TL;DR
- The Trap: Growth hacking is a bag of tactics copied from someone else’s context. Tactics decay. Systems compound.
- The Difference: A hack optimizes one metric once. A growth loop feeds itself, so every turn makes the next turn cheaper.
- The Loop: One channel, one metric, one input that compounds. Test weekly, kill the flat experiments fast, reinvest in what repeats.
- The Tell: If growth stops the moment you stop pushing, you have a campaign. If it keeps spending effort for you, you have a loop.
- The Start: Prove one loop repeats for 30 days before you add a second. Width without a working loop just multiplies the noise.
Every growth team I audit keeps a graveyard of hacks. A referral widget that spiked signups for two weeks. A viral waitlist that produced a spike and a retention cliff. A pricing tweak that lifted a metric until the novelty wore off. The work was real. The growth was not durable. That gap is the entire problem with how most teams practice growth hacking.
Growth hacking started as a genuinely good idea: bring the speed and rigor of engineering to marketing. Somewhere along the way it collapsed into a hunt for tricks. A hack is a tactic lifted from someone else’s context and applied to yours, usually without the loop that made it work for them in the first place. It can produce a spike. It cannot produce a slope.
The teams that keep compounding do not run more hacks. They build growth loops: systems where the output of one turn becomes the input to the next. That is the difference between a campaign and a machine. Most teams are running campaigns and calling it growth.
The Growth Hack Trap
A hack is designed to win a screenshot, not a year. It is optimized for the demo: the lift you can show in a standup, the test that ships before the quarter closes. That is why hacks cluster around the metrics that move fastest, traffic, clicks, signups, and almost never around the metrics that compound, activation, retention, referral, and revenue per account.
The deeper problem is borrowed context. A tactic that worked for a consumer app with viral sharing does not transfer to a B2B product sold by a sales team. When you copy the hack without the surrounding system, you copy the visible move and miss the invisible loop that gave it power. The tactic fails, and the team concludes that growth hacking “does not work here” instead of concluding that they borrowed the wrong part.
A growth hack is a tactic without a loop. It can spike a metric once, but nothing about it makes the next spike cheaper. If the second attempt costs the same as the first, you have a tactic, not a growth engine.
Why Hacks Do Not Compound
Compounding is the only thing that separates a spike from growth. A return that repeats and feeds itself beats a bigger one-time win every time, because the loop gets cheaper to run as it scales while the hack gets more expensive. The economics make the case before the tactic does: acquiring a new customer costs a multiple of what it costs to keep one, and a small lift in retention moves profit far more than the same lift in acquisition, per Harvard Business Review. Loops live on the retention and referral side of that equation. Hacks usually chase the acquisition side, which is the most expensive and least durable place to spend effort.
There is a simpler test. Ask what happens when you stop pushing. A campaign stops the moment the spend stops. A loop keeps turning because each turn supplies the fuel for the next. Here is how the two compare in practice.
| Dimension | The hack | The loop |
|---|---|---|
| Optimises for | The screenshot | The compounding slope |
| Cost over time | Rises with scale | Falls with scale |
| When you stop pushing | Growth stops | Growth keeps turning |
| Transferability | Borrowed from other contexts | Built for your context |
| Metric it feeds | Usually acquisition | Retention, referral, revenue |
| Half-life | Weeks to a quarter | Years, until the channel saturates |
None of this means speed is wrong. It means speed pointed at the wrong thing just gets you to a dead end faster. The fix is not to test less. It is to test inside a loop, so every experiment either strengthens the loop or gets killed before it becomes a distraction.
The Growth Loop Framework
A growth loop is not a funnel. A funnel is linear and leaks; you pour in at the top and hope some fraction falls out the bottom. A loop is circular and self-feeding; the output of one turn becomes the input of the next. Four jobs make one work.
Not a dashboard. One number that, if it moved, would change how the business spends money. Usually it is a revenue-adjacent metric, not a vanity one: activated accounts, expansion revenue, qualified conversations per week. If you cannot say which single number the loop exists to move, you are optimizing a hobby.
Growth comes from dominating one channel before diluting effort across five. Saturation means you have exhausted the easy wins in a channel: the audience knows your name, the format is refined, the response rate has plateaued at a good level. Most teams never reach that point because they move to the next shiny channel at the first dip.
Every durable loop reinvests its own output. Content creates signals that create conversations that create proof that creates more content. Referrals create customers that create referrals. The input has to be something you produced on the last turn, otherwise you are not in a loop, you are in a treadmill.
Measure two things: how long one full turn takes, and how much it costs to complete. The goal of every subsequent experiment is to shorten the turn or lower its cost, not to find a new loop. A loop that turns monthly and cheaply beats a loop that turns daily and expensively, because the cheap one compounds without you.
What I Actually Think
I have run growth on both sides of this line, and the asymmetry is not subtle. The hacks get the applause. The loops get the revenue. The best “growth hack” I ever ran looked boring from the outside: one channel, one format, one weekly cadence, measured for a year. It never spiked. It just kept going up, because each post created the audience for the next one. The tricks I tried around it came and went. The loop is still paying.
Nobody shares a growth loop on a slide. That is why hacks get all the attention and loops get all the revenue. Build the boring thing that compounds.
Your First 30 Days
You do not need a growth team, a new tool, or a bigger budget to start. You need one loop that works end to end. This is the sequence I would run in the first month.
List every growth activity running right now. For each, ask the tell question: if we stop pushing, does it keep producing? Mark the ones that do. That short list is where your compounding lives, and it is usually shorter than anyone expects.
Every experiment that has not moved the one metric in a full cycle gets cut. This is the hard week, because these are often the most fun activities. Killing them frees the effort that the loop needs. Be ruthless: a flat experiment is not neutral, it is a tax on the ones that work.
Take your one promising channel and wire the full turn: the output, the reinvestment, the measurement. Do not add a second loop yet. One loop running cleanly beats three running at half power, and a half-fed loop does not compound.
Compare the loop’s output this week to last week. If the turn is getting cheaper or faster, it is working and you scale it. If it is flat, you have not found a loop yet, and the honest move is to say so rather than dress up a tactic as a system.
Growth hacking works best when it stops being a hunt for hacks. If you want the channel discipline behind this, read why you should saturate one channel before adding another. For where the compounding input comes from, see how to build a first-party signal engine. And to keep the loop honest, start with why most teams measure output instead of impact.
The concept is not new, which is exactly why it is easy to ignore. Growth teams have been arguing for loops over funnels for years, and the clearest framing is still Brian Balfour’s “growth loops are the new funnels” over at Reforge. The other half of the argument is about decay: Andrew Chen’s “law of shitty clickthroughs” explains why every tactic gets less effective the more it is used, at Andrew Chen’s blog. Put the two together and the conclusion is unavoidable: tactics decay, so the only durable growth is the kind that reinvests in itself before the tactics run dry.
Stop collecting hacks. Build one loop: one channel, one metric, one input that compounds, then shorten the turn every week. A loop that feeds itself will beat a bag of tricks every time, because it is the only thing on the list that gets better without you pushing harder.
