A hypothesis is not the same as an idea
“Let's try Reels” is an idea. “If we show in Reels how a furniture workshop builds an order, more people will ask about pricing in Direct, because today customers don't understand what they are paying for” is a hypothesis. It contains an action, a metric and a reason.
The reason matters most. When results differ from expectations, it tells you which belief about your customer was wrong. Before launch, write down which result would confirm the hypothesis and which would reject it. Otherwise any number can be explained in your favour afterwards.
Where to find marketing hypotheses worth testing
The best hypotheses sit in your own customer conversations, not in someone else's case studies.
Export the latest chats, including those that ended without a purchase, and read them in a row. An hour or two later the list is usually ready.
- Questions that keep repeating in Telegram and Instagram Direct
- Objections your sales team hears on the phone
- Comments under ads and customer reviews
- The funnel stage where people drop off
- Competitor ads and listings on OLX and Uzum
- Results of earlier tests
How to decide what to test first
There are always more hypotheses than budget and time. ICE scoring helps: impact on the result, confidence in the effect and ease of testing, each rated from 1 to 10. Add or multiply the scores, and the ideas with the best return on effort rise to the top.
For example, rewriting the first screen of a clinic's website in Tashkent has high impact and is easy to check. Launching delivery to the regions may have a big effect, but testing it needs logistics, so it sits lower. Scores are subjective, so set them together with your sales team rather than alone.
How to test without spending much
Start with a minimal test: a part of the audience, a short period, one change at a time. Several formats work:
Do not stop a test on day two because a number looks great or terrible. Mind seasonality too: a before-and-after comparison around Ramadan, Navruz or New Year can mislead, so keep a control group where possible.
- A/B tests of ads and landing pages in Meta Ads and Google Ads
- A pilot in one branch or one customer segment
- A Telegram bot broadcast to part of your customer base
- Short customer interviews for a qualitative check
Why hypotheses need end-to-end analytics
If leads from Instagram, phone calls, Telegram messages and Click or Payme payments are not tied to a source, you only see clicks. A cheap lead may never reach a payment, and a hypothesis will look confirmed exactly where there is no money.
So pick a success metric as close to revenue as possible: a paid order, a visit, a repeat purchase. The end-to-end analytics that markdev builds shows the path from ad to sale and lets you compare hypotheses by real revenue instead of lead counts.
Backlog, rhythm and the right to be wrong
Keep a simple table: the hypothesis wording, ICE scores, status, result and conclusion. Every week or two, pick hypotheses to run and review the outcomes. Regularity matters more than volume.
A hypothesis that fails is still a result: it kept your budget away from scaling a weak idea. Record these conclusions as carefully as the wins, or in six months you will test the same thing again. And if people are blamed for failures, the team stops proposing bold ideas.
Where to start
Pick three hypotheses from your latest customer conversations, score them with ICE and run the strongest one over the next few weeks. If you need to connect ads, CRM and payments into one report so test results are trustworthy, you can discuss the task with the markdev team.
FAQ
How many hypotheses can I test at once?
As many as will not interfere with each other: in different channels or on different segments. On a single page or campaign, change only one thing.
How do I know a test has run long enough?
Set the duration and a minimum number of leads in advance, and end the test when that condition is met, not when the number looks good.
Does a small business need analytics?
At least a table that splits leads by source and marks payments. Even that simple link noticeably improves decisions about hypotheses.
Let’s look at your task for free
Leave your contact — in a 15-minute call we’ll show where leads get lost and what to fix first.