Growth hacking described a specific thing when the phrase was coined: rapid, cheap experiments run across product and marketing together, by somebody with the access to change both. Most agencies now using the phrase sell one of three quite different services under it, and the useful first step is working out which, because only one of them is what the phrase originally meant and it is the one that requires something most clients cannot give an agency.
The three things sold under one name
First, experimentation across acquisition channels, which is competent paid and organic marketing with a testing discipline. Second, conversion rate optimisation on the website, which is a real and narrower service. Third, product-led growth work touching onboarding, activation and referral inside the product itself. The third is what the phrase originally meant, and it needs engineering access most agencies never get. Ask which of the three is being proposed.
Experiments need a rate, a baseline and somewhere to write them down
The discipline that makes experimentation valuable is boring: a stated hypothesis, a defined success measure, a decision rule agreed before the result, and a record of what was tried and what happened. Without that, an experimentation program becomes a series of changes with a narrative attached afterwards. Ask to see the experiment log from an existing client, anonymised. Its absence is the finding.
Most sites do not have the traffic for the method
Testing needs enough volume to distinguish a real effect from noise, and a great many businesses buying this service do not have it on the pages that matter. Where volume is low, the honest approach is qualitative research and sequential changes judged on judgement rather than statistics, which is a perfectly good way to work and is not what is usually sold. An agency that proposes multivariate testing on a page with a handful of conversions a week has not looked at your numbers.
The tactics that gave the phrase its name have mostly closed
The original examples relied on platform loopholes, scraped contact data and automation that platforms now prohibit and in some cases that the law reaches. Anything a supplier proposes involving scraping personal data, automating actions on platforms against their terms, or contacting people who never gave you their details, is a risk transferred to you. Ask what the agency refuses to do and listen for specifics.
Judge it on what was learned, not on what was launched
A quarter of experimentation that produced three clear negative results and one modest positive is a good quarter, because you now know things you did not. A quarter that produced twenty launches and no conclusions is not, however busy it looked. Agree at the start that the report is a list of hypotheses with outcomes, which changes what gets proposed as much as how it gets judged.
Questions people ask about growth hacking agency
Is growth hacking still a real discipline?
The experimentation discipline is real and valuable. The phrase has drifted to cover ordinary performance marketing and conversion work, and several of the original tactics are now prohibited by platforms or reached by privacy law. Judge the specific services proposed rather than the label.
What does a growth agency cost?
Usually a retainer, sometimes with an experiment volume attached. Published floors across the agencies in this index start in the hundreds a month and run to five figures. The variable that matters is how much of the fee is senior analytical time rather than execution, because the value here is in the judgement about what to test.
Do we have enough traffic for experimentation?
Ask the agency to work it out from your own numbers before you buy: how many conversions a week on the pages in question, and therefore how long a test would take to reach a conclusion. If the honest answer is months, the right program is qualitative research and sequential judgement rather than testing.
Can an agency do product-led growth work?
Only with real access to the product team and the ability to ship changes, which most agencies never get. Without it, what gets delivered is recommendations that sit in a document. If product-led growth is what you need, the question is whether engineering capacity exists for it, not which agency to hire.