The Rented Audience: How Perplexity's 80% Collapse Exposes the Biggest Lie in AI Growth

There's a chart circulating on X and LinkedIn right now that should terrify every AI founder, every growth marketer, and every VC who's written cheques based on hockey-stick user numbers.

It shows Perplexity, one of the most celebrated AI startups of the past two years, watching their global app downloads collapse by 80% in just six weeks.

Eighty percent. Six weeks.

This isn't a blip. It's a diagnosis. And if you're building in AI, it might be yours too.

Let's be clear about what Perplexity built. On paper, the metrics are staggering: 50 million app downloads. 30 million monthly active users. 780 million queries at peak. Partnerships with Airtel, Snap, PayPal. Celebrity campaigns featuring Ronaldo and Lewis Hamilton. Premium $50+ CPM advertising rates that made media buyers raise their eyebrows.

Gizmodo called it a Google killer. VCs valued it at billions. The growth charts went up and to the right in exactly the way pitch decks promise they will.

And then ChatGPT launched a free product in India, and more than half of Perplexity's growth vanished almost overnight.

Here's what the chart doesn't show you: roughly 50% of Perplexity's downloads since July 2025 came from a single source, an Airtel India partnership that gave away free Perplexity Pro subscriptions. When you combine a promotional tsunami with heavy paid acquisition through celebrity endorsements and platform deals, you get impressive numbers that mask an uncomfortable truth.

Those weren't users. They were tourists.

They came because it was free. They came because a telco bundled it. They came because an algorithm served them an ad featuring a footballer they admire. And when a competitor offered something similar, also free, also convenient, they left without a backward glance.

Because they had nothing invested. No reputation built. No relationships formed. No identity attached. No reason to stay except the product itself.

And in AI, the product is never enough.

When’s the last time you updated your digital marketing strategy?

If you’re relying on old-school PPC tactics you might be missing out on a major revenue opportunity.

Levanta’s Affiliate Ad Shift Calculator shows how shifting budget from PPC to creator-led partnerships can significantly improve conversion rates, ROI, and efficiency.

Discover how optimizing your affiliate strategy can unlock new profit potential:

Built for brands ready to modernize how they grow.

Get your free calculation.

Here's the brutal arithmetic of building in AI right now: any feature you ship today, your competitors can replicate in months. Sometimes weeks.

Perplexity's core innovation, an "answer engine" that synthesises web sources and provides citations, is no longer unique. OpenAI offers it. Google offers it. Microsoft offers it. A dozen startups offer variations. The underlying large language models are increasingly accessible. The retrieval-augmented generation patterns are well-documented. The user interface paradigms are converging.

When ChatGPT launched its competing free product in India, it didn't need to be better than Perplexity. It just needed to be comparable and free. That was sufficient to trigger an exodus.

This is the commoditisation trap, and almost every AI company is walking into it with their eyes closed, distracted by the intoxicating early metrics of a market that's still finding its shape.

The question isn't whether your AI product can be replicated. It will be. The question is: what remains when it is?

For Perplexity, the answer appears to be: not enough.

Let me name what's actually happening here, because it extends far beyond one company.

Perplexity has a rented audience.

Every major growth lever they've pulled puts users in their product temporarily, with no mechanism to convert that attention into allegiance:

Rented through partnerships. The Airtel deal delivered millions of downloads. But those users' primary relationship is with Airtel, not Perplexity. When the promotion ends, so does the relationship.

Rented through celebrities. Ronaldo and Hamilton can generate awareness, but awareness without identity is just noise. Users acquired through celebrity endorsement have no more loyalty to Perplexity than to any other product those celebrities might promote next month.

Rented through platform distribution. Snap integration, PayPal partnerships: these provide reach, but the users belong to those platforms. They're borrowed, not owned.

Rented through advertising. At $50+ CPMs with no attribution analytics, Perplexity's ad product is an experimental line item for brands, not a committed channel. Advertisers are testing, not investing.

The pattern is consistent: scale without depth, reach without relationship, users without community.

And when competitive pressure arrives, as it always does, there's nothing holding anyone in place.

There's a deeper wound here that the growth collapse has exposed.

Despite impressive ratings (4.8/5 on Product Hunt) and genuine enthusiasm from power users, Perplexity faces growing criticism from technical users and researchers about hallucinated citations and over-confident answers. Late 2025 benchmarks showed cases where Perplexity fabricates answers and misuses sources more than some competitors.

This creates what I call the hallucination paradox: the product's core promise is trustworthy, cited answers, but the AI underneath sometimes invents sources or misrepresents them. Individual users can't easily verify this. The citations look authoritative even when they're not.

For a rented audience, this is fatal. Users who came for utility and have no other investment will leave at the first sign of unreliability. There's no reservoir of trust, no community vouching for the product, no collective stake in making it better.

But here's the counterintuitive opportunity buried in this problem: community can solve what AI alone cannot.

Individual AI checking sources is fallible. Collective human verification is more robust. If Perplexity had built a community of researchers who could flag, annotate, and improve answers, creating reputation around contribution quality, they would have both a solution to their accuracy problem and a switching cost that no competitor could replicate.

They didn't. And now they're paying for it.

I've spent years studying what makes companies defensible. Not theoretically defensible in a strategy document, but actually, practically immune to the competitive dynamics that kill most ventures.

The answer is never technology alone. Technology is replicable. The answer is rarely brand alone. Brand without community is just expensive awareness.

The answer is community-based competitive advantage: what I call a moat built on Mission, Organic Leadership, Authentic Engagement, and Tribal Identity. The MOAT Framework.

Let me show you what this could have looked like for Perplexity, and what it can still look like for any company willing to do the work.

Perplexity's current positioning is functional: "answer engine." It describes what the product does, not what it stands for.

Contrast this with what a mission-aligned Perplexity might declare: "We're building the most trustworthy knowledge tool on earth, and our community holds us accountable."

That's not a feature. That's a movement. It's something people can believe in, contribute to, and identify with. It transforms the hallucination problem from a liability into a differentiator: we're the company that admits AI isn't perfect and builds collective intelligence to make it better.

When users believe in your mission, they don't leave when a competitor offers a similar feature for free. They stay because they're part of something.

Buried in Perplexity's data is an asset they've never properly leveraged: power users who genuinely love the product.

These are researchers, journalists, marketers, and analysts who've integrated Perplexity into their daily workflows and rave about it on social media. They exist. They're creating content. They're telling others.

But Perplexity treats them as customers rather than leaders.

An organic leadership programme would identify these people, elevate them into visible community roles, give them genuine influence over product direction, and equip them to teach others. These become voices the company doesn't control but amplifies, and voices that competitors cannot hire away because their reputation is built within the Perplexity ecosystem.

Ronaldo can endorse anyone who pays him. A researcher who's built their reputation as a Perplexity expert has a vested interest in the platform's success.

Here's a question that reveals everything: can Perplexity users connect with each other?

The answer is no. Engagement is transactional: query, answer, done. There's no community layer where users share discoveries, build on each other's research, follow interesting minds, or collaborate on complex questions.

Compare this to what authentic engagement could look like:

Each of these creates investment. Investment creates switching costs. Switching costs create defensibility.

A user who's spent six months building a research collection with 500 followers isn't going to abandon that for a marginally better AI. A researcher whose reputation for verification accuracy is visible on their profile has invested something they can't take elsewhere.

This is the engagement Perplexity doesn't have, and the absence is why their users could disappear so quickly.

What does it mean to be a "Perplexity person"?

Right now, probably nothing. There's no shared language, no visible signals of membership, no in-group identity that distinguishes Perplexity users from ChatGPT users or Google users.

Tribal identity is the culmination of the other three pillars. When there's a mission worth believing in, leaders worth following, and engagement worth investing in, identity emerges naturally. Users start saying "I'm a Perplexity researcher" the way people say "I'm an Apple person" or "I'm a Notion user."

That identity is the ultimate moat. It's emotional, not rational. It survives feature parity. It survives price competition. It survives promotional offers from competitors.

And it's the one thing Perplexity never built.

Here's what's still possible for Perplexity, and for any company seeing their own version of this rented audience problem.

The users who stayed after the Airtel promotion ended, who remained when ChatGPT went free in India, who still use Perplexity despite the competitive alternatives: these are the foundation. They're not tourists. They're the beginning of a community.

A serious transformation looks like this:

Months 1-2: Emergency Foundation

Stop the bleeding. Identify the true believers who stayed. Launch a community leader programme from this cohort. Create features that let power users publish and share their work. Redirect spend from celebrity endorsements to community infrastructure.

Months 3-5: Build Investment Mechanisms

Launch profiles with visible contribution history. Enable collections and research trails. Introduce community verification features. Start vertical communities for high-value segments: researchers, marketers, journalists.

Months 6-9: Activate the Organic Flywheel

Community leaders creating content that attracts new users. Shareable collections ranking in search and social. Referral tied to status, not discounts. User-generated growth replacing paid acquisition.

Months 10-12: Prove Defensibility

Document community contribution to accuracy. Show organic growth metrics versus paid. Measure retention curves for invested users versus acquired users. Demonstrate to boards and investors that the moat is real.

This isn't theoretical. This is the difference between a company that survives the AI commoditisation wave and one that gets crushed by it.

Here's what I know about the next 18 months in AI:

Features will converge. What seems differentiated today will be table stakes tomorrow. Every major tech company is building answer engines, AI assistants, research tools, and knowledge products. The technology gap is closing faster than most founders want to admit.

Paid acquisition will get more expensive. As more AI companies chase the same users through the same channels, CAC will rise. The promotional partnerships that drove early growth will become bidding wars that only the best-capitalised players can win.

Commoditisation will accelerate. Open-source models are improving. Development frameworks are maturing. What took eighteen months to build will take six. Then three.

In this environment, the only sustainable advantage is one that can't be replicated with code or capital: a community of users who've invested their reputation, their relationships, and their identity in your platform.

Perplexity's 80% collapse isn't an anomaly. It's a preview. The same dynamics that drove their rented audience away will affect every AI company that's built scale without building community.

The question is whether you'll learn from their experience or repeat it.

If you recognise any of these patterns in your own growth metrics, the time to act is now:

These are symptoms of the rented audience problem. They don't show up in vanity metrics. They show up when competitive pressure arrives, and by then, it's often too late.

At BuildMoat, we help AI-era companies build the only competitive advantage that survives commoditisation: community-based moats.

We've developed the MOAT Framework (Mission Alignment, Organic Leadership, Authentic Engagement, Tribal Identity) as a systematic methodology for creating defensibility that technology and capital cannot replicate.

If you're seeing patterns similar to what's now publicly visible about Perplexity, such as volatile growth, promotional dependency, weak retention, or competitive vulnerability, we should talk.

We offer a MOAT Diagnostic: a focused assessment of your current community defensibility, the gaps that leave you vulnerable, and a prioritised roadmap for building genuine competitive advantage.

For organisations facing urgent competitive pressure, we provide full strategic advisory: community architecture, leadership programmes, engagement design, and the implementation support to make it real.

The companies that thrive in the AI era won't be the ones with the best models or the biggest marketing budgets. They'll be the ones who stopped renting audiences and started building tribes.

The clock is running. The question is what you'll do with the time you have left.

Emeka Ogbonnaya is the founder of BuildMoat, an AI defensibility consultancy helping businesses build community-based competitive advantages through the MOAT Framework. With 15+ years of product and strategy experience at major financial institutions, he now works with growth-stage companies to create the moats that technology alone cannot provide.

To discuss your organisation's community defensibility, reach out at [email protected] or visit https://buildmoat.co.uk

We appreciate you taking the time to read this. Building community isn't just theory for us, it's a framework we've developed and refined with companies at every stage.

Have questions about adopting The Moat framework for your business? We'd love to help. Book a discovery call to speak directly with one of our consultants. We'll audit your current community engagement process and show you exactly where the opportunities are.

Share this article: The founders and investors who need to see this won't find it unless you share it. If this diagnosis resonated, pass it on.

Subscribe: For ongoing analysis of AI defensibility, community strategy, and the moats that matter, join the BuildMoat newsletter.

The Moat
Where Connection Becomes Your Competitive Edge