growth loops

AARRR — Acquisition, Activation, Retention, Revenue, Referral — is the most widely taught growth framework in PM courses. It is also the reason most PMs think about growth incorrectly.

The pirate metrics framework is a funnel. It is linear. Users enter at the top, leak out at each stage, and the survivors reach the bottom. To grow, you pour more users in. When you stop pouring, growth stops. This is not a growth engine. It is a bucket with holes.

Growth loops are the corrective. In a loop, the output of the system feeds back as the input. A user creates value. That value attracts another user. That user creates more value. The system reinforces itself. The difference between a funnel and a loop is the difference between renting growth and owning it.

After working with growth teams across Indian startups and running hundreds of PMs through this material at Pragmatic Leaders, I can tell you the single biggest mistake: teams optimise funnel stages without asking whether their system loops at all.

Why funnels fail as a growth model

A funnel is a diagnostic tool, not a strategy. It tells you where users drop off. It does not tell you how to build a system that grows on its own.

The problem with funnel thinking:

1. It assumes a constant external input. Every user in a funnel comes from outside — paid ads, SEO, partnerships. The moment you reduce spend, the top of the funnel shrinks. There is no self-reinforcement.

2. It treats referral as the last stage. In AARRR, referral sits at the bottom. Most teams never get to it because they are busy patching leaks in activation and retention. But referral — or more precisely, the mechanism by which your product generates its own demand — should be the first thing you design, not the last.

3. It hides the real question. The real question is not “what is our conversion rate at each stage?” The real question is: “does our product generate more input than it consumes?” If the answer is no, you are running on a treadmill. You can optimise the treadmill’s speed, but you are still running.

What a growth loop actually is

A growth loop has three parts:

Input: A new user, a piece of content, a dollar of revenue — something enters the system.

Action: That input produces value — a review, an invitation, a transaction, a piece of content.

Output that becomes new input: The value produced attracts or funds the next round of inputs. The output is not an endpoint. It feeds back into the system.

The key test: if you stopped all external marketing today, would your product still grow? If the answer is yes, you have a loop. If the answer is no, you have a funnel with a referral feature bolted on.

Most products have elements of multiple loops. But one loop dominates. The growth PM’s job is to identify which loop is actually driving growth, strengthen it, and stop pretending the others are working.

The four loop types

1. Viral loop: user invites user

The output is an invitation. A user gets value from the product, invites another user, that user gets value and invites another.

The mechanics: User A uses the product → User A invites User B (because the product is better with others, or because sharing is built into the core action) → User B signs up → User B invites User C.

Indian examples:

• WhatsApp — the purest viral loop in India. You cannot use WhatsApp alone. Every message is an implicit invitation. Every group is a growth mechanism. WhatsApp did not spend on user acquisition in India. The product was the acquisition channel.
• Meesho — social commerce where resellers share product catalogues on WhatsApp. The act of selling is the act of distributing. Every reseller is a distribution node.
• Dream11 — fantasy sports where you create a league and invite friends. The game is meaningless without opponents. Inviting is not a feature. It is the product.

The viral coefficient (k): This is the average number of new users each existing user brings in. If k > 1, your product grows exponentially without any external input. If k < 1, the loop decays — each generation is smaller than the last. Very few products sustain k > 1. WhatsApp did. Most do not.

But here is the part most PMs miss: even k = 0.3 compounds. If every 10 users bring in 3 new users, and those 3 bring in 1, and that 1 brings in 0.3 — you are getting 14.3 users for every 10 you acquire. Over time, that 43% bonus compounds on every cohort. A paid funnel gives you exactly what you pay for. A viral loop with k = 0.3 gives you 43% more. Run that for twenty cohorts and the difference is enormous.

2. Content loop: user creates content that attracts new users

The output is content. A user creates something — a review, a post, a listing, a question — that gets indexed, discovered, and brings in a new user who creates more content.

The mechanics: User A creates content → content gets indexed by Google or distributed on social → User B discovers it → User B signs up → User B creates content.

Indian examples:

• Zomato — restaurant reviews and photos. Users write reviews. Reviews rank on Google. New users find Zomato through those reviews. Those users write their own reviews. The loop is so strong that Zomato’s SEO moat is built almost entirely on user-generated content.
• Pratilipi — Indian language storytelling platform. Writers publish stories. Readers discover them through search and social. Some readers become writers. The content library grows, attracting more readers.
• ShareChat — regional language social content. Users create short videos and memes. Content gets shared on WhatsApp and Instagram. New users come to create their own content. The creation-distribution cycle is the growth engine.

Why content loops are powerful: They create a compounding asset. Every piece of content is a permanent acquisition channel. Paid ads stop working when you stop paying. A Zomato review from 2019 is still driving traffic in 2026. Content loops have the best unit economics of any loop type — but they are slow to start and require a critical mass of creators.

3. Paid loop: revenue funds acquisition that generates more revenue

The output is money. Revenue from existing users funds the acquisition of new users who generate more revenue.

The mechanics: User A pays → revenue funds ads or sales → ads acquire User B → User B pays → revenue funds more ads.

Indian examples:

• Zepto and Blinkit — quick commerce where average order value funds the next round of performance marketing. The loop works as long as LTV (lifetime value) exceeds CAC (customer acquisition cost) with enough margin left over to fund operations.
• Nykaa — beauty e-commerce where repeat purchase revenue funds Google and Instagram ads that acquire new buyers who also repeat-purchase.
• upGrad — high-ticket ed-tech where ₹2-5 lakh course fees fund aggressive performance marketing on YouTube and Google.

The math that matters: A paid loop is sustainable only when:

LTV > CAC + COGS + operating margin

If your customer’s lifetime value is ₹500 and your acquisition cost is ₹400, you have ₹100 to cover everything else. That is not a loop — it is a trap. The loop only works when the surplus is large enough to reinvest in acquisition and cover costs and generate profit.

4. Sales loop: revenue funds a sales team that closes more revenue

The output is a closed deal. Revenue from existing customers funds sales hiring, and new salespeople close new deals that fund more hiring.

The mechanics: Sales team closes Deal A → revenue funds hiring → new salesperson closes Deal B → more revenue → more hiring.

Indian examples:

• Freshworks — started with inbound product-led growth, then built an enterprise sales team funded by SMB revenue. The enterprise deals funded a larger sales team that closed bigger deals.
• Darwinbox — HR-tech that grows through enterprise sales. Each closed deal funds the sales team that closes the next deal. The sales cycle is long (3-6 months) but deal sizes are large enough to sustain the loop.
• Zoho — built a sales organisation across India and APAC where revenue from existing accounts funds expansion into new geographies and segments.

When this loop works: Sales loops work for high-ACV (annual contract value) products where the deal size justifies human involvement. If your ACV is under ₹5 lakhs, a sales loop is hard to sustain — the cost of a salesperson exceeds the revenue they generate. Above ₹25 lakhs ACV, a sales loop becomes the dominant model.

Most Indian startups misidentify their loop

This is where I see the most confusion. A referral program is not a viral loop. A blog is not a content loop. A sales team is not a sales loop. The loop is defined by what actually drives growth, not what features you have built.

How to tell which loop you actually have: Look at your last 10,000 new users. Where did they come from? Not where your attribution model says they came from — where did they actually come from? If 70% came from paid channels, you have a paid loop regardless of what your referral dashboard shows. If 50% came from organic search landing on user-generated pages, you have a content loop. The loop is an empirical fact, not an aspiration.

The paid loop trap in India

Cheap digital advertising in India created a generation of startups that mistook a paid funnel for a growth engine. Meta and Google ads in India have historically been 3-5x cheaper than in the US on a per-click basis. This made paid acquisition look sustainable when it was not.

Byju’s — spent aggressively on performance marketing and tele-sales, acquired millions of users, but the cost of acquisition exceeded the lifetime value for a large percentage of customers. When funding dried up and ad costs rose, the loop collapsed. There was no organic engine underneath.

Lido Learning — spent heavily on Facebook ads to acquire students for live tutoring. The unit economics never worked — CAC was too high relative to LTV. When they could not raise more money to fund acquisition, they shut down. The “growth” was a paid funnel, not a loop.

The pattern: Cheap venture capital subsidises cheap acquisition. Growth looks real. But the loop test fails — if you stop spending, growth stops. The moment capital becomes expensive (as it did in 2022-2023 in India), every startup running a subsidised paid funnel hit a wall simultaneously.

The corrective: always ask “what is our organic growth rate?” Strip out all paid acquisition. What is left? If the answer is close to zero, you do not have a loop. You have a spending habit.

Compounding vs draining: the mathematics

This is worth making concrete. Suppose you acquire 1,000 users per month through paid ads.

Funnel model (no loop): Every month, you get exactly 1,000 users. After 12 months, you have 12,000 users (minus churn). Growth is linear. If you stop spending, growth goes to zero.

Loop model (k = 0.3): Every month, you acquire 1,000 users through paid ads. Each cohort of 1,000 generates 300 additional users through the loop. Those 300 generate 90. Those 90 generate 27. The total from each monthly cohort is not 1,000 — it is approximately 1,429 (the geometric series converges to 1,000 / (1 - 0.3) = 1,429).

After 12 months with the same ad spend:

• Funnel: ~12,000 total users acquired
• Loop (k=0.3): ~17,143 total users acquired

That is 43% more users for the same spend. And the gap widens every month because each cohort’s loop output stacks on top of the previous ones.

Now consider what happens when you stop spending:

• Funnel: Zero new users immediately
• Loop: The existing users continue generating 300 per 1,000 active users per month, decaying but not stopping

Even a weak loop (k = 0.2) is worth more than a strong funnel over any time horizon longer than a year. This is why identifying and strengthening your loop — even if it is weak — matters more than optimising your funnel conversion rates.

How to strengthen your loop

Once you have identified your actual loop, the growth PM’s job is to reduce friction at every point in the cycle.

For viral loops: Reduce the effort required to invite. Reduce the friction for the invited user to get value. Increase the number of natural invitation moments in the product. WhatsApp did not add a “refer a friend” button. They made the product unusable without inviting someone.

For content loops: Make it easier to create content. Make content more discoverable (SEO, recommendations, social sharing). Reduce time-to-first-creation for new users. Zomato gamified reviews with badges and leaderboards — not because gamification is a growth strategy, but because it increased creation frequency, which fed the loop.

For paid loops: Increase LTV (retention improvements, cross-sell, upsell). Decrease CAC (better targeting, creative testing, channel diversification). Improve payback period so you can reinvest faster. The paid loop is a capital efficiency game.

For sales loops: Increase deal size. Decrease sales cycle length. Improve win rate. Reduce ramp time for new salespeople. Every improvement to sales efficiency directly funds more sales capacity.

The reframe conversation

The hardest part of loop thinking is not the analysis. It is the conversation with leadership when the analysis reveals an uncomfortable truth.

Designing a loop where none exists

Some products genuinely do not have a loop. They are pure funnels — new users come from paid acquisition or sales, use the product, and that is the end of the chain. No content is created. No invitations are sent. No revenue surplus funds further acquisition.

If that is your situation, the question is not “how do we optimise the funnel?” The question is: “can we design a loop into the product?”

Three approaches:

Add a content layer. Can your users create content that would attract other users? Reviews, templates, public profiles, shared dashboards. Notion grew partly because users shared templates publicly. Canva grew because users shared designs on social media. The content was not the core product — but it was the loop.

Make the product multiplayer. Single-player products do not loop. Multi-player products can. If your product works better with colleagues, the act of onboarding a colleague is the viral loop. Slack grew this way. Figma grew this way. The question is: does your product have a legitimate multi-player mode, or would you be bolting one on artificially?

Build a referral mechanism that is structural, not incentive-based. A ₹100 cashback referral is an incentive. It is a cost centre, not a loop. A structural referral is when the act of using the product inherently exposes it to non-users. A payment link from Razorpay, a scheduling link from Calendly, a shared Notion page — these are structural. The product markets itself through normal usage.

If none of these fit naturally, accept that you have a funnel business. Not every product can loop. Some of the most successful B2B companies in India run on sales loops, which are the slowest-compounding loop type. That is fine — as long as you are investing in the right engine and not pretending you have a different one.

Where to go next

• Understand the role that owns loops: Growth Product Management
• Optimise the moment users first experience value: Activation Optimization
• Build the loop that keeps users coming back: Retention Loops
• Choose the right metrics for your loop type: Metrics and KPIs