Testing6 min readChecked
A CRO guide for adult stores in Australia: hypotheses, tests you can trust, and where to begin
Conversion rate optimisation (CRO) for adult stores in Australia is only as good as its method. This lab note explains how to form a hypothesis, when an A/B test makes sense and when it does not, how to avoid fooling yourself with results, and which tests an adult store should run first.
01Start with a hypothesis
Every test should begin with a sentence you could be wrong about: "Because shoppers leave at the delivery step, showing delivery cost on the product page will increase completed orders." It names the problem, the change and the measure. Without one, a test is a guess with extra steps.
02Decide whether to A/B test
- A/B testing needs enough orders for a difference to show up; small stores often do not have them.
- With low traffic, test bigger changes, or compare before and after while noting anything else that changed.
- Never test something you would not be willing to keep, and never test anything misleading.
03Read results without fooling yourself
Decide the length and sample size before starting, and do not stop the moment a result looks good. Run for at least one full week to include weekends. Judge by completed orders and revenue; clicks alone mislead. If a result seems too good to be true, run it again before rolling it out.
04Four tests for an adult store
| Test | Why it is worth trying |
|---|---|
| Show delivery cost on product pages | Late costs are the leading reason for abandonment in Baymard research |
| Offer guest checkout | Forced account creation also ranks high among reasons for leaving |
| State billing and packaging plainly | Removes a hesitation specific to adult purchases |
| Tidy the mobile checkout | Fewer fields, larger tap targets, clearer errors |
05Keep a test log
Record every hypothesis, what was changed, the dates, the result and the decision. Over a year this log becomes the most valuable document you own about your customers. The service we offer around this is on conversion optimisation, and landing pages have their own lab note.
06Hypotheses from Australian stores
| Observation | Hypothesis to test |
|---|---|
| Many carts abandoned at the shipping step | Showing delivery cost by postcode on product pages will reduce abandonment |
| Few buyers on phones after 10 pm | A shorter guest checkout will lift late-night mobile orders |
| Questions about packaging in support email | A photo of the outer parcel near the buy button will increase conversion |
| Low uptake of a subscription | Stating the renewal price and cancellation method plainly will raise trust and sign-ups |
Illustrative examples. Each starts from something observed, not from a guess.
07Tests that stay within the law
Some popular tests are risky now and will be riskier after 1 July 2027: fake scarcity messages, timers that reset, pre-ticked extras, surcharges revealed only at the last step, and wording that shames people for declining an offer. Leave them out of the test plan entirely. Persuasion through clear facts, real reviews and honest reassurance can be tested freely, and tends to hold its gains.
08Before and after, when traffic is thin
Many Australian adult stores do not have the orders to split traffic reliably. A careful before-and-after comparison can still guide decisions: change one thing, compare the same weeks before and after, account for pay cycles, holidays and promotions, and treat the result as a signal rather than proof. Note anything else that changed in the same period, such as a sale or a stock shortage, so you do not credit the wrong cause.
09Pay cycles and seasons
Australian spending follows fortnightly pay cycles for many people, and adult stores often see clear peaks around paydays, Valentine's Day, end of financial year, Black Friday and Christmas. A test that runs across a payday and a quiet week can be misread. Run tests over full fortnights where possible, avoid launching them in the middle of a major sale, and compare like periods when reading results.
10What to test first
- Price display: totals with GST and any surcharge shown early.
- Delivery: cost and time visible before the cart.
- Checkout: guest option and fewer fields.
- Product facts: the same key details in the same place on every page.
- Reassurance: packaging photo and statement name near the buy button.
11Sharing results with the team
A test log only helps if people read it. Share a short summary each month: what was tested, what happened, what will change. Include tests that failed; they stop the same idea being tried again next year. Over time the log becomes the most useful record of what your customers respond to. The service behind this guide is conversion optimisation.
12Qualitative research, cheaply
Numbers show where shoppers leave; people tell you why. Read support emails and chat logs, ask a handful of recent customers what nearly stopped them buying, and watch a few anonymised session recordings with personal data hidden. Adult shoppers often hesitate for reasons they will not type into a search box: worry about the parcel, the statement, or whether a product will suit them. Those reasons make the best hypotheses.
13Privacy in testing tools
Testing and analytics tools can collect more personal information than a store needs. Configure them to mask form fields and personal details, keep data only as long as necessary, and describe them in your privacy policy. Under the reformed Privacy Act, collecting less is the safest default, and adult customers are more sensitive than most about being watched.
14From test to standard practice
When a change wins, make it the default everywhere it applies as well as on the page where it was tested: if showing delivery cost early helped one product, apply it to all. Record the decision in the test log and check a month later that the gain held. The landing-page side is in our guide to landing pages that convert, and the wider plan in our growth marketing guide.
15A test log template
| Field | Example entry (illustrative) |
|---|---|
| Observation | Support emails ask what the parcel looks like |
| Hypothesis | A parcel photo near the buy button will lift conversion |
| Change | Photo and one line of text added to all product pages |
| Dates and method | Two fortnights before and after, same promotions |
| Result and decision | Conversion up modestly; kept as standard |
16Stopping rules
Decide before a test starts when it will end and what would count as a clear result. Stopping early because a test looks good on day three is the commonest way to adopt a change that does nothing. If a test runs its full length without a clear answer, choose the simpler option and move on; not every idea deserves a second run.
17When to bring in help
Most stores can run simple before-and-after comparisons themselves. Outside help is worth it when traffic is large enough for proper split tests, when a checkout rebuild is planned, or when nobody has time to keep a test log. Whoever runs the work, the rule is the same: start from evidence, change one thing, and keep honest records.
18Sources cited
The guide draws on these publishers. Visit them for detail and for any recent change.
QQuestions
How long should an A/B test run?
At least one full week, to cover weekday and weekend behaviour, and until the sample you planned for has been reached. Stopping early is the commonest mistake.
What if we do not have much traffic?
Test bigger, bolder changes, or use before-and-after comparisons with care, noting anything else that changed in the same period.
Should we test prices?
Carefully, and never by showing different people different prices without a clear, lawful reason. Test presentation and clarity first.