn his techniques, I asked Claude to build them into its own instructions. Now it asks itself three questions before it starts on any proper piece of work, and I just carry on asking the way I always do. 1. What's the goal beyond the goal? "Write me a LinkedIn post" is really "get the right people to my website". Sometimes the answer is that a post isn't the best way to do that. 2. What would the best person in this field do? A good copywriter starts from what customers actually say, their emails and reviews, not from a summary of the product. 3. What would someone from a different field see? A newsletter is really "getting busy people to read one email a month". Journalists are good at exactly that. It only tells me when one of those changes the plan. Otherwise it just gets on with the better version. You don't need Claude Code for this. Paste the three questions into your custom instructions in ChatGPT, or a project's instructions in Claude, and ask it to run them before it starts. His fourth point was never to let the AI mark its own homework. I already had that one. Anything I'm going to rely on gets checked by a second Claude that didn't do the work. The end of his article is about self-efficacy, people doing more when they believe something's possible. One of the ways he suggests is giving yourself permission to fail. I suspect that's what trips most of us up, more than any prompt. Ryan says all the time that he's not technical, and look what he's built.