Growth Campaign

Origin & evolution

The growth campaign is the operational unit of growth hacking, a discipline Sean Ellis named around 2010 after building the early growth engines at Dropbox, LogMeIn, and Eventbrite (Lenny's Newsletter). Ellis reframed growth as a high-tempo loop of ideas, prioritisation, tests, and learning🧠LearningValidationAn insight gained from an experimentView reference →, codified later in Hacking Growth (2017, with Morgan Brown). A campaign in this world is a batch of experiments run against a single objective🏁ObjectiveStrategyA strategic goal (OKR)View reference →, not a creative artefact shipped to an audience.

Prioritisation is what keeps the tempo honest. Ellis created the ICE model, scoring each idea on impact, confidence, and ease so a team can rank a backlog quickly with a "good enough" accuracy (airfocus). Intercom later extended it to RICE by adding reach, weighting impact by how many users an idea touches. A growth campaign is, in practice, the highest-scoring cluster of ICE or RICE bets bundled for one cycle.

The frame then widened from tactics to systems. Brian Balfour and Reforge argued that durable growth comes from growth loops♾️Growth LoopGrowthA self-reinforcing growth cycleView reference →, reinvesting outputs to drive the next round of inputs, rather than a linear sequence of one-off pushes (Reforge). A modern growth campaign is read against the loop it feeds: a test of one mechanism inside a self-reinforcing system.

How it works in practice

A consumer-app growth team wants to lift week-one retention. They open a campaign with a single objective, retained users at day seven, and a six-week window. From a backlog of fifteen ideas they ICE-score the top three: a personalised onboarding checklist (impact 8, confidence 6, ease 7), a day-two re-engagement push (7, 7, 8), and a streak mechanic (9, 4, 3).

The re-engagement push scores highest on ease and confidence, so it runs first as an A/B test against a held-back control, targeted only at the segment of users who installed but never returned on day two. After three weeks the treatment lifts day-seven retention from 22 to 26 percent on that segment, significant at the campaign's threshold. The streak mechanic, lower confidence, gets staged next rather than shipped on faith. The campaign closes with one validated lever and a clearer read on the next.

Growth campaign vs. its neighbours

• Marketing campaign broadcasts a message to an audience on a schedule, judged on reach and brand lift. A growth campaign runs controlled tests against a product metric, judged on conversion and retained behaviour. One persuades; the other learns.
• Experiment plan📐Experiment PlanValidationAn experiment plan describing the hypothesis, setup, success criteria, and methodology before a test runs.View reference → is the design document for how a test will be run: hypothesis, variants🆎VariantGrowthA variant in an A/B testView reference →, sample size, success metric. A growth campaign is the wrapper that may hold several such plans aimed at one objective. The plan is the method; the campaign is the mission🧭MissionStrategyThe purpose and reason the product existsView reference →.
• Behavioral segment🐾Behavioral SegmentGrowthA user segment based on behaviourView reference → is the population a campaign targets, defined by what users do rather than who they are. The campaign acts on a segment; the segment is the where, the campaign the what and why.

In the graph

In the Unified Product Graph, a growth campaign is a container in the Business, GTM & Growth region. An acquisition channel🎣Acquisition ChannelGrowthA channel through which users are acquiredView reference → reaches it via acquisition_channel_runs_growth_campaignAcquisition ChannelrunsGrowth Campaignhierarchy, locating where the push happens. Its work hangs off three edges: growth_campaign_tests_variantGrowth CampaigntestsVarianthierarchy for the A/B arms it runs, growth_campaign_tests_via_experiment_planGrowth Campaigntests viaExperiment Planhierarchy for the formal design behind those arms, and growth_campaign_targets_behavioral_segmentGrowth CampaigntargetsBehavioral Segmentcross-domain for the audience. Modelling the campaign as a container, with the plan and variants beneath it, keeps the experiment-driven structure visible: a growth campaign that connects to no variant and no plan is a marketing push wearing a growth label, and the graph shows the difference.