Growth loop design

Growth Loop Design

Before starting

Confirm (ask or infer) before running:

• NSM status — has the North Star Metric been defined? (loops are designed to compound the NSM)
• Current acquisition channels — where do new users come from today? (determines whether a loop exists)
• Retention signal — do users who get value come back? (loops require retained users as inputs)
• Product type — does the product have a network effect, content output, or referral mechanic? (determines loop archetype)
• Growth motion — PLG / SLG / CLG? (each motion has characteristic loop types)

---

Contract

This skill guarantees:

• A loop diagnosis is performed before loop design (you may not have a loop, you may have a funnel)
• The k-factor (loop coefficient) is calculated or estimated before any loop investment is recommended
• Only loop archetypes appropriate to the product type and motion are recommended
• Loop improvement is prioritized at the weakest conversion step, not at the most visible step
• The output distinguishes between what the team can build now vs. what requires product changes

---

Role: Growth Systems Architect. Most growth teams have funnels. Funnels are linear — you pour users in the top, some convert, growth is proportional to top-of-funnel investment. Loops are compounding — outputs of one cycle become inputs to the next, and growth accelerates without proportional reinvestment. The difference in long-term growth rate is enormous. Your job is to determine which you have, identify which archetype applies, and improve the loop at its weakest point.

---

The core distinction

FUNNEL (linear):
  Acquisition → Activation → Retention → Revenue
  
  Growth rate is proportional to acquisition investment.
  If you stop investing in acquisition, growth stops.
  CAC must be paid for every new user.
  
  Typical growth curve: linear or sub-linear.

LOOP (compounding):
  Acquisition → [User gets value] → [User generates a new acquisition input]
             ↑_____________________________________________↓
  
  Outputs of one cycle feed the inputs of the next cycle.
  Each user acquired contributes to future acquisition.
  With a k-factor > 1, the loop is self-sustaining (true virality).
  With a k-factor < 1 but > 0, each user brings in a fraction of a new user.
  At scale, this creates meaningful compounding.
  
  Typical growth curve: exponential or super-linear.

THE DIAGNOSTIC QUESTION:
  "When a user gets value from the product, does that create inputs for
   acquiring the next user — without any additional marketing spend?"
  IF yes (and the mechanism is designed, not accidental) → you have a loop.
  IF no → you have a funnel; you may be able to build a loop on top of it.

---

Step 1 — Loop diagnosis

Run this before designing anything. Most teams think they have a loop when they have a funnel with some referrals.

LOOP PRESENCE TEST (run for each candidate mechanism)

Question 1: Is the output of one user's value moment an input to another user's acquisition?
  Example YES: "User creates a public dashboard → other users find it via search → they sign up"
  Example NO: "User gets value from the tool → happy user → they might tell someone someday"

Question 2: Is the mechanism observable and instrumentable?
  IF no → you have a hoped-for loop, not an actual one
  IF yes → proceed to measure it

Question 3: Is the loop step native to the product experience?
  IF the user has to leave the product to complete a loop step → the loop has high friction
  IF the loop step happens inside or through the product → low friction, higher coefficient

Question 4: Does the loop complete? (Does the referral actually result in a new user?)
  Trace the full cycle. Many "loops" have a step that is assumed but never verified.
  Example: "User shares content" → [is the content findable?] → [is it compelling to non-users?]
  → [do non-users click?] → [do they sign up?] — trace each step.

DIAGNOSIS OUTPUT:
  [ ] Full loop exists and is instrumented → proceed to Step 2 (measure)
  [ ] Full loop exists but is not instrumented → instrument first, then Step 2
  [ ] Partial loop (one step is missing or assumed) → identify the gap; design to close it
  [ ] No loop → select an archetype in Step 3 to build toward

---

Step 2 — Measure loop strength (k-factor)

K-FACTOR DEFINITION:

k = (invitations sent per user) × (conversion rate of invitations to new users)

If k = 0.3: every 10 users bring in 3 new users → loop supplements acquisition
If k = 0.7: every 10 users bring in 7 new users → loop is significant; meaningful compounding
If k = 1.0: every user brings in exactly one new user → perfect replacement growth
If k > 1.0: every user brings in more than one new user → truly viral; exponential growth

REALISTIC K-FACTOR BENCHMARKS:
  B2B SaaS (invite/share mechanics): 0.1–0.3
  B2B dev tools (content/community loop): 0.15–0.4
  Consumer social / collaboration: 0.3–0.8
  Marketplace (supply creates demand): 0.4–0.9
  True viral consumer apps: 0.7–1.5+

WHY K > 1 IS RARE AND FRAGILE:
  Viral products at k > 1 grow explosively but also collapse quickly when k drops
  below 1 (audience exhaustion, feature changes, platform policy shifts).
  B2B products rarely sustain k > 1. Target k = 0.3–0.7 as a sustainable range
  for compounding growth that doesn't depend on virality.

HOW TO MEASURE K:
  Step 1: Count invitations/shares generated per cohort of 100 users in 30 days
  Step 2: Count how many of those resulted in new signups
  Step 3: k = (new signups from loop) / (100 users in cohort)
  
  IF shares/invitations are not tracked → instrument first.
  You cannot improve a k-factor you cannot measure.

LOOP CYCLE TIME:
  The time between a user getting value and the loop completing (new user acquired).
  
  Short cycle (hours to days): viral / product loop
  Medium cycle (days to weeks): content loop, referral loop
  Long cycle (weeks to months): community loop, sales-led referral
  
  Shorter cycle time amplifies k-factor impact. The same k = 0.4 loop with a
  7-day cycle is far more powerful than the same k with a 60-day cycle.

---

Step 3 — Loop archetype selection

Five archetypes. Select the one(s) that fit your product type. Do not design a loop that requires users to do something unnatural.

ARCHETYPE 1 — VIRAL / INVITE LOOP
How it works: user gets value → invites others to collaborate or see the output → invitees sign up
Classic example: collaborative documents, project management, multiplayer products

Requires:
  - Value is better with multiple people (network effect inside the product)
  - Invitation is natural to the core workflow (not a "refer a friend" button)
  - Invitees experience value before being asked to create an account

Design requirements:
  - Share/invite step is native to the product action (not a modal after the fact)
  - The shared output is useful to the recipient even without an account
  - Friction from "received invite" to "created account" is minimal (< 3 steps)

IF this archetype fits: instrument invite-to-signup rate as the primary loop metric
TYPICAL K-FACTOR RANGE: 0.3–0.8 (B2B); higher for consumer

---

ARCHETYPE 2 — CONTENT / SEO LOOP
How it works: user creates content (or the product generates it) → content is indexed publicly → new users
discover via search or social → they sign up to create their own content

Classic example: resume builders, portfolio tools, public dashboards, API documentation sites

Requires:
  - Users naturally produce content as part of using the product
  - That content is publicly accessible and indexable (not behind a login wall)
  - The content is valuable enough that non-users want to see it

Design requirements:
  - Public-by-default (or easy to make public) output pages
  - SEO-friendly URLs and metadata on user-generated content
  - Clear CTA on public content pages: "Create your own [thing] for free"

IF this archetype fits: track organic search traffic → signup rate as the loop metric
TYPICAL K-FACTOR RANGE: 0.15–0.4 (long cycle, but very durable)

---

ARCHETYPE 3 — PRODUCT / PLG LOOP
How it works: product delivers value → user shares usage data, results, or outcomes with their team →
team members sign up to use the same product

Classic example: analytics tools, developer tools with shareable outputs, data products

Requires:
  - Output is something worth sharing (result, report, analysis, code)
  - The recipient has a reason to want their own account (not just to read the output)
  - Sharing is a natural step at the end of the core workflow

Design requirements:
  - "Share" is the last step of the core workflow, not an afterthought
  - Shared outputs include implicit or explicit CTAs for the recipient to sign up
  - Free tier or trial available for new users arriving from a shared output

IF this archetype fits: track shares-per-active-user and share-to-signup rate
TYPICAL K-FACTOR RANGE: 0.2–0.5

---

ARCHETYPE 4 — COMMUNITY LOOP
How it works: users build expertise in the community → community content ranks in search → new users
discover via content → join community → become users → contribute content

Classic example: OSS tools, developer platforms, forum-driven products

Requires:
  - Community produces content of genuine value to non-members
  - Content is publicly indexed (Stack Overflow-style, not Discord-only)
  - A path exists from "content consumer" to "product user"

Design requirements:
  - Community content is findable via search (not locked in Discord or Slack)
  - Strong "getting started" path for users arriving from community content
  - Contributor recognition that reinforces community participation

IF this archetype fits: track community-content-search-to-signup as the loop metric
TYPICAL K-FACTOR RANGE: 0.1–0.3 (long cycle; compounds over months, not days)

---

ARCHETYPE 5 — SUPPLY / DEMAND LOOP (marketplace)
How it works: more supply attracts more demand → more demand attracts more supply →
both sides grow through the other

Classic example: marketplaces, two-sided platforms

Requires:
  - Two distinct user types with complementary needs
  - Network value increases for both sides as the other side grows
  - A mechanism to incentivize early supply-side participation

Design requirements:
  - Solve the supply-side first (empty marketplace has no demand)
  - Geographic or vertical segmentation to concentrate supply and demand in the
    same market before expanding (thin supply across 50 markets = zero value)

IF this archetype fits: track supply density and demand-to-supply ratio by market segment
TYPICAL K-FACTOR RANGE: 0.4–0.9 in active segments

---

Step 4 — Loop improvement (find the weakest step)

The loop is as strong as its weakest conversion step. Improving the strongest step first is the most common loop optimization mistake.

LOOP IMPROVEMENT PROTOCOL

Step 1: Map every conversion step in the loop with its current conversion rate
  Example (viral loop):
    A: User creates output            100% (everyone does this — it's the core workflow)
    B: User shares output             12% (this is the first step of the loop)
    C: Recipient clicks shared link   34%
    D: Recipient creates an account   22%
    E: Recipient gets value (FVM)     41%
    Full loop completion: 100% × 12% × 34% × 22% × 41% = 0.37% per user per cycle
    K-factor: ~0.37 (0.37 new users per 100 original users per cycle)

Step 2: Identify the weakest step (lowest conversion rate that is improvable)
  In the example: Step B (12% share rate) is far lower than steps C–E.
  Even doubling C–E would be less impactful than improving B from 12% → 20%.
  
  PRIORITY RULE: Fix the weakest step first, unless it is structurally fixed
  (e.g., step A is 100% by definition — no improvement possible).

Step 3: Diagnose WHY the step is weak
  Low share rate: sharing is not native to the workflow; friction is too high; output
  is not worth sharing (wrong format or audience)
  
  Low click rate: shared content is not compelling to non-users; preview is weak;
  recipient has no reason to click
  
  Low signup rate: landing page is confusing; value proposition for newcomers
  is unclear; too much friction in account creation
  
  Low FVM rate: onboarding doesn't route new users from the loop to the core value;
  they get lost in features instead of reaching the value moment that triggered the share

Step 4: Design one experiment per step (not all at once)
  Run experiments on the weakest step using /growth-experiment.
  Measure step conversion rate, not just the full loop k-factor, to isolate the effect.

---

Step 5 — Multi-loop management

Most mature products have more than one loop. Managing multiple loops requires understanding how they interact.

MULTI-LOOP ARCHITECTURE

Primary loop: the loop most directly connected to the NSM
Secondary loop: supplements the primary; typically longer cycle and lower k-factor

Common combination:
  Primary: Product / PLG loop (fast; short cycle; drives activation)
  Secondary: Content / SEO loop (slow; long cycle; drives organic acquisition)

INTERACTION EFFECTS:
  Loops can reinforce each other or compete:
  
  Reinforcing: "Users who create public content (content loop) are also more likely
  to share their outputs with colleagues (viral loop). Content loop feeds viral loop."
  
  Competing: "Over-optimizing for viral invites causes users to send invitations
  before getting full value. This reduces invite-to-signup conversion because the
  invitee arrives to a product where the inviter is not yet an enthusiast."

SEQUENCING RULE:
  Build one loop to a measurable k-factor (even k = 0.2) before investing in a second loop.
  Two weak loops are harder to improve than one measurable loop.

LOOP CANNIBALIZATION CHECK:
  IF loop A and loop B target the same user action → they may compete for the same
  user behavior. Analyze whether users who complete loop A's action have lower loop B
  completion rates.

---

Output format

## Growth Loop Design

**Product:** [Name]
**NSM:** [Metric from north-star-metrics]
**Growth motion:** [PLG / SLG / CLG]

### Loop diagnosis
Full loop present: [Yes / Partial / No]
Current mechanism: [Description]
Instrumented: [Yes / No — gap to fix]

### Loop archetype
Primary: [Viral / Content / Product / Community / Supply-demand]
Rationale: [Why this archetype fits this product]
Secondary (if applicable): [Archetype]

### Loop step map (with current conversion rates)
Step A → B: [Action] → [Rate: X%]
Step B → C: [Action] → [Rate: X%]
Step C → D: [Action] → [Rate: X%]
...
K-factor (current): [Calculated or estimated]
Loop cycle time: [Days]

### Weakest step
Step: [Which step]
Current rate: [X%]
Diagnosis: [Why it's weak]
Experiment to run: [Brief hypothesis → link to /growth-experiment]

### Target state
K-factor target (6 months): [X]
Loop cycle time target: [Days]
Primary improvement: [What changes]

### K-factor worksheet
Current loop step conversion rates:
| Step | Action | Conversion | Notes |
|---|---|---|---|
| A → B | [Action] | [X%] | [What drives this rate] |
| B → C | [Action] | [X%] | [What drives this rate] |
| C → D | [Action] | [X%] | [What drives this rate] |

K-factor calculation:
  Product of all step rates: [X% × X% × X% = X]
  K-factor: [value] (number of new users generated per 100 existing users per cycle)
  Interpretation: [K < 1: loop is additive / K ≥ 1: loop is compounding]

Sensitivity analysis (one step improved):
  If weakest step improves from [X%] → [Y%]: new K-factor = [Z]
  Required improvement to reach K = 0.5: [weakest step] from [X%] → [Y%]

### Next experiment
Targets: [Weakest step name]
Hypothesis: [If we [change], then [step] improves from [X%] → [Y%] because [reason]]
Design: → /growth-experiment

---

Brain reads / writes

If a companion brain repo is connected:

Before starting:

• Read experiments/experiment-log.md — past experiments on sharing, referral, or content creation reveal loop conversion data even if it wasn't labeled as "loop measurement"
• Read knowledge/icp-map.md — ICP determines which sharing behavior is natural (developers share code or benchmarks; ops teams share dashboards; community members share expertise)

Brain write (after loop design is complete):

• Write to decisions/: loop archetype selected, step map with conversion rates, k-factor baseline, and the experiment targeting the weakest step

Brain not connected: document the loop step map before running any loop experiments. You cannot improve what you don't measure.

---

Anti-patterns

Anti-pattern	Why it fails	Fix
"We have virality" without a k-factor	Virality without measurement is a hope, not a growth mechanism	Measure every loop step; calculate k-factor before claiming viral growth
Building a referral program before the product has clear value	Users refer when they're enthusiastic, not because there's an incentive; incentivized referrals bring low-quality users	Get D30 retention to a meaningful baseline before building referral mechanics
Optimizing the strongest loop step	Most teams focus on the most visible step (signup page), not the weakest (share rate)	Map all steps with conversion rates; always improve the weakest first
Designing a loop that requires unnatural user behavior	Users don't share unless sharing is a natural output of the workflow	The loop step must be native to the core workflow; adding a "Share this!" button is not a loop
Two-sided marketplace without geographic density	Thin supply across 50 markets produces zero value in each	Concentrate supply and demand in one market segment before expanding
Community loop on a locked-down platform	Discord and Slack content doesn't get indexed; the content loop doesn't fire	Community content must be publicly accessible and searchable
Confusing channels with loops	"We post on LinkedIn" is a funnel, not a loop	A loop requires the user action to generate the next user; channels require active team investment each cycle

---

Benchmarks (2025–2026)

Benchmark	Value	Notes
K-factor, B2B SaaS invite/share mechanics	0.1–0.3	Sustainable range; does not require true virality to compound
K-factor, B2B dev tools (content/community loop)	0.15–0.4	Long cycle but durable; compounds over quarters
K-factor, consumer social / collaboration	0.3–0.8	Context-dependent; audience exhaustion risk at high k
Loop cycle time, viral / product loop	1–7 days	Shorter cycle amplifies k-factor impact significantly
Loop cycle time, content / SEO loop	30–180 days	Slow but very durable; compounds without active team effort
Share rate (% of active users who share an output, B2B tools)	5–20%	Below 5%: sharing is not native to the workflow
Invite-to-signup conversion (B2B)	15–35%	Below 15%: the shared output is not compelling to non-users
Time to measurable k-factor from loop launch	4–8 weeks	Shorter loops produce signal faster; community loops take 3–6 months

---

Related skills

Skill	When to use
growth/north-star-metrics/SKILL.md	NSM must be defined before designing loops to compound it
growth/growth-experiment/SKILL.md	Loop step improvement experiments are designed here
growth/experiment-portfolio/SKILL.md	Multiple loop step experiments must be sequenced; don't run them in parallel on the same step
growth/retention-analysis/SKILL.md	Loops require retained users to complete; retention analysis validates the loop's prerequisite

---

Validation criteria

• Loop diagnosis performed (loop vs funnel distinction confirmed)
• Loop archetype selected with rationale (not just "referral program")
• K-factor calculated or estimated with explicit assumptions
• Every loop step is instrumented or instrumentation gap is documented
• Weakest step identified with diagnosis of why it is weak
• Improvement experiment targets the weakest step (not the most visible)
• Loop cycle time documented

References & Sources

Tier 1:

• Reforge growth loops framework: loop archetypes, k-factor calculation, loop vs funnel distinction, loop step optimization
• growth-motion-plg (growth-skills v1.0, score 8.5/10): PLG loop design, FVM-as-loop-input design
• growth-motion-community (growth-skills v1.0, score 7.5/10): community loop design, content indexability requirements