Business coach, live events · Performance
How did a business coach get 120 live workshop buyers at $160 each?
By rebuilding its Meta ads around the formats that had actually sold and scaling by duplicating winners instead of pushing budgets. The June 2026 workshop launch brought in 120 front-end buyers at $160 each. The launch before it, in February 2026, brought 70 at $257.
120front-end buyers, June 2026 launch
$160cost per buyer, down from $257
71%more buyers than the February launch
1.70xblended return on ad spend, December 2025 launch
- What we did
- Kept the formats that sold: the founder on camera for cold audiences and the full sales video for retargeting. Cut the ones that did not, including static images, carousels and motion graphics. Scaled by duplicating proven ad sets, which held cost per buyer near $160 while daily spend rose from about $400 to $2,600.
- What the data showed
- Slicing the results by country found Canada costing $415 per buyer against $126 in the United States, a 3.5x gap backed by a weaker click-through rate. The blended number hid it. We recommended moving that spend to the US.
- What we learned
- A workshop launch is judged on what buyers spend after the workshop, not on the front-end cost alone. The December 2025 launch returned 1.70x on $13,442 in ad spend once follow-on program sales were counted.
Source: launch benchmark readouts from Meta Ads and checkout data, December 2025 to June 2026. Front-end buyers exclude test purchases and warm email carts.
Physician-led info product · Performance
Why did one audience look more valuable than another?
It was not the audience. Two ad audiences for a physician-led info-product business looked $2.73 apart in average order value. Broken down, their front-end revenue per order was within three cents. The whole gap was one $49 add-on, taken 7.2% of the time by one audience and 1.8% by the other.
$2.73apparent order value gap
3¢actual gap on the core product
7.2% vs 1.8%take rate on the $49 add-on
- What we run
- Meta ads, the sales funnel, order form and upsell tests, split tests on the sales page, and a daily sales readout posted three times a day in the client's own time zone, with ad and organic revenue kept separate.
- Why it mattered
- "Spend more on the better audience" and "make the add-on easier to take" are different decisions. Reading the order value split before acting turned an audience question into an offer question.
- What happened next
- In October 2026 the add-on moved from a checkbox to a one-click upsell with a lower-priced downsell, run as a recorded test with a fixed cut point. It is too early to report the result.
Source: ad set order-value readout, September 2026; order form test register, October 2026.
Local contractor, Waterloo Region · Local
What happens when a contractor's website is built for Google and AI answers?
More people find it. We rebuilt one local contractor's site as 23 pages written for SEO and AEO, with a page for each city he serves. In the six weeks after it launched in September 2026, Google showed the site in searches 70% more often than in the six weeks before, and clicks from Google rose by about a third.
+70%times shown in Google search, 6 weeks after vs before
+32%clicks from Google search, same weeks
23pages, one for each city served
- What we built
- City pages with a real local angle rather than copies with the name swapped, such as older stone foundations in one city and building rules for new homes in another. Every page answers the questions people actually ask, in a form Google and AI tools like ChatGPT can quote. Alongside it, an AI receptionist that answers from his own pages, checks every sentence against the claims he approves, and passes calls to his cell when needed.
- How we are spending
- Search ads launched in October 2026 at $20 a day, to learn cost per lead, per booked inspection and per finished job before any increase.
- Results so far
- Search numbers are early and still small, about 17 clicks a week, but moving the right way. Leads and booked jobs get added here once a full quarter of data is in.
Source: Google Search Console, weeks starting July 27 to August 31, 2026 (3,653 impressions, 76 clicks) against September 14 to October 19, 2026 (6,194 impressions, 100 clicks); client project records, August to October 2026.