Justin Welsh’s personalized course launch did not win because an email greeted people by name. His team asked what kind of business each subscriber had and what stage it was in, then changed the sales argument in both the launch emails and the sales page.
Justin reports that the control group purchased at 8.69% and the personalized group at 12.01%—a 3.32-percentage-point increase, or roughly 38% relative lift. He attributes about $650,000 in additional launch revenue to that difference.
That is a striking result. It is also a result from one course launch, not a promise about what personalization will do for every launch. The useful part of this case study is the design: the data had a job, the treatment crossed email and website, and a holdout kept the original experience available for comparison.
What the reported 38% lift actually measured
The Creator MBA interest list had about 18,000 people. According to Justin’s account, the team first wrote a standard six-day launch sequence for the most common persona in the survey: someone who had not started a business yet but wanted to sell digital courses.
That became the baseline. The personalized treatment then changed multiple parts of the experience:
- Every launch email included some sentences adapted to the reader’s survey data.
- Six additional emails addressed specific audience segments, including people who had not started, side-hustle operators, and full-time business owners.
- The sales-page header and subheader reflected what the team knew about the visitor.
RightMessage’s full Justin Welsh customer story reports that 15% of the interest list was held back from those changes. That group received the default emails and default sales page. The other 85% received the personalized launch treatment.
Justin reports an 8.69% purchase rate for the control group and 12.01% for the personalized group. Dividing the 3.32-point difference by the 8.69% control rate gives the reported 38% relative lift.
The boundary matters: this was a combined-treatment experiment. It can tell us how the personalized email-and-page experience performed against the original experience. It cannot tell us how much of the difference came from the emails, the sales page, a particular segment, or a particular line of copy. The public accounts also do not provide raw delivered-recipient counts, randomization details, confidence intervals, or the calculation behind the $650,000 estimate.
So the evidence-safe conclusion is narrower than “personalization adds 38%.” Justin reports that this launch’s personalized group purchased at a higher rate than its holdout, and that the difference helped produce about $650,000 in additional revenue.
Collect audience context the launch can use
Before writing variants, Justin used a RightMessage survey to learn where subscribers were in business, how they primarily made money, what platform they used, what challenges they faced, and more.
The launch eventually centered on two dimensions:
- Business stage: not started, side hustle, or full-time.
- Primary income source: coaching, agency services, courses, software, or something else.
Those answers were useful because they changed the case for buying. A side-hustle operator could receive a message about making the move to full-time. A course creator could see sales-page language about building a course business instead of the vague phrase “online business.”
The survey was not a research project that ended in a spreadsheet. Responses synced to the subscriber’s record in Kit, and people who had already answered were not asked the same questions again when they joined The Creator MBA interest list.
RightMessage is smart. If someone had already completed my survey when they first joined my list, they wouldn’t be asked those questions again when they joined The Creator MBA interest list.— Justin Welsh
RightMessage’s customer story says Justin reported an 86% completion rate for the subscriber survey. That is worth treating as his result, not as a survey benchmark. More important than the percentage is what happened next: each chosen answer had a planned use in product research, copy, or routing.
If you are designing the data model for your own audience, start with the message or decision a segment will change. Do not collect company size, business model, or current goal just because those fields look sensible in a form.
Turn two dimensions into one coherent launch
The team did not write a completely separate funnel for every possible combination of answers. They kept the baseline sequence, personalized selected sentences, and added six emails for the segments that warranted dedicated treatment.
That kept the work proportional. It also meant the generic launch could remain a real fallback for subscribers without enough data. Personalization fails operationally when the “unknown” audience gets broken copy or no path at all.
The same context followed the click. A course creator who received a course-specific email saw a related header and subheader on the sales page. An agency owner did not click a message about scaling an agency and land on a page that suddenly forgot everything it had just said.
That continuity is the heart of website personalization in a launch: adapt the important parts of the regular page you already maintain so the promise, proof, and destination agree. The team did not need a pile of disconnected sales pages for every segment.
Survey replies also influenced the product itself. Justin said personalized questions in his behind-the-scenes emails revealed demand for a community, which he then offered as a higher tier:
I ended up adding a course community simply because the personalized asks I included in my emails led people to share that this was something important to them. I included this as a higher tier offering, which alone added over $280,000 in new revenue.— Justin Welsh
Do not add that $280,000 to the reported $650,000 personalization lift as if they were independent buckets. One figure describes revenue Justin associated with the community tier; the other is his estimate of revenue associated with the difference between experiment groups. They come from overlapping parts of the same launch story.
Protect the original experience as a holdout
A personalized message can sound obviously better in a copy review and still fail to improve purchases. That is why the default path matters.
For this launch, RightMessage reports that the 15% holdout got neither the personalized emails nor the personalized page. Keeping both surfaces aligned made the comparison understandable: original launch experience versus personalized launch experience.
It also set the limit on what the team could learn. Because email and page changed together, the experiment measured the package. If the question had been “Does the personalized sales-page header improve purchases after the same email?”, both groups would have needed the same emails and different page treatments.
Before copying the 15/85 allocation, check whether your audience is large enough to learn from it. A smaller launch may produce too few purchases in a 15% control to distinguish a real difference from noise. The published case does not provide enough raw data to prescribe a sample split or assess statistical confidence.
Decide the claim you want the test to support, choose a primary outcome before launch, and preserve an untreated experience that differs only where the hypothesis requires. Then report absolute rates as well as relative lift. “Up 38%” sounds much bigger without the 8.69% and 12.01% rates beside it.
What to borrow for your next course launch
The reusable playbook is not “add a survey and expect $650,000.” It is a sequence of decisions:
- Start with one sales decision. Identify where a reader’s stage, business model, goal, or objection would materially change the pitch.
- Ask only for usable context. A short quiz or survey should write answers somewhere the email and website can use them.
- Build a strong default first. You need a coherent experience for unknown visitors and a credible baseline for the test.
- Change a few high-leverage moments. Adapt the email argument, proof, headline, or subhead instead of multiplying the entire launch.
- Keep the destination consistent. The page should continue the specific conversation the email started.
- Hold something back. Compare purchases or another commercial outcome against the original, then keep the limits attached to the result.
RightMessage’s account says the tool setup and Kit connection took about an hour, while segmentation and writing personalized copy took about 20 additional hours. That distinction is refreshingly practical. Moving data can be quick. Deciding what to say to each group is the work.
If you want the complete customer narrative and implementation detail, read the Justin Welsh customer story. Or see how RightMessage helps you adapt the pages you already have, keep a default path, and measure whether the personalized experience actually sells more.
