Notes from the work · CEO, Sneeze It
David runs Sneeze It, the performance marketing agency for multi-location fitness, wellness and med spa brands that builds its own software.
In six weeks this fall, Meta launched an agent that buys things, Amazon blocked it, and agent orders more than quadrupled. Part 1 of Beyond the Four Walls: what changes when the customer stops walking in and starts sending software.
An AI assistant sent to book a call with us would have been turned away at the door. Here is what was wrong, what we changed in one afternoon, and the one point we chose not to chase.
Our free benchmark report opens your website on a phone and checks nine things. Here is what each one is, why it matters to someone who just tapped an ad, and how to fix it.
How many ads to launch, which kinds to test first, and how to judge them at the end of the month without fooling yourself.
Two of the lower cost ad looks in our data are also two of the easiest to script. Here is how to make each one, and how to tell if it is working.
Headline-only ads cost $34.35 per lead in our data, 83% more than before-and-after. Feature and benefit lists were close behind. Here is how to rebuild both into ads that show instead of tell.
Phone-shot ads became winners at more than twice the rate of high-production ads in our data, and at a lower cost per lead. The group is small. Here is how to test it without a crew or a budget.
The way an ad opens changed its cost per lead more than any other label we tracked, from $14.30 to $73.48. Here is what each opening did, which groups are too small to trust, and how to test your own.
Before-and-after ads brought the lowest cost per lead of any look we labeled. The group is small, so here is how to test one properly, and how to make it without stretching the truth.
Some kinds of ads bring cheap leads but rarely scale. Others scale often but cost more per lead. Here is how to read both numbers and pick the one that fits the job in front of you.
Cost per lead is the most useful number in a lead-gen account and the easiest one to misread. Here is how to compare it fairly, and how to tell a real difference from a lucky week.
Meta decides which ads get the money. You decide which ads get the leads you can afford. In our data those two picks agree less than half the time, so every ad needs two grades.
Nearly half of all the money in 18,920 Meta ads went to the 6.2% that became winners. Here is how to manage that concentration, spot a tiring winner and replace it before it fades.
Two of every three Meta ads we graded spent under $100 in their whole life. Here is how to read a new ad early without cutting a winner before it starts.
Accounts that launched four or more new ads a week found about one winner a month. Accounts under one a week found about one a quarter. Here is how to pick your pace.
Our free scanner asks whether an AI assistant can actually book or buy on your site. Here is exactly how it decides, and the rules that keep the score honest.
Ideate takes one hunch through six steps and ends with the cheapest test you could run this week. Claude adds ideas when asked. It is never allowed to rank them.
When someone on our team fixes something an AI wrote, that fix becomes a lesson six Sneeze It tools read before they write again. Here is how the Brain decides what counts, and what never goes agency-wide without a human yes.
We let our team ask every Sneeze It tool a question from inside Claude or ChatGPT. Before we built the door, we wrote down who may never walk through it, and what may never come back out.
Coach uses AI to write the plain-English paragraphs of a client report. Code computes every figure, a checker rejects any number the AI was not given, and a person reads it before it goes.
A report built to find good news teaches everyone to miss the bad news. So we took the Wins section out of Coach and put five plain comparisons in its place.
The three-mile radius is the most common map in site selection, and it is wrong in ways you cannot see on the map. Here is how Audience draws the market instead, and what changed when it did.
Before building a forecast for new sites, we tested the industry's favorite inputs against 865 ZIP codes and about $102,000 of real ad spend. Here is what Audience uses instead.
Our team spends its day in two places, the public Ad Library and other people's websites. So we built a Chrome extension that brings our own tools to both, in four versions over two days.
Meta grades your ads by where it puts the money. We grade them a second time, by what a lead actually cost, and that second grade decides what our team makes next.
Our team can take an ad design that is working in the market and rebuild it in your brand, your words and your approved offer. The AI does the design work. Software guards your logo, your prices and your reviews.
Before a new ad reaches your inbox, Studio checks it against what has actually brought you leads at a good cost. If the read it is checking against is more than a week old, it refuses.
Most ad approvals happen on a flat image in an email. In Studio you approve the ad inside the real Facebook, Instagram or Google frame, and the version you approve is the version that runs.
One" lives in our company Slack. It answers questions, takes notes, changes facts and files requests in plain words, and every change it makes is recorded in the Hub under the name of the person who asked.
Our request desk now asks for the deadline, the other locations and the files up front, files every request on the location it is about, and stays in sync with the Jira desks our team already works in. Here is what is live, and the one part we have not proven yet.
On a franchise marketing call, "not connected" is the most useful thing a location page can say, as long as it says which kind. Here is how the Hub tells them apart, and why each one is a different job.
Our creative, reporting, market, call center and AI tools all ask the Hub who a client is. Here is how that works, the rules that keep it clean, and what it changes when you tell us something once.
Our first portal was organized around a process. A day later we started again with one page per gym, studio or spa. Here is what that page holds, and why a record beats a workflow.
Every tool our team uses to make, run and report your marketing is one we built, at *.sneeze.it. Here is what each one does, who uses it, and how they all read from a single record.
Gyms, studios, med spas, salons and recovery chains, scored on whether an AI assistant can find them, understand them and book. The median was 46 out of 100. Not one site lets an agent book a class or appointment.
Banks, card networks and Cloudflare spent the last year building ways for an AI agent to prove who it is. Most local business websites are still checking everyone at the door the old way. Part 9 of Beyond the Four Walls.
Google, Microsoft, Anthropic, Perplexity and Opera have all put an agent inside the browser that can click, type and fill in forms. OpenAI built one and shut it down. Part 8 of Beyond the Four Walls: what that means for the form on your website.
Google can already call a salon for prices, book through three beauty and wellness platforms, and take payment for a hotel inside a chat. Part 7 of Beyond the Four Walls: why your Business Profile now matters more than your homepage.
Amazon built its own shopping agent, sent it to buy on other brands' sites, and fought to keep outside agents off its own. Part 6 of Beyond the Four Walls: what that playbook teaches a gym, studio or spa.
Meta is automating both ends of local marketing: machines that buy the ads and machines that answer the people who click them. Part 5 of Beyond the Four Walls: what that means for a gym, a studio or a med spa.
The web reserved a "Payment Required" code in 1997 and never used it. Now Coinbase, Cloudflare, Stripe and Visa are building rails for software that pays a few cents at a time. Part 4 of Beyond the Four Walls.
In one year Stripe co-wrote a shopping protocol with OpenAI, gave agents their own wallet, and quietly made checkout pages for 7.8 million businesses readable by software. Part 3 of Beyond the Four Walls: what that means for the studio or spa that takes payments online.
robots.txt, structured data, llms.txt, MCP and WebMCP, in plain English, and which ones a gym, studio or spa should fix first. Part 2 of Beyond the Four Walls.
Can it get in, can it understand you, can it do something, and can it finish? A plain guide to what makes a local business website agent ready, with the points each fix is worth.
We graded every Meta ad we ran for gyms, studios, med spas and salons. The popular formats are not the cheap ones, and Meta's favorite ad is often not yours.
Book a walkthrough with David Steel. Bring last month's lead count and one location you are worried about.
