Running Growth Experiments With a Tiny Team
By CodexierPublished 5 min read
Growth experiments sound like something for start-ups with data teams and millions of visitors. The method is simpler than that, and it suits small teams well: choose the one thing holding growth back, try a specific change, look at what happened and write it down. What small teams lack is not tools but discipline, because it is tempting to change five things at once and then not know what worked. This guide gives you a lightweight system you can run in a couple of hours a week.
Pick one constraint to work on
Draw your customer path in four or five steps and put a rough number on each: people who find you, visit, contact or add to cart, get a quote or reach checkout, and buy. The step with the biggest drop relative to what is normal in your industry is usually the constraint. Improving anything else gives little until that step moves.
| Step | Typical constraint sign | Example experiment area |
|---|---|---|
| Found | Few visitors or searches | Google Business Profile, local SEO, one new channel |
| Visit to contact | Visitors but few enquiries | Offer clarity, contact options, forms |
| Contact to quote | Enquiries that go nowhere | Response time, follow-up, qualification |
| Quote to customer | Quotes not accepted | Quote format, pricing presentation, follow-up |
| Customer to repeat | One-off customers | Reminders, referrals, packages |
See what counts as a good conversion rate for how to read the visit-to-contact step.
Writing a testable idea
Vague ideas produce vague results. Write every idea in the same format before you start, so you know what you are testing and what would count as success.
- Change: exactly what will be different, for example replacing the contact form with a bookable fifteen-minute call.
- Expected effect: which number should move, for example booked conversations per week.
- Reason: why you believe it, for example several customers said they prefer to talk before sending details.
- Measure: where the number comes from, and that it is tracked before you start.
- Effort: hours needed, so you can prioritise cheap ideas first.
Score ideas roughly on expected impact, confidence and effort, and start with those that are cheap and plausible. Measurement must be in place first; our GA4 setup guide covers the basics.
One experiment at a time
With little traffic, you cannot split visitors into groups and compare them statistically the way large sites do. Instead you compare periods: before and after the change. That only works if you change one thing at a time on the step you are measuring, and keep everything else as stable as you can.
- Run each experiment for a fixed period, often two to six weeks depending on volume.
- Avoid starting experiments in periods that are unusual anyway, such as the Swedish summer holiday or just before Christmas, unless that is what you are testing.
- Note anything external that happened: a campaign, a competitor closing, press coverage.
- Different steps can have experiments running in parallel if they do not affect each other.
Stop rules and small numbers
Small numbers swing a lot by chance. Going from four enquiries a week to six can be luck. Decide before starting what result will make you keep, drop or extend the change, and stick to it.
| Result after the period | Decision |
|---|---|
| Clear improvement, and customers' comments point the same way | Keep the change and move to the next idea |
| No visible difference | Keep it only if it is simpler or cheaper; otherwise revert |
| Worse results | Revert, and note what you learned |
| Too few events to tell | Extend once, or measure an earlier step with more volume |
When numbers are small, qualitative signals matter more: what customers say on calls, questions that stop coming, objections that change. Write them in the log next to the numbers.
Keeping an experiment log
- One row per experiment in a shared spreadsheet: date, step, idea, change, measure, result, decision.
- A link or screenshot of what the page, email or offer looked like before and after.
- A short sentence on what you learned, even when the answer was no.
- A monthly half-hour review of the log to choose the next experiments.
Decision rule: if you cannot name the constraint and measure it, fix measurement before experimenting. When you do not need outside help: a founder who spends two hours a week on this and keeps the log will learn a lot on their own. If you want someone to run experiments, measurement and reporting with you each month, that is what our growth retainer does; book a short call to see whether it fits your stage.
Frequently asked questions
Can I run A/B tests with little traffic?
True A/B tests need enough visitors to separate a real effect from chance, which most small sites do not have. Before-and-after comparisons with one change at a time, plus customer feedback, are more practical.
How many experiments should a small team run per month?
One or two per bottleneck is plenty. The limit is usually the time to set them up properly and wait for results, not the number of ideas.
What if an experiment fails?
Log it and revert. A failed experiment that is written down is still progress, because it stops you trying the same idea again in six months.
Which tools do we need?
Analytics with working conversion tracking, a spreadsheet for the log, and whatever tool you already use for the website. More tools rarely help a small team.
Not sure where your growth is stuck?
In a free 15-minute call we map your customer path, point out the most likely constraint and suggest the first two experiments worth running.
Book a free 15-minute call