The End of Marginally Better?

Joan Westenberg
5 min readSep 23, 2023
Photo by ilgmyzin on Unsplash

Conventional wisdom has dictated that startups need to relentlessly optimise their products for maximum growth and monetisation. The accepted playbook involves meticulously fine-tuning onboarding flows, running drip marketing campaigns, A/B testing every page, and building “pro” features and premium subscriptions to capture revenue. This philosophy of iteration and incremental improvement has dominated Silicon Valley for nearly three decades under the assumption that the path to startup success requires launching products that are marginally better than the competition.

But the rapid ascendance of AI juggernauts like ChatGPT and Midjourney reveals the first cracks in this long-held belief. Despite little optimisation, intentional virality, or polished marketing, these companies gained tremendous traction in record time, signalling a monumental shift in the world of product development.

The Conventional Startup Playbook

Software startups are broadly engineered for rapid, data-driven growth. The prevailing mindset has been that great ideas alone are not enough — they should be translated into products that deliver maximum stickiness and monetisation. This has given rise to the “growth hacking” framework, the MVP / Lean Startup approach and an array of playbooks tailored to startup success.

Fundamental tenets: an obsessive focus on onboarding and first-run experiences to hook users quickly. Companies going through YC spend months fine-tuning initial user flows before launching publicly via gamified waitlists. Startups rely heavily on drip campaigns and lifecycle marketing to gradually guide customers through product features. Optimising sign-up funnels, engagement, and conversions are prime directives.

Nearly every screen and user journey is A/B tested extensively to enhance designs and copy incrementally. Building “pro” tiers with premium features is the go-to, monetising power users first before expanding to the masses. Startups turn growth and engagement into a science that flirts with alchemy, poring over analytics to squeeze every drop of optimisation possible.

This ethos took hold because it consistently worked. Focusing on frictionless onboarding and stickiness was critical to product-market fit in a world saturated with apps and services. Startups realised incremental improvements could mean the difference between breakout growth or fading into obscurity.

ChatGPT and Midjourney as Counterexamples

The launch trajectories of ChatGPT and Midjourney upend many of these conventions. ChatGPT debuted with no onboarding, marketing website, and sign-up flow for the first few days. Midjourney’s only access point was a Discord bot with no formal UI.

These sparse environments did nothing to impede explosive organic adoption. ChatGPT crossed one million users within five days, faster growth than any consumer app in history. Midjourney quickly accumulated over a million users through grassroots interest alone.

Different rules apply when core technology represents a seismic leap forward. ChatGPT’s natural language capabilities surpassed any previous conversational AI. Midjourney generated imagery of stunning detail and coherence. These products’ sheer novelty and utility were so compelling that refinement barely mattered.

They provide existential proof that a groundbreaking product can supersede the incremental gains startups typically fight tooth and nail for. Good ideas executed well do not require hundreds of performance optimisations. Traction can scale exponentially with little intervention when the experience exceeds people’s wildest expectations.

The problem: very few startups boast technical foundations capable of propelling this overwhelming early adoption, a particularly true statement in Web3. Without inherent viral momentum, conventional tech’s only option is a hard sell on a marginally improved experience, product or service.

Limitations and Open Questions

ChatGPT and Midjourney will likely evolve to incorporate a more conventional startup philosophy. ChatGPT now offers registration, documentation, and API access in response to surging demand. Midjourney is expected to launch more robust interfaces and access controls as well.

But their initial success triggers profound questions about product experience expectations in an age of rapidly advancing AI. Are their minimalist early days an aberration justified by exploding early traction rather than a new norm?

If software has seemed largely incremental for the past decade, ChatGPT and Midjourney provide a glimpse of paradigm-shifting innovation. Will users now expect most products to be lightyears ahead of alternatives rather than marginally better?

Based on AI integrations quickly built into every product from Notion to Grammarly — within five years, all software will incorporate some level of AI to meet basic expectations. If so, the playbook for startups may need to shift from iterating on existing ideas to building entirely new categories powered by AI. Disruption may supersede optimisation as the prevailing growth model in exponentially improving technology.

The coming wave of AI startups will provide more data on whether the ChatGPT and Midjourney launch models are unique cases or a broader turning point. If proliferating AI capabilities can reliably drive jaw-dropping early traction, tactics like growth hacking and nudging may lose relevance.

The next era of iconic companies could emerge — once more — from dorm rooms and small teams armed with powerful ideas rather than growth teams and optimisation roadmaps squeezed through the homogenising YCombinator meat grinder. But placing bets of any real magnitude on the cut-through power of a base technology alone is relatively risky. For most mortals, there are limits to dismissing the merits of operating at the point of being marginally better.

ChatGPT and Midjourney prove that with enough fundamental innovation, products can go viral on novelty alone rather than incremental optimisation.

Their trails to the top were so unorthodox that only some companies may be able to replicate the same path. The normalised playbook of growth hacking and marginal gains will likely be around for a while, even for startups with an AI edge. Customers are hungry for experiences that exceed expectations rather than marginally outperform competitors.

The era of dismissing conventional wisdom in favour of launching epoch-defining products built on technology alone has only just begun. To the bold founders taking that leap, the playbook as we know it need not apply.

I’m Joan. Transgender. Solopreneur. Tech writer. I help founders, CEOs, and leaders embrace emerging tech.

🍕 Newsletter: