Growth Hacking: What It Means and a Small-Scale Experiment Loop
Growth hacking is a way of working in which a small team runs many quick, measured experiments across the customer journey, from first visit to repeat use, and keeps what moves a number. The term is usually credited to Sean Ellis around 2010. In practice it is a loop of ideas, tests and decisions, and it does not mean tricks, shortcuts or spam.
What does growth hacking actually mean?
Strip away the reputation, and it is a discipline of testing. You pick one number you want to improve, list ideas that might move it, rank them, test the best one quickly, and decide whether to keep it. The word “hacking” points to speed and improvisation across roles, so a growth person might change a headline, adjust an onboarding step and write an outreach message in the same week.
It differs from marketing in scope and from engineering in purpose. Marketing often sits at the top of the funnel, while growth work follows the user through activation, retention and referral as well. The common thread is that every idea is a hypothesis with a number attached, not a belief.
What is the experiment loop?
- Choose the number. One metric for the period, such as the visitor-to-signup rate on your landing page.
- List ideas. Ten, without judging them. Headline, hero image, form length, a different community, a different first email.
- Score them. A common method is ICE: rate each idea from 1 to 10 for impact, confidence and ease, then average the three, or multiply them, but pick one rule and keep it.
- Run one. Change a single variable, note the start date, and decide in advance how many visits or days you will wait.
- Read and record. Keep, drop or retest, and write down what you learned even when it failed.
| Idea | Impact | Confidence | Ease | Average |
|---|---|---|---|---|
| Rewrite the headline in the customer’s own words | 7 | 6 | 9 | 7.3 |
| Add a 30-second demo clip above the fold | 8 | 5 | 4 | 5.7 |
| Post in one niche community with a tagged link | 6 | 6 | 8 | 6.7 |
| Redesign the whole page | 7 | 3 | 2 | 4.0 |
The scores are opinions, which is the point of writing them down: you can argue with the numbers and revise them after each test. Here the headline rewrite goes first because it is cheap and probably useful, and the redesign goes last.
How do you run experiments when traffic is tiny?
Classic split tests need many visits per variation before a difference is trustworthy, and a new product rarely has them. Adapt in three ways. Favour big changes over small ones, since a rewritten promise can shift behaviour visibly while a button shade cannot. Use qualitative checks, such as five people thinking aloud on a call, which show you why a page fails faster than any chart. And compare time periods only when the source of traffic is the same, using UTM parameters to keep the groups clean.
Write the decision rule before you start: “If at least 8 of the next 100 visitors sign up, keep it; otherwise revert.” The threshold is arbitrary, but fixing it in advance stops you from stretching a weak result into a win. Report the counts as well as the rate, which the conversion rate entry explains.
Which “growth hacks” should a small maker leave alone?
- Fake votes, bought accounts or vote rings. Launch boards commonly ban them and may remove a listing, so read the rules first. Real ways to earn votes are in how to get upvotes without spam.
- Mass unsolicited messages. They damage your name before you have one, and communities ban the account.
- Fake testimonials or invented user counts. Visitors check, and the damage lasts longer than the lift.
- Dark patterns. Hidden costs, forced shares and guilt-trip buttons raise a number once and cost trust permanently.
- Copying another product’s tactic without its audience. A trick that worked for a consumer app means little for a developer tool.
Honest experiments are less dramatic and more reliable: a clearer headline, a faster first session, a better reply on a maker comment, a post in the right community. A product that people want responds to these. A product that nobody wants does not respond to tricks either.
Where should a launch-stage maker start?
With one number and a log. Pick the step where most people drop off, usually between a visit and a signup or between a signup and first use, and run one experiment on it each week. After a month you will have four small facts about your audience, which is more than most launches leave behind. The launch metrics to track list helps you choose the number, and getting your first hundred users shows the channels to test.
When you launch your product free, the launch listing is your first experiment: one headline, one audience, one tagged link. Record what happens and run the next one. For inspiration on the sort of problems people try to solve with small products, browse the social app ideas on the board.
Frequently asked questions
Who coined the term growth hacking?
It is usually credited to Sean Ellis, who used the phrase growth hacker around 2010 to describe someone whose main job is finding ways to grow a product. The idea predates the name in practice, and today it is used loosely, which is why it helps to define it as a loop of measured experiments.
Is growth hacking just marketing?
It overlaps with marketing but is wider. Growth work follows the user from first visit through activation, retention and referral, and it relies on measured experiments. A marketer may spend most of a day on top-of-funnel channels, while a growth practitioner might change onboarding or pricing in the same week.
What is ICE scoring?
ICE is a quick way to rank experiments by rating each idea from 1 to 10 for impact, confidence and ease, then averaging or multiplying the three. It helps you pick what to try first. The scores are opinions, so revise them after each test as you learn how accurate your guesses were.
Can a solo maker do growth hacking?
Yes. The loop needs only one number, a list of ideas and a log. With little traffic, favour large changes, use direct conversations to learn why a page fails, and decide in advance what result would make you keep a change. Skip tactics that break platform rules or mislead people.
Written by Post your build ideas with AI assistance and edited for this site. Where a page names another product, the facts came from that product’s own pages on the date shown above; terms change, so check them before you rely on them. About the publisher.