In a 2004 business school class, I recall an esteemed (and knighted) guest speaker explain to us why Google's recent IPO would fail. His argument was straightforward: Google didn’t own the end-to-end customer journey; its entire purpose was to send you somewhere else. Therefore, every successful search was, by definition, a user leaving Google. Through the lens of classic strategy… you know, the kind built on decades of P&G and Coca-Cola case studies at HBS… this was a terrible business. You controlled the demand but gave away the destination with no brand lock-in, no owned experience, nor any switching costs.
The same logic would have killed Android before it launched - why give away a mobile operating system? Why let Samsung, Verizon, and HTC own the customer relationship? It would have killed Shopify too; why build infrastructure for millions of stores that compete with each other, instead of owning the storefront yourself?
The Atomic Assumption
These strategies made no sense under the legacy frameworks, because they were predicated on the assumption that products are atomic. A consumer does not assemble a beverage from one company’s carbonation, another’s sweetener, and a third company’s distribution. The product was the experience, and the experience was atomically consumed.
Therefore, when product is atomic, you must own the destination - to centralize demand, control the experience end-to-end, and capture value through attention and brand gravity. Every classical strategy framework assumes this.
Then, tech broke it
Whether hardware or software, the difference is that complex systems are inherently composable. Software products naturally decompose into interoperable layers (identity, payments, search, logistics, hosting, inference), while hardware carries a different, but analogous, value chain (design, fabrication, subsystem separation). What this does is divorce capabilities from experiences, and thus value migrates between layers of the stack.
That composability created a competitive dynamic that literally did not exist before tech.
Portals vs. Protocols
The Portal Strategy is the inherited playbook to centralize demand, own the destination, and capture value through attention, habit, and brand gravity. Come to us, and we'll provide whatever you need.
The Protocol Strategy is the tech-native alternative. Don’t own the destination, and instead become the embedded capability that every destination depends on. Show up everywhere, and provide the underlying service.
To be clear, by “protocol” I don’t necessarily mean an open standard like TCP/IP, but I do mean the strategic posture of becoming the infrastructure layer that others build upon - whether open or proprietary. By this definition, Stripe, Twilio, Shopify would all fit the definition of a "protocol".
The Strategic Karaoke Trap
When markets concentrate around a dominant portal, they drive the compounding effects of scale aggregation, user habit, and centralized network effects. Let's call this Scale Gravity. Then, when the portal market matures, Scale Gravity is what makes it functionally unbeatable on its own terms.
Yet, for some reason, challengers keep trying. AltaVista and MSN fought Yahoo for homepage relevance. Microsoft thought Zune would kill the iPod. Jet.com tried to build a better Amazon. Smaller streaming services attempt to out-bundle Netflix.
Predictably, they all fail the same way; they bring a slightly cleaner UI, a smarter algorithm, or a cheaper bundle. It's like bringing a knife to the proverbial gun fight, which in this case is our friend, Scale Gravity. Playing the incumbent’s game, on the incumbent’s board, with the incumbent’s pieces, is always going to be rough.
This is Strategic Karaoke, in that you’re not writing your own song; you’re singing along to the incumbent’s hit and hoping the crowd prefers your version. They won’t. They never do.
The instinct is understandable. Portal Strategy is the playbook CPG-heavy business schools taught, making it sound smart and provide defensibility to the board. Who could ever disagree with NOT wanting direct customer relationship?? It’s how every incumbent built itself, so it must be how you catch them, right?
“But our brand is our differentiator.”
In reality, if your competitor owns the demand, your brand is the consolation prize. Which is exactly why this move is both challenging and counterintuitive.
Shifting the axes of competition by giving up the one thing your b-school prof told you to protect - the direct customer relationship - feels like surrender. It feels like handing away the brand. But in a composable market dominated by a massive portal incumbent, the direct customer relationship is already gone. You just haven’t admitted it.
The choice isn’t between owning the customer and becoming infrastructure. The choice is between becoming infrastructure and becoming irrelevant.
Shifting the Axes of Competition
The right question isn’t “How do we build a better portal?” The question is “How do we make the destination matter less and the protocol matter more?”
We've seen this pattern recur in the tech industry for the better part of 30ish years:
- Search: Yahoo owned the homepage, but Google didn’t beat Yahoo by building a better Yahoo; it shifted value to the indexing and relevance layer beneath the homepage.
- Commerce: Amazon owns the destination. Shopify didn’t beat Amazon by building a bigger Amazon; it built the infrastructure that lets a million stores route around Amazon.
- Payments: PayPal owned P2P payments. Stripe didn’t beat them with its own competitor; it turned payments into an API.
- Communications: Telecos owned the end-to-end relationship with customers. Twilio turned communications into a programmable primitive.
The canonical pre-internet example is Intel. PC manufacturers (IBM, Dell, HP, Compaq, Gateway) competed at the portal layer by owning the consumer-facing brand, marketing, retail presence. Intel competed at the infrastructure layer, standardizing architecture and ecosystem compatibility. In a composable market, the layer beneath the portal can become more strategically valuable than the portal itself.
The AI Accelerant
If composability created Protocol Strategy, then AI is about to inject it with rocket fuel.
Portals win by owning the user’s single entry point. Historically, the destination app controlled the workflow because the human had to choose where to start. But AI agents don’t have a single entry point. Agents dynamically assemble workflows across providers. They don’t care about your homepage, your icon position, or your push notification cadence.
Humans care about portals, while agents care about capabilities, APIs, and machine-readable context. As agents increasingly orchestrate workflows on behalf of users, the structural advantage of portals (owning attention and habit) becomes disintermediated, and the portal’s grip on the user relationship weakens.
The pendulum swings again
The pendulum of power in tech swings in waves, and right now AI is swinging it violently back toward protocol economics. If you move to the infrastructure layer during this transition, you ride the wave with the wind at your back. If you fight it, you’re fighting gravity and the wind.
The window won’t stay open forever. The incumbents who currently sit on top of the portal economy are not stupid. They are racing to absorb the protocol layer themselves (Apple absorbing maps, payments, identity; Google absorbing inference; Amazon absorbing logistics). Every quarter you spend trying to win the portal game is a quarter the portal incumbents spend locking down the protocol game beneath you.
When Protocol Strategy Works (and When It Doesn’t)
Protocol Strategy isn’t a silver bullet. It works when:
- The market is highly composable.
- Ecosystems matter more than end-to-end control.
- Interoperability increases adoption.
It fails when:
- The product is genuinely non-composable (luxury fashion, Michelin-star restaurants, branded CPG).
- The customer experience can’t be separated from the capability.
- The infrastructure layer is already commoditized.
The Strategic Choice
If you’re an incumbent, your job is to defend the portal and refuse to lose the protocol layer beneath you. Apple has done this exceptionally well.
If you’re a challenger, competing against an incumbent portal, the picture is much simpler. You will not win the portal war. You have already lost it; the only question is whether you accept that or keep paying for the privilege of confirming it. The Protocol Window is open right now. The companies that shift to the infrastructure layer during this AI transition will define the next decade, while the rest keep paying their branding agencies.
Don’t build a smaller portal. Don’t hedge. Don’t ship the BlackBerry Storm.
