“My course launch flopped” — reading the numbers before you rebuild

Do not relaunch yet. Divide your launch registrations by your list size in thousands. Under 10 registrations per 1,000 list members is a reach failure, not an offer failure, and rebuilding the offer will not fix it. And if the launch was only ever going to produce one expected sale, finishing at zero means nothing — that happens about 37% of the time.

• The one number: R/1k = launch registrations ÷ (list size ∕ 1,000). It separates a reach problem from an offer problem from a launch-delivery problem.
• The floor: below three expected sales, a zero is not evidence about your offer.
• Costs nothing: the whole diagnosis is two exports — registrations and purchases — and about an hour.
• Do this before: touching the sales page, cutting the price, or re-recording a module.

Was it a failed launch, or a launch too small to test anything?

Most launches called failures were never large enough to produce a readable result. Work out the expected sales first: registrations × attendance (or reply) rate × buy rate. If buyers arrive independently of one another, the chance of finishing at zero is roughly e^−expected, which is where the numbers below come from.

Worked through with real inputs: a 1,200-person list produced 42 registrations, 40% of them turned up, and your own historical buy rate on a live room is 6%. That is 42 × 0.40 × 0.06 = 1.0 expected sales. Selling nothing had roughly a one-in-three chance before you wrote a single word of the sales page. A launch that could only ever have produced one sale did not test your offer; it tested a coin. Substitute your own three numbers — you already have all of them.

How do I tell if it was my list, my offer, or the launch itself?

These three failures produce the same feeling and need opposite responses, which is why launch post-mortems go wrong. R/1k tells them apart because it is measured before the offer is ever seen: registrations happen off the invitation, not off the sales page. Someone who never registered never evaluated your course.

The bands below are a decision rule, not an industry benchmark. They are anchored to what ordinary email performance alone produces: GetResponse’s benchmark report — 2023 sending data, the most recent edition GetResponse has published — puts the average click-through rate for the education sector at 3.17% per send, so a single email to 1,000 contacts generates about 32 clicks, and a four-to-six email invitation sequence generates several times that. If a whole sequence produced fewer than 10 registrations per 1,000, the invitation never reached people — that is deliverability, segmentation or a dead list, not a weak promise.

R/1k is the only launch number that is unaffected by your sales page, because it is counted before anyone sees it.

“My online course didn’t sell” — what to do in the next 24 hours

None of this costs money and none of it involves buying anything. Export two lists: everyone who registered, and everyone who bought. Calculate R/1k and expected sales. Then pull the three timestamps that explain most launch collapses — when the cart opened, when each registrant last opened an email, and how long after registering the first human contact happened, if it happened at all.

Do not refund-panic, do not email an apology to the list, and do not drop the price publicly. A discount issued in the first 48 hours prices every future launch you run, and it is almost always a response to a sample size rather than to evidence. The single most expensive thing you can do in the 24 hours after a flat launch is change the price.

The next 7 days: work the people who already raised their hand

Your registrant list is the most responsive audience you will have for about a fortnight, and then it decays. Before rebuilding anything, have a real conversation with every non-buyer — not another broadcast email. Ask what they expected to get, what they thought it cost, and what they did instead. Ten of those conversations outrank any amount of analytics.

Older enquiry lists are worth working at the same time, because a flat launch usually sits on top of years of unworked interest. Across our own database reactivation work on dormant lists, the Colliers-era campaigns converted at 4.4% on average with an 8.9% peak — that is our record on our own campaigns, not an industry benchmark, and it is a conversation rate, not a purchase rate. The mechanics of running one are documented in our guide to how to run a database reactivation campaign.

Why fixing the sales page is usually the wrong first move

The sales page is the first thing creators rewrite and the last thing that binds at small list sizes. It can only convert people who reach it, so its influence is capped by R/1k. At 20 registrations per 1,000, a sales page rewritten from good to exceptional might move you from one expected sale to two — a change you cannot detect, because the chance of a zero only falls from 37% to 14%.

The page becomes the binding constraint when R/1k is above 60 and the room is full of the right people. Below about 3,000 engaged subscribers, the sales page is rarely the constraint — the number of qualified people who saw it is. That is also why “my launch flopped” and “my sales page is broken” are usually different problems wearing the same clothes.

What it actually costs to run this diagnosis yourself

The arithmetic is an hour. The expensive part is the follow-up: 40 non-buyers is one evening on the phone; 400 is a fortnight, and you do not have a fortnight inside the two weeks while the launch is still recent enough to be worth mentioning. That decay is the honest crossover. Below roughly 150 non-buyers, do it by hand — you will learn more from your own calls than any system will tell you. Above that, the constraint is not insight, it is dial time.

Running it as a system means owning list hygiene, a dialler or messaging stack, a booking calendar, and someone available when people reply. That is a real operating cost in tools and hours, and it is the reason most creators do the first 40 calls and abandon the rest. Where it is handed over, our own model for online course launch marketing is pay-per-result — you pay on booked qualified appointments rather than on a retainer or a seat count — which exists because launch work is lumpy and a monthly fee prices the quiet months. Do your own arithmetic on which side of 150 you sit; it decides more than any vendor comparison will.

When to relaunch, and what has to be different

Relaunch when you can state, in advance, the R/1k you expect and the number of expected sales it implies — and that number is at least three. If it is not at least three, you are not relaunching, you are re-running a coin toss with more effort attached. Change one variable at a time: reach, or offer, or launch mechanics. Changing all three teaches you nothing, because the next result is unattributable.

Know the real cost per enrolment before you commit, using published cost per student enrolment benchmarks for course creators rather than the last launch’s emotion. A second flat launch on the same list is the expensive way to learn something an export and an hour would have told you.

Frequently asked questions

Why did my course launch fail?

In our experience it is one of three things, and they are routinely confused: not enough qualified people saw the invitation, the offer was refused by people who did see it, or registrants were never followed up after the cart opened. R/1k tells you which. Anchor it against ordinary email performance — GetResponse’s benchmarks report, built from more than 4.4 billion messages sent by its customers in 2023 by senders with at least 500 contacts, puts the education sector at a 41.33% open rate and a 3.17% click-through rate.

How big does my email list need to be before a launch can tell me anything?

At 35 registrations per 1,000, 40% attendance and a 6% buy rate, every 1,000 subscribers produces about 0.84 expected sales. To clear the three-expected-sale floor you need roughly 3,600 engaged subscribers. Below that, a launch is a fundraiser, not an experiment — run it if you need the revenue, but do not draw conclusions from it.

Should I relaunch the same course?

If R/1k was under 30, yes — the offer was never tested, so changing it discards information you never collected. Fix reach and re-run it. If R/1k was above 60 and nobody bought, the offer is the thing to change, and you should interview ten non-buyers before deciding what to change about it.

Do I need to fix my sales page before I relaunch?

Only if enough people reached it. A sales page cannot convert people who never registered, so its effect is capped by R/1k. Below 30 registrations per 1,000 the page is not your constraint; above 60 with a full room and no sales, it is one of the few remaining candidates.

Is another email list the answer, or paid traffic?

Neither is an answer until you know which failure you had. Paid traffic bought to fix an offer problem converts strangers at a lower rate than your own list did, so it makes the loss larger and faster. Reach spend is only rational once R/1k is healthy and the drop-off is downstream of registration.

I run a nationally recognised or CRICOS-registered course — does this still apply?

The arithmetic does, but your constraint may be regulatory rather than commercial. ASQA has suspended new CRICOS applications, including applications to add new courses to an existing CRICOS registration, from 19 May 2026 for 12 months until 19 May 2027, with exemptions for government schools, state-owned VET providers and Table A providers. If your plan was to launch a new course to international students, check your eligibility with ASQA first; this is general information, not regulatory advice.