Most growth hacking articles are a museum tour
Dropbox's referral programme. Hotmail's email signature. Airbnb's Craigslist integration. PayPal paying users to sign up.
Every guide on this topic tells the same four stories, and all four are between fifteen and twenty-five years old. They happened in conditions that no longer exist: an unsaturated referral market, an open Craigslist API, a venture budget for paying strangers to register. They are worth knowing as history and they are not a plan.
This page separates the two. Section one is the strategies that still work in 2026, ranked by what they need from you. Section two is the famous case studies, and why each one cannot be repeated. If you only read one, read the first.
Methodology: ranked by return against the two resources that actually constrain a growth team, which are engineering time and money. Each strategy states which of the two it needs. Where a widely repeated tactic has stopped working, it is in the "stopped working" section rather than quietly omitted.
What growth hacking actually is
A testing loop, not a bag of tricks. You find the one constraint holding growth back, design the cheapest possible test against it, run it, and keep or kill it on the evidence.
The reason the discipline gets a bad name is that most people copy the tactic that worked for somebody else, which skips the first step entirely. A referral programme is not a growth hack if referral was never your constraint.
The useful frame is the AARRR funnel: acquisition, activation, retention, referral, revenue. Find your worst stage, work only on that one, and ignore everything else until it moves. We cover the fundamentals in what is growth hacking if you want the groundwork first.
The strategies compared
| # | Strategy | Needs engineering | Needs budget | Time to signal | Best stage |
|---|---|---|---|---|---|
| 1 | Fix activation before acquisition | No | No | 2 to 4 weeks | Activation |
| 2 | Free tool as an acquisition channel | Yes | Low | 8 to 16 weeks | Acquisition |
| 3 | Programmatic landing pages | Some | Low | 8 to 20 weeks | Acquisition |
| 4 | Onboarding to first value | Some | No | 2 to 6 weeks | Activation |
| 5 | Original data and research | No | Medium | 6 to 12 weeks | Acquisition |
| 6 | Integration and marketplace listings | Some | No | 4 to 12 weeks | Acquisition |
| 7 | Referral, done with real incentives | Yes | Medium | 6 to 12 weeks | Referral |
| 8 | Community as a retention engine | No | No | 12 to 24 weeks | Retention |
| 9 | Reverse trial | Some | No | 4 to 8 weeks | Activation |
| 10 | Pricing page experiments | No | No | 4 to 8 weeks | Revenue |
| 11 | Win-back and resurrection | No | No | 2 to 4 weeks | Retention |
| 12 | Being cited by AI answer engines | No | Low | 8 to 20 weeks | Acquisition |
Read the "needs engineering" column first. If you have no engineering time, five of these are closed to you, and the remaining seven are where your quarter should go.
1. Fix activation before you touch acquisition
The cheapest growth in almost every business is not more signups, it is more of your existing signups reaching the point where the product is useful.
Find your activation moment: the action that separates users who stay from users who leave. Then measure what share of signups reach it, and how long they take. If that share is under 40%, every dollar you spend on acquisition is leaking out of the same hole.
Why it ranks first: no engineering, no budget, and a measurable answer within a month. It is also the least glamorous item on this list, which is why it gets skipped for a referral programme that will not fix anything.
2. Build a free tool
A calculator, a checker, a generator that solves one small problem completely and asks for nothing. It acquires users through search, through sharing, and through links, and it keeps doing so after you stop working on it.
The rule: single purpose, instant result, no signup to see the answer. Gating the result behind an email destroys most of the acquisition value, because nobody links to or recommends a form.
The honest cost: this needs engineering time and it needs the tool to be genuinely good. A mediocre free tool acquires nobody.
3. Programmatic landing pages
Generate hundreds or thousands of pages from a structured data set, each targeting a specific long-tail query: one per city, per integration, per use case, per comparison.
This works when the underlying data is genuinely useful per page. It fails, badly and sometimes permanently, when the pages are the same paragraph with a variable swapped, which search engines now identify quickly.
The test before you build: would a human landing on page 400 find something on it they could not get from page 1? If not, you are building a liability.
4. Compress the time to first value
Distinct from activation rate. This is about how long it takes, not how many get there.
Remove setup steps, pre-populate with sample data, defer the parts of configuration that can wait. Every step between signup and the first useful moment loses a share of users, and the losses compound.
The highest-leverage version: find the single setup step with the biggest drop-off and remove it entirely, rather than improving all of them slightly.
5. Publish original data
Numbers nobody else has, from your own product data, a customer survey, or something public that nobody has counted. It earns links, it earns press, and it now earns citations in AI answers, which cite statistics far more readily than opinions.
The rule that decides whether it works: it must produce one sentence somebody else wants to quote. If the finding cannot be stated in a single line with a number in it, it will not travel.
6. Get listed where your customers already are
Integration directories, app marketplaces, template galleries, comparison sites. Each listing is a page ranking for intent you would otherwise pay for, on a domain far stronger than yours.
Deeply unglamorous and consistently underrated. For most B2B products this is the highest-return acquisition work available in a single quarter, and it needs no engineering beyond whatever integration you already built.
7. Referral, with an incentive that is actually worth something
Referral still works. What stopped working is the $10 credit, because everybody offers one and nobody cares.
Two conditions for a referral programme to do anything. The product must already have people who love it, which you can verify before building anything. And the incentive must be worth more than the social cost of the recommendation.
Do not build this if your activation rate is poor. You will pay to acquire users who churn, which is worse than not growing.
8. Build a community
Slow, unfashionable, and the strongest retention mechanism available to a small company. A place where customers talk to each other creates a reason to stay that has nothing to do with your feature set.
The honest timeline: six months before it feels alive, and it needs a real person hosting it. Do not start one you will not sustain, because an abandoned community is worse than none.
9. The reverse trial
Give everyone the full product for a fixed period, then drop them to the free tier rather than cutting them off. Users experience the value first and downgrade into a limitation they now understand.
This consistently outperforms both a feature-limited free tier and a hard trial expiry, and it is a pricing change rather than a product build.
10. Experiment on the pricing page
The highest-traffic, highest-intent page you own, and usually the least tested. Ordering, plan naming, annual default, the number of tiers, what the comparison table emphasises.
Small changes here move revenue immediately, with no acquisition cost. The constraint: you need enough traffic for significance. Under a few thousand pricing page visits a month, you will be reading noise, and the honest answer is to skip this until you have volume.
11. Resurrect the users you already lost
Your churned list is the cheapest audience you will ever have. They know what the product is, they wanted it once, and something specific stopped them.
Ask what changed rather than offering a discount. The replies are worth more than the recovered accounts, because they tell you which of the twelve strategies on this list you should actually be working on.
12. Get cited by AI answer engines
The newest item here and the one most guides have not caught up with. A growing share of buying research now happens inside ChatGPT, Perplexity and Gemini, and being named in those answers is a distribution channel with almost no competition yet.
What earns citations: direct answers near the top of a page, specific numbers, comparison content that names alternatives fairly, and pages structured so a single paragraph can be lifted and quoted. What does not: long throat-clearing before the answer.
Watch for the signature in your own data. A page with thousands of impressions, a decent position and a click-through rate near zero is usually a query being answered without the click, and the fix is to be quotable rather than to rewrite the title.
The famous case studies, and why you cannot copy them
Named specifically, because they appear in every article on this topic without the caveat.
- Dropbox's referral programme (2008). Worked because storage was scarce and expensive, and because referral incentives were novel. Both conditions are gone.
- Airbnb's Craigslist cross-posting (2010). Worked because Craigslist had an unprotected posting flow. It was shut down and would be a terms-of-service violation today.
- Hotmail's email signature (1996). Worked once, on a channel nobody had thought to use as a growth surface. Everyone has thought of it now.
- PayPal paying $10 per signup (2000). Worked because they had venture money and no competitors doing it. It is a customer acquisition cost strategy, not a growth hack.
- Gymshark's early influencer seeding (2012). Worked before influencer marketing had rate cards.
The pattern is worth extracting even though the tactics are not: each found a channel that existed but nobody was using as a growth surface. That is the transferable lesson. The transferable lesson is not "build a referral programme".
What has stopped working
- Growth teams with no engineering. Half the list above needs code. A growth function without it is a marketing function.
- Vanity A/B tests. Button colours on low-traffic pages produce noise dressed as insight.
- Scraping and cold-emailing at volume. Deliverability and regulation have both closed this down.
- Viral loops bolted onto products nobody loves. Virality amplifies whatever is already there, including indifference.
- Copying a competitor's funnel. You cannot see their retention numbers, which is where the funnel actually succeeds or fails.
How to run this as a loop
Week 1. Measure every stage of the funnel and find the worst one. Do not skip this because you already have an opinion about which it is.
Week 2. Pick one strategy from the table that targets that stage and that fits your engineering budget. One, not three.
Weeks 3 to 6. Build the smallest possible version and ship it.
Weeks 7 to 10. Measure against a control. Keep, iterate, or kill. Killing is the normal outcome and it is not a failure of the loop.
Then repeat. Four cycles a year, each one attacking a measured constraint, beats twenty tactics executed simultaneously on instinct. Our guide to funnel hacking covers the measurement side in more depth, and automated lead generation software covers tooling the acquisition stage.
Frequently asked questions
What is the best growth hacking strategy? The one that targets your worst funnel stage. For most companies that is activation rather than acquisition, which makes strategy 1 above the correct answer far more often than the interesting ones.
Does growth hacking still work in 2026? The method does. Most of the famous tactics do not, because they depended on conditions that no longer exist.
Do I need engineers to do growth hacking? For half of this list, yes. The other half, including the highest-ranked strategy here, needs none.
How long before a growth experiment shows results? Two to four weeks for activation and retention work, eight to twenty for anything acquisition-led through search. Judging a search-led experiment at four weeks is the most common way good ideas get killed early.
What is a good conversion rate to aim for? Aim for improvement against your own baseline. Cross-industry benchmarks are close to useless because they average across wildly different products and traffic sources.
How much should I spend on growth experiments? Enough to run the test properly and little enough that killing it is easy. If an experiment is too expensive to abandon, it is not an experiment.
Where to start this month
Measure your funnel and find the worst stage. Then pick one strategy from the table that targets it and that you can afford in engineering time.
For most companies reading this, that will be activation, and the work will be unglamorous: removing two setup steps and rewriting an onboarding email. It will also do more than a referral programme, and it can be done this month rather than next quarter. Our breakdown of lead generation cost is useful for pricing the alternative before you commit to acquisition spend.
If the acquisition stage is genuinely your constraint and content is the channel, Distribb runs the publishing and link side of strategies 3, 5 and 12 on a schedule, so the experiment runs long enough to produce a signal rather than stalling in month two.