Set a baseline before you run a sales experiment
Record the existing result, audience, period, and conditions so a test has something honest to compare against.
You change the price of a digital product on Monday, send an email on Tuesday and see three sales by Friday. It feels like the experiment worked. But compared with what? If you did not record what was happening before the change, you have a result without a reliable baseline.
What a baseline is for
A baseline is a short record of what normally happens before an experiment begins. It gives you a reference point, not a promise that the next period will repeat it. Sales vary because traffic, timing, audience intent and the offer itself vary. The baseline helps you separate an ordinary fluctuation from a result worth investigating.
For example, suppose a product page received 240 visits and made 6 sales over the previous 14 days. Its observed conversion rate was 6 ÷ 240 = 2.5%. That is the starting observation. It is not the product's permanent conversion rate, and it does not tell you why those six people bought.
| Record | What to write down | Why it matters |
|---|---|---|
| Result | Sales, revenue, visits, leads or another chosen measure | Defines what success means |
| Audience | Where visitors came from and who they were | Shows whether you are comparing similar people |
| Period | Exact start and end dates | Prevents vague memory from becoming the comparison |
| Conditions | Price, page, promotion, product version and delivery | Makes the test reproducible |
Choose one question before you change anything
A sales experiment should answer a specific question. “Can I improve this offer?” is too broad. “Does showing the outcome in the first section of the product page increase purchases from people arriving through my weekly email?” is narrow enough to measure.
- 1Write the current version: price, title, description, images, checkout path and delivery method.
- 2Choose the one change you want to test, such as a clearer promise, a different price or an added bonus.
- 3Choose one primary measure, such as completed purchases per relevant page visit.
- 4Set the comparison period before you start, rather than ending the test when the numbers look pleasing.
- 5Record unusual events, including a launch, mention by another creator, technical fault or unusually large email.
Make the baseline specific to the audience
A baseline made from all traffic may be misleading if your experiment targets one source. Visitors from a search result may be looking for a solution, while visitors from a social post may only be discovering your work. Combining them can hide a change in either group.
Record the source when you can: email, search, social, referral or direct visit. Also note whether the audience already knows you. If your test is a discount sent to existing subscribers, compare it with earlier subscriber emails rather than with every visitor to your storefront.
The same principle applies to the product format. A course, coaching call and downloadable template can each have a different buying decision. If you sell several formats from store.fan, keep their baselines separate rather than combining every sale into one average.
Keep the conditions visible
Write down anything that could plausibly affect the result. Include the price and whether a discount was active, the wording and layout of the page, the product version, the call to action, the checkout process and how the buyer received the product. If you use email, record the send date, audience and subject line.
This is also where your storefront setup matters. A change to your page design, checkout or delivery can alter behaviour even when the sales copy stays the same. A storefront on store.fan keeps the product and payment journey in one place; if you test the page itself, record the relevant version and use the available analytics to check the same measure before and after.
Compare the numbers without moving the goalposts
Use the same calculation in both periods. If the baseline is 6 sales from 240 visits, the rate is 2.5%. If the test produces 9 sales from 300 visits, the rate is 3%. The difference is 0.5 percentage points. The relative increase is 0.5 ÷ 2.5 = 20%, but that does not mean the change caused a 20% improvement.
Check absolute totals as well. A higher conversion rate with fewer visitors may produce fewer sales overall. Revenue can also move differently from sales if the price changed. Decide in advance whether the experiment is meant to improve purchases, revenue, profit, leads or something else. Then keep that primary measure fixed.
Use a period that reflects your normal buying cycle and traffic pattern. The important part is to set the dates before the test and record enough observations to make the comparison meaningful. If your sales are infrequent, expect the answer to remain uncertain for longer.
You can, but you will have less ability to explain the result. Changing the price, page headline and bonus at the same time may improve sales, yet you will not know which change mattered. Test one central change first unless your question is specifically about a combined offer.
Use a simple document or spreadsheet with the dates, metric, audience, conditions and result. Add a short note about unusual events. The best system is one you will update consistently, not one with more fields than your experiment requires.
After the test, write one sentence about the decision: keep the change, undo it, or run a cleaner follow-up. If the result is mixed, say which part is uncertain. This turns a sales experiment into a sequence of small decisions instead of a hunt for a dramatic number.
Set up a storefront for your next measured offer and keep the conditions in one place.
Start with store.fanTurn your knowledge into income
Launch your Store.Fan in minutes — sell digital products, courses, and calls straight from your bio. Free to start.
