framework · metrics
AARM framework: measure product success without the vanity trap
Best for: "How would you measure success for X?" questions at Google, Meta, and Amazon
AARM is Lewis Lin’s answer to a specific gap in AARRR: Referral is a downstream effect, not a PM-controllable metric in most product contexts, and “Revenue” is ambiguous enough that candidates routinely cite total revenue rather than per-user economics. Lin stripped the framework to four stages, Acquisition, Activation, Retention, Monetization, introduced in his 2013 book Decode and Conquer. The key insight is not the acronym but the distinction between lazy registration and real activation. Most candidates skip that distinction entirely.
The four stages
Acquisition. Count new users who complete a defined entry action, not raw downloads or page visits. Downloads include churned installs. The relevant number is lazy registration: installing and granting permissions, or creating an account. This is the denominator that the activation rate divides against.
Activation. This is the stage most often missed or mis-defined. Activation is the first moment the user gets real value. Lin’s own example: for Google+, activation is uploading a profile photo or completing a profile, not creating an account. Account creation is lazy registration; profile completion is activation, because without it the social graph is empty and the product cannot work. You must define a specific behavioral event. “User engages with the product” is not an activation definition. “User completes their first playlist and plays it for more than 30 seconds” is. Track the percentage of acquired users who reach that event within 7 days. Below 20% means the funnel is broken before retention is relevant.
Retention. Track D7, D30, and weekly active rate. Also check depth: are retained users hitting the core loop, or just passively browsing? A user who logs in weekly but never uses the feature they activated for is not retained in any meaningful sense.
Monetization. Lin specifies ARPU as the primary metric. For freemium, also track free-to-paid conversion rate. For B2B, track net revenue retention: does the account expand at renewal or churn? Even when a product does not yet monetize, candidates should name this stage and explain why it is or is not the current priority. Skipping it signals a candidate who does not think about viability.
AARM vs AARRR: two deliberate changes
AARRR (Dave McClure, 2007) has five stages. Lin made two changes: he dropped Referral because it is a downstream amplifier, not something a PM can tune directly the way you can tune onboarding; and he renamed Revenue to Monetization to push candidates toward per-user economics (ARPU, conversion rate) rather than total revenue, which tells you nothing about unit economics.
Worked example: Spotify
- Acquisition. Users who install and complete account setup. Not App Store installs.
- Activation. User searches for a song, plays it for more than 60 seconds, and saves it or adds it to a playlist. This signals they connected the product to a specific listening intent.
- Retention. D7 and D30 active rates, plus weekly listening hours per retained user to distinguish daily commuters from occasional listeners.
- Monetization. ARPU segmented by market (ad-supported vs. paid tiers differ sharply), and free-to-paid conversion rate.
For a mature product like Spotify, the priority is retention depth and monetization. An early-stage product inverts that: activation is the constraint because without a high activation rate, retention data is meaningless.
strong
"I'd use AARM to structure this. Acquisition: users who complete app install and account setup, not downloads, because downloads inflate every downstream rate. Activation: the first moment of real value for this product is [specific event, e.g. completing their first AI-generated summary and acting on it]. I'd track the percentage of acquired users who hit it within 7 days. If that rate is below 20%, acquisition spend is irrelevant until we fix onboarding. Retention: D7 and D30, but also depth: are retained users completing the core loop or just browsing? For an AI product, a user who returns but never completes a task is not retained in a way that predicts revenue. Monetization: ARPU and free-to-paid conversion rate. Given this product is early-stage, I'd prioritize activation: if users are not reaching the magic moment, no acquisition spend fixes retention, and monetization follows from retention."
weak
"I'd measure acquisition with signups, retention with DAU, and monetization with revenue." This fails for four reasons: it skips activation entirely; it uses vanity metrics without anchoring them to specific user behavior; it doesn't connect any metric to the product's job-to-be-done; and it lists metrics rather than diagnosing where the lifecycle breaks. Interviewers conclude the candidate has never debugged a broken funnel.
The 2026 angle: trust is the new activation
For AI-native products, activation is the hardest stage because trust is now the magic moment. A user who runs one query and gets a plausible-sounding wrong answer has technically “activated” by old definitions but has not experienced real value. Interviewers at AI companies now expect candidates to define activation as the moment a user completes a task they could not complete before, or delegates a workflow they previously handled manually.
Retention splits into session-based (did the user return?) and workflow-embedded (is the AI now a step in their daily process?). The latter predicts viable businesses. Monetization has also shifted: consumption-based pricing (tokens, tasks) makes ARPU dynamic. The cleaner signal is cost-per-outcome versus revenue-per-user. Feasibility is no longer the constraint. Any team can ship a working product. Viable and lovable are where the bar sits, and AARM applied with specificity is the diagnostic for whether you are clearing it.