(2025-06-18) Balfour The Next Great Distribution Shift

Brian BalfourThe Next Great Distribution Shift. Here's something that should keep every product leader up at night: We're living through one of the most significant technology shifts in history, yet we're still distributing products like it's 2015.

The AI revolution has transformed how we build products. We can now create experiences that would have been impossible just two years ago. But there's a glaring problem—this technology shift has come without a distribution shift.

In fact, AI is actively destroying the distribution/sales channels we've relied on for decades. SEO traffic is plummeting as users shift to answer-engine platforms.

But I believe we're on the precipice of something bigger. The next great distribution shift is coming

there will be extraordinary windows of opportunity. There will also be casualties.

Every major new platform follows the same playbook: They start open and generous, practically begging developers to build on their platform. (enshittification?)

But once that moat has escape velocity? The walls go up.

The difference between the companies that thrive and those that get crushed? They understand the game being played. They see the pattern. They prepare for the inevitable shift from open to closed.

I'm going to show you exactly how this cycle works.

We'll walk through the rise and fall of multiple major distribution platforms from Facebook to Google to LinkedIn.

I’ll predict who I think will be the next distribution platform and why.

Most importantly, we'll cover what you need to do now—before the window closes—to position your product for success in the AI platform era.

You can't opt out of this game.

Technology Shifts vs Distribution Shifts

In December 2023, Casey Winters published an important essay that I’ve been thinking about since. In "On Platform Shifts and AI," he made a distinction that every product leader needs to understand. What separates a major platform shift from a minor one isn't the technology itself. It's whether that shift enables both new ways of building things AND new ways of reaching people.

Winters' insight gets really interesting, and where most people miss the pattern:
"What I realized having gone through the internet and mobile platform shifts is that the technological and distribution shifts did not happen at the same time. Platform shifts that create both technological and distribution opportunities happen in a sequence, not all at once."

Which brings us to AI.

We're clearly in the middle of a massive technological shift. AI has enabled entirely new ways of building products—from conversational interfaces to autonomous agents to personalized experiences that adapt in real-time. The technology is here, and it's spectacular.

But where's the distribution?

ChatGPT has 700 million users, yet it's not a distribution channel—it's a destination. If anything, AI is destroying distribution channels. SEO is collapsing as answer engines like Perplexity, ChatGPT Search, and others keep users in its experience rather than sending traffic out

The old channels are dying, but the new ones haven't emerged.

But—and this is crucial—we shouldn't expect the distribution shift to have happened yet. If history is our guide, we're right on schedule. The AI technology shift started gaining mainstream adoption in late 2022 with ChatGPT. If it follows the mobile timeline, we wouldn't expect to see the distribution shift until about 2026 or 2027.
Except I think it's going to happen faster this time.

Every major distribution platform in history has followed the same three-step cycle.

Step 1: Identify the Moat

For Facebook, it was the social graph—who knows whom. For Google, it was search data—what people want. For Apple, it was having an application ecosystem to attract premium device owners.

Step 2: Open the Gates

create "open" ecosystems, practically begging developers to build on top of them. Free API access. Viral growth mechanics. Revenue sharing that seems too good to be true.

Step 3: Close for Monetization

the rules change. What was free becomes paid. What was permitted becomes restricted

Sometimes this is deliberate and planned, sometimes this is a result of an emergent business need.

Critical Warnings About This Cycle

The timeline is unpredictable but accelerating.
Facebook went from open to closed in about two years. Apple took four. Google stretched it over two decades.

Step 3 is a bloodbath... entire business models. Most developers are completely unprepared for how aggressive the platform will become.

Knowledge is survival.

The companies that thrive through platform shifts aren't necessarily the biggest or best-funded. They're the ones who understand the game being played. They build with the end in mind. They extract value during Step 2 while preparing for Step 3.

Let's look at exactly how this cycle has played out across major distribution channels. The patterns are so consistent it's almost eerie.

Facebook Developer Platform

In May 2007, Facebook wasn't the inevitable winner we know today. They had 25 million users—impressive, but hardly dominant.

The competitive landscape was brutal:

They needed something to hit escape velocity before another platform locked in the global social graph.

Step 1: Identifying the Moat

Facebook recognized their moat was the social graph itself

Building that graph user by user, friend by friend, was too slow. They needed acceleration.

Step 2: Opening the Gates

In May 2007, Facebook launched f8 and their developer platform.

By late 2007, over 7,000 apps existed on the platform, with 100 new apps launching every single day. By mid-2008, there were 33,000 apps

Facebook grew from 25 million to 250 million users in just two years,

Step 3: Closing for Control

The closing happened in waves, each more restrictive than the last.

• 2009-2010: The Viral Channels Disappear
• 2010-2011: The Tax Man Comes... Facebook takes 30% of all transactions
• 2011-2012: The Feature Absorption

It wasn’t just apps as well as Casey Winters notes:
Facebook pages worked the same way. They incentivized companies to grow page likes to get free distribution. Then distribution turned paid even if a user liked your page.

Apple App Store

Apple entered the smartphone market as an underdog. In 2008, when the App Store launched, the mobile landscape looked nothing like today's iOS-Android duopoly:

• Nokia: 40% global market share with Symbian OS
• Android: Just launching, but backed by Google

Apple needed something to make the iPhone indispensable—to justify the premium price and lock in users before Android gained momentum.

Step 1: Identifying the Moat

Apple recognized their moat wasn't just hardware—it was the ecosystem. If they could make the iPhone the platform with the best apps,

Step 2: Opening the Gates

In July 2008, Apple launched the App Store with what seemed like a developer-friendly proposition at the time. (Note the iPhone itself launched the previous year.)

2010, the App Store had 225,000 apps and had generated $2.5 billion in revenue

Step 3: Closing for Control

The restrictions came slowly, each justified by "user experience" or "security," but the pattern was unmistakable:

2011: The In-App Purchase Mandate

2012-2015: The Slow Squeeze

• Bans on apps that "duplicate core functionality"
• Introduction of Search Ads in the App Store reducing organic distribution and visibility

2016-Present: Maximum Control

• App Tracking Transparency decimates ad-based business models

Google Search

In the early 2000s, search was fragmented. Yahoo served as the starting point for millions. AltaVista appealed to engineers.

Step 1: Identifying the Moat

Google's moat was elegantly circular: data and content. The more people searched, the more data they gathered. The better their results, the more people searched. But this virtuous circle had a critical dependency—they needed the entire web's content.

Step 2: Opening the Gates

Google's early philosophy bordered on religious devotion to openness. Larry Page captured it perfectly in 2004:
"We want to get you out of Google and to the right place as fast as possible."

For over a decade, it felt like genuine partnership. Google's revenue came from clearly marked ads in yellow boxes or sidebars. The main results remained sacred

Step 3: Closing for Control

Phase 1 (2010-2015): The Ad Creep Ads moved from sidebar to prime real estate above organic results.

• Google killed keyword data with "not provided," blinding websites to their traffic sources.

Phase 2 (2015-2020): The Feature Takeover

• Google Featured snippets answered questions without clicks. Knowledge Graph pulled data without attribution

Phase 3 (2020-Present): The Closed Garden

• Today, a commercial search reveals Google's true form. You'll see ads, then shopping results, then maps, then featured snippets, then "People Also Ask" boxes

Google began competing directly in lucrative verticals. Travel queries that fed Expedia now show Google Flights.

LinkedIn

LinkedIn's platform shift is happening right now, in real-time. It's also the fastest we've seen—from open to closed in less than four years

For most of its existence, LinkedIn was essentially a digital resume database. Users updated their profiles maybe once a year. Engagement meant accepting connection requests. But there was an internal mandate: transform LinkedIn from a static directory into a daily destination.

Step 1: Identifying the Moat

LinkedIn's moat is professional data—not just who you are, but what you do, what you care about, and how you behave professionally. This data is gold to three lucrative customer segments: marketers targeting B2B buyers, recruiters hunting talent, and salespeople seeking warm leads.

But static profiles only reveal so much. LinkedIn needed dynamic data—what you read, what you share, what makes you engage.

Step 2: Opening the Gates

Around 2020, LinkedIn made a strategic decision to prioritize "creator" content in the feed. The deal was irresistible: publish business content directly on LinkedIn, and they'd give you extraordinary organic reach.

For about two years, it was a content heaven. B2B marketers abandoned Twitter and blogs for LinkedIn posts.

Step 3: Closing for Control

The first signs of closure appeared in late 2022. Creators started noticing declining reach. Posts that once reached 50,000 professionals now struggled to hit 5,000

Then came the smoking gun: Thought Leader Ads.
Launched in June 2023

Today, LinkedIn's feed is starting to look more like driving down the 101. Sponsored content dominates

The Impending New Distribution Channel: ChatGPT

My personal belief is that ChatGPT is the next distribution platform. I think it will happen within the next 6 months.

Step 0: The Competitive Environment

we're still in a genuinely competitive environment.

Users still have relatively low switching costs.

This is what this stage of the cycle always look like—before someone figures out the moat.

Step 1: Identifying the Moat

OpenAI has identified their moat, and it's not the model quality. It's context and memory.

MCP only addresses context portability, not memory.

The accumulated history of interactions, the learned preferences, the refined understanding of each user—that's not portable. And that's where the real lock-in lives.

Every document, every conversation, every preference, every workflow. (Not quite true, see what Simon Willison has exported.)

Step 2: Opening the Gates

They can't capture this context and memory alone. No single company can build integrations with every tool,

There are a few early signals that they are about to open the gates:

They recently launched connectors w/ Deep Research to tools like HubSpot, Box, GitHub, and more. (Note Slack.com had blocked everyone(?) the week before this piece.)

Step 3: The Inevitable Closure

Once OpenAI achieves context lock-in—when switching to another AI means losing months or years of accumulated context—the platform dynamics will shift.

I’m not implying OpenAI is “evil” or deliberately planning this from the start. It’s just the reality. The cycle always ends the same.

Why OpenAI / ChatGPT Over Others?

If you are a Reforge alum, you should have one thing burned into your brain - those with the highest retention & engagement win categories. Retention is the god metric.

ChatGPT clearly has the highest retention

They have attained the rare and magic “smile curve” of retention.

It will fuel the engagement loops that will drive integrating more context and capturing more memory

The Accelerating Timeline

Casey Winters observed something crucial:
"With every generation, companies that reach massive scale have gotten more efficient at preventing other companies from growing on top of them, at least for free."

How long will the AI window last? My bet: two years. Maybe less.

There Is No Opt’ing Out

Here's the uncomfortable truth: knowing the game doesn't mean you can opt out.

In isolation, integrating with ChatGPT makes no sense. Why would HubSpot want to become a database to ChatGPT's interface?

But we don't operate in isolation. We operate in competitive markets. And markets are ruthless about efficiency.

When customers start asking why your product doesn't work with ChatGPT like your rival's does, what's your response?

How do you use the platform while building your own moat?

This is where product strategy matters. You need to create value that persists even when platform access gets restricted. Build direct relationships. Capture first-party data. Create moats within your own product.

The companies that survived previous platform shifts didn't just integrate—they integrated strategically. They used Facebook's viral channels while building email lists. They leveraged Google's traffic while developing brand loyalty. They sold through the App Store while creating web experiences. (Hmm smells like a just-so story. Could he have predicted the winners and losers at the time? But yeah, you have to follow the cycle and figure out how to stay alive.)

This time won't be different. Use ChatGPT's distribution, but don't depend on it.

The stampede is coming. If I’m right, in six months, it’s likely every SaaS product and consumer application will be rushing to complete a ChatGPT integration

The clock isn't just ticking, it's accelerating. And in this game, the house always wins.

But if you play it right, you can win too—at least for a while.

Appendix: How This Prediction Could Be Wrong

I give it 80% odds that ChatGPT/OpenAI becomes the next major distribution platform. But I'm 100% certain that whoever wins will follow the same open-grow-close pattern we've documented.

there are legitimate reasons why ChatGPT might not be the one.

Apple Gets Their Act Together

Apple is the sleeping giant that could change everything

As Aaron White said on Unsolicited Feedback:
“If you own the device, you see everything the user sees — that’s why Apple is in the best position.” (And Google Gemini plus Android and Google Apps)

If Apple suddenly shipped a compelling AI assistant that leveraged all that context, integrated with all their services, and opened it to developers, they could leapfrog everyone.

But "if" is doing heavy lifting here. Apple's AI efforts have been underwhelming

OpenAI clearly sees this threat—hence their reported device collaboration with the acquisition of Jony Ive.

OpenAI Fumbles It

This is the most likely failure mode. OpenAI could absolutely blow their advantage by moving too aggressively, too fast. (Feels like another just-so story.)

OpenAI Already Has Escape Velocity And Doesn’t Need A Platform

That might be true in isolation, but once again we live in a competitive environment. The rational thing for Claude, Google, others to do in this case is to build a platform themselves

MCP Makes Context More Portable

In theory, it commoditizes integrations

But theory isn't reality. Developers and businesses are resource-constrained. They'll focus on the largest platform first, and ChatGPT has at least 20X Claude's user base and is growing faster.

Plus, MCP only addresses context portability, not memory. I’ll repeat - the accumulated history of interactions, the learned preferences, the refined understanding of each user—that's not portable. And that's where the real lock-in lives.

There Are Major Other Players With Lots of Distribution

Microsoft has Copilot embedded across Office. Google has Gemini integrated everywhere

What matters isn't size—it's engagement depth, platform readiness, and trajectory.

My hypothesis is that if we could see the engagement numbers between platforms, ChatGPT has the deepest engagement

Regulation

Data privacy laws, AI safety regulations, antitrust enforcement—any of these could derail a platform play.

But betting against technology platforms finding ways around regulation is historically a losing proposition. They'll adapt, lobby, and innovate their way through.