Building a Growth Hacking Strategy From Scratch

Sean Ellis coined the term “growth hacker” in a 2010 blog post arguing that startups needed something different from a traditional VP of marketing: someone whose only job was growth, measured relentlessly, rather than brand awareness or campaign output. Andrew Chen’s widely shared 2012 essay pushed the idea further into the mainstream, and by now most of the tactics associated with it, Dropbox’s referral program, Airbnb’s early integration with Craigslist listings, PayPal paying users to refer friends, have become industry folklore. What rarely gets built alongside the folklore is the actual process behind it. Here’s how to construct one from nothing.

Step 1: Pick one number that matters

Before running a single test, decide on a single metric that best represents real growth for the business, sometimes called a north star metric. For a marketplace it might be completed transactions; for a content product it might be weekly active users who return without a prompt. The point of choosing one number is discipline: every experiment gets judged against whether it moved that number, not whether it generated buzz, traffic or vanity engagement that doesn’t translate into it.

Step 2: Map the funnel with AARRR

Dave McClure’s AARRR framework, often called pirate metrics for how it reads aloud, breaks growth into five stages: Acquisition (how people find the product), Activation (whether they have a first good experience), Retention (whether they come back), Referral (whether they tell others) and Revenue (whether any of this makes money). Walk through actual data for each stage and find where the biggest drop-off happens. Most businesses assume their problem is acquisition, not enough visitors, when the real leak is often activation or retention, in which case pouring more people into the top of the funnel just wastes money on a funnel that already loses most of what it has.

Step 3: Build a backlog of testable ideas

Once the weakest stage is identified, generate specific, falsifiable ideas to fix it rather than vague ones. “Improve onboarding” isn’t testable. “Reduce signup from five fields to two” is. Pull ideas from wherever they come: competitor research, direct customer interviews, support tickets that reveal friction and the well-known playbook of prior growth hacks adapted to the specific product. Dropbox’s referral program worked because it gave both the referrer and the new user extra storage, a real, felt benefit tied directly to the product itself rather than a generic discount, which is the pattern worth studying, not the specific mechanic.

Step 4: Prioritize by impact, confidence and ease

A backlog of ten or twenty ideas is only useful if the team can decide what to run first. A simple scoring approach, often called ICE, rates each idea on its potential Impact, the team’s Confidence it will work and how Easy it is to build and ship. Score each idea roughly on a scale, sort by the total and run the highest-scoring, cheapest tests first. This keeps a small team from burning weeks on an ambitious idea with uncertain payoff before testing several cheap ones that might move the number just as much.

Step 5: Run fast, measure honestly, kill what doesn’t work

Speed is the actual advantage growth hacking claims over traditional marketing planning cycles. A test should be built to answer one question as cheaply as possible, not to be a polished feature. Ship it, measure it against the north star metric from step one and be willing to kill it fast if the data doesn’t support it. The habit that separates functioning growth teams from ones that stall is refusing to let ego or sunk cost keep a failed experiment alive past its verdict.

Step 6: Turn winners into loops, not campaigns

The best growth hacks aren’t one-time promotions, they’re mechanisms built into the product that keep generating new users on their own. PayPal’s early referral bonus and Dropbox’s storage-for-invites program both worked this way: using the product naturally created an incentive to bring in the next user, so growth didn’t require constantly refilling an ad budget. When a test succeeds, the next question isn’t just “should we keep this,” it’s “can this become a loop that runs itself.”

Step 7: Make it a weekly habit, not a project

Growth hacking stops being a phase and becomes a real capability when a team reviews the north star metric, the current backlog and the latest test results on a fixed weekly cadence, the way GrowthHackers.com and similar early communities institutionalized the practice. A single successful campaign proves an idea works once. A standing process that keeps generating, testing and scaling new ideas is what compounds, and that discipline, not any individual clever trick, is the real difference between a company that grew this way and one that just got lucky once.