Navigating the SaaS-pocalypse Part 3: Winning In The Long Term

This is the third and final post in our SaaS-pocalypse series.

A quick recap:

• In Part 1 we covered how the rules of the game that we all play aren't changing, how you play it is.

• In Part 2 we covered how AI impacts the short-term tactics that keep you alive.

Now we're covering what makes you succeed and hopefully not die in the long term.

Will all SaaS companies die? Probably not.

Will some of them die? Probably yes.

We're going to use the 7 powers framework from Hamilton Helmer's book. There are a million bad books on strategy. This is one of the best ones.

As we covered previously, all business is fundamentally arbitrage, unless you have a way of spending a dollar and getting more than a dollar back, you don’t have a business.

The best way of thinking about strategy is how you protect and increase the arbitrage you've created.

This framework is a great lens through which to look at that.

AI is asymmetrically reshaping the seven powers. Some are collapsing. Some are strengthening.

Knowing which is which determines whether you grow, shrink or die.

The Obvious Points - Codebases and Engineers

To get the obvious things out of the way, the relative value of complex codebases and engineers is shrinking fast.

Once upon a time (i.e., six months ago) a company's codebase and a large engineering team gave them inherent advantages in a market.

A million lines of code, years of accumulated logic, thousands of edge cases handled used to be a moat

Engineers were also strategic resource because they were scarce and expensive, and the thing they built — the codebase — was very expensive to replicate.

AI has completely changed this dynamic.

These large code bases and large teams working in old processes are maybe at best a net neutral, possibly a disadvantage now.

Lets go through each power

1. Scale Economies → Eroding for Software Companies

Definition: Per-unit costs decline as volume increases. Bigger = cheaper per customer.

Examples:

• AWS spreads massive infrastructure capex across millions of customers. No smaller provider can match their unit economics.

• Walmart's distribution network gets cheaper per unit with every new store.

• Salesforce spreads its R&D across 150,000+ customers. The pattern is the same: fixed costs divided by a massive base creates a structural cost advantage.

Before AI: In SaaS, scale economies lived primarily in engineering.

You spread your R&D cost — your 200-person engineering team — across your customer base. The bigger the base, the cheaper each customer's share of that R&D.

More R&D budget → more engineers → more features → more/happier customers

Now: AI compresses the cost of building software so dramatically that this advantage shrinks.

A 2-person team with Claude Code can ship in a week what previously took 200 people to ship in a year.

Infrastructure and compliance still scale. Good go to market still scales.

But the R&D cost moat — which was the primary scale advantage for most SaaS companies — gets much shallower.

Your large engineering team is no longer a moat. It might even be a liability if it slows you down.

The advantage shifts to whoever can direct AI-augmented teams most effectively — which is about process and judgment, not headcount.

2. Network Effects → Holding + Data Network Effects Become More Valuable

Definition: The product becomes more valuable as more users adopt it. Every node in the network raises the usefulness of the whole network.

Examples:

• Slack becomes more valuable to you as a user the more of your colleagues join.

• Uber gets more valuable with more drivers (shorter wait times) and more riders (more demand for drivers).

• Airbnb listings attract guests, guests attract hosts.

• Any social network is more useful to you the more people that you know who are on it.

Before AI: Network effects are hard to build, but extremely defendable and make companies very valuable. Social networks, marketplaces, these all exhibit classic network effects, which just copying the codebase won't replicate.

Slack doesn't get less valuable because AI exists. Uber's driver network doesn't erode because someone can vibe-code a ride-sharing app.

NFX has an excellent post on network effects and the 16 different types.

Now: Data network effects have always existed but AI makes them more valuable to more companies.

A lot of companies in the past have claimed that their data is valuable, and most of them were lying.

Companies that use customer interactions to improve their AI features build compounding advantages.

More usage → better AI → more usage → better AI. This now applies to more companies than ever before.

Companies sitting on proprietary interaction data, especially in valuable niche’s, now have a path to network effects they never had. If anything, this power may have gone up in the ranking.

3. Counter-Positioning → The Biggest Opening for Attackers

Definition: A newcomer adopts a superior model that the incumbent can't copy without materially damaging their existing business.

Examples:

• Figma's browser-based collaborative design counter-positioned Adobe — Adobe couldn't cannibalize its desktop Creative Suite licensing to match.

• Netflix's delivery model counter-positioned Blockbuster — Blockbuster couldn't go delivery-only without destroying its store business.

• Zoom counter-positioned Webex and Skype by being dead simple and free to start — incumbents couldn't match the price without cratering their enterprise revenue.

Before AI: Counter-positioning existed but was rare because building the alternative was expensive.

Figma needed years of engineering investment to build a credible browser-based design tool.

Netflix needed massive streaming infrastructure.

The cost of building a competitive product was itself a barrier — you needed serious funding and engineering talent just to get to the starting line, which limited the number of attackers who could attempt it.

Now: This is the single biggest structural threat to mid-market SaaS. Three forms of counter-positioning are opening simultaneously:

1. Micro-SaaS replaces bloated platforms. A vibe-coded tool that does 20% of Salesforce for 5% of the price. Salesforce can't race to the bottom without cannibalizing revenue.

2. Custom-built replaces off-the-shelf for skilled companies. Companies vibe-code internal tools tailored to their exact workflows. Why pay $50k/year for horizontal software you can build in a week?

3. Usage-based AI-native tools vs. seat-based pricing. Incumbents locked into per-seat revenue can't shift without cannibalization.

Its early, but customer expectations are shifting from "a place for me to create" to "do the work for me."

Incumbents built around the old expectation face a classic innovator's dilemma. Harvey (legal AI) and Cursor (coding AI) are pure-play challengers exploiting exactly this.

4. Switching Costs → Under Serious Pressure

Definition: The pain of switching to a competitor creates retention even when alternatives are comparable.

Examples:

• Workday's deep integration into payroll, benefits, compliance, and org structure makes switching a multi-year, multi-million-dollar project.

• SAP implementations take years and cost millions — nobody switches on a whim.

• Even something as simple as switching from Slack to Teams involves migrating years of conversations, channel structures, and integrations.

Before AI: Switching costs were the default SaaS moat for two decades. Once a customer was integrated, trained, and had years of data in your system, leaving was so painful that they stayed even when frustrated. The deeper you embedded, the stickier you got.

Now: AI attacks all three traditional sources of switching costs:

1. Data migration. AI can parse, restructure, and migrate data between systems in days instead of months.

2. Workflow replication. An LLM can study how you use Tool A and replicate those workflows in Tool B. Or build you a custom tool that matches your exact workflow.

3. User retraining. Natural-language interfaces make all tools feel similar. AI assistants guide users through new UIs. The learning curve — which used to be a moat — flattens.

Switching costs were the default SaaS moat for the last two decades. That's crumbling.

If your retention strategy is "it's too painful to leave," this might not last.

You need to shift to "it's too valuable to leave" — proprietary data, AI that gets smarter with use, ecosystem integrations that add value.

5. Brand → More Important, Not Less

Definition: The ability to charge a premium compared to similar products with the same features rooted in trust/name recognition.

Examples:

• Stripe charges comparable rates to other payment processors, but its brand for developer experience and reliability wins deals where raw product differences are minimal.

• Apple charges a massive premium for hardware that benchmarks similarly to competitors.

• Shopify has become synonymous with "start an online store" — that's brand, not features.

Before AI: Brand has always mattered, but features also mattered won.

In SaaS especially, buyers could evaluate products directly through trials and demos.

The better product typically beat the better-known one. Brand was a tiebreaker in close decisions, not the primary purchasing driver.

Now: Brand is more important as a power when AI floods every category with new and similar entrants.

When anyone can ship an app, customers need trust signals to decide what's reliable, secure, and maintained.

Enterprise buyers won't trust a vibe-coded tool with compliance data without a strong brand behind it.

As mentioned in Part 2, some acquisition channels are likely getting more expensive, which is going to raise CAC and put pressure on margins.

However, the best channels - word of mouth, organic traffic, community, referral loops - aren’t going to deteriorate as much when a hundred new competitors show up.

This is where brand really helps.

Brand both allows you to charge a higher price and lowers your customer acquisition costs, this improves your unit economics from both sides.

Higher revenue per customer, lower cost to acquire them. In a world where everyone can build the same features, that economic advantage compounds.

Invest in brand now, while your competitors are still focused on features.

6. Cornered Resource → Shifts from Code to Data & Distribution

Definition: Preferential access to a valuable asset that competitors can't replicate.

Examples:

• Bloomberg Terminal — the moat isn't the software, it's the proprietary financial data feeds and physical space on the floors of traders

• Google's search index is a cornered resource no competitor can fully replicate.

• Oil companies have exclusive leases to land that competitors can’t drill on

Before AI: High quality engineers were the cornered resource.

Not just any engineers — good ones with specific skill sets. In the early 2010s, if you had a team of iOS engineers, that was a genuine competitive advantage.

There weren't enough to go around. Companies hoarded mobile talent because the ability to ship a native app was itself scarce.

Same thing happened with ML engineers, data engineers, DevOps engineers at various points.

The talent was the moat. And the thing the talent built — the codebase — was also a cornered resource.

Now: That's over. LLMs can replicate most application logic. The scarce skill is no longer "can write code" but "knows what to build and why."

The cornered resources that matter now:

• Proprietary datasets that make AI features uniquely good

• Distribution and audience — newsletter subscribers, community members, captive attention

• Domain expertise embedded in AI — fine-tuned models trained on proprietary workflows

• Regulatory moats — compliance certifications, government approvals

Audit your actual cornered resources today. If the answer is "our codebase" or "our engineering team," you should think again.

7. Process Power → The Sleeper Advantage

Definition: Embedded organizational routines that enable superior output, but are so complex and tacit that competitors can't copy them even when visible.

Examples:

• Toyota Production System — competitors have studied it for decades and still can't replicate it.

• Netflix's culture of freedom and responsibility combined with chaos engineering produces deployment reliability competitors can observe but consistently fail to replicate.

• Apple's design culture: they can reliably produce hardware devices that people love using.

Before AI: Process power in software was mostly about culture and judgement within various R&D teams such as product, design, engineering, etc.

It was real but slow to build, and the payoff was incremental. Most SaaS companies didn't think of their internal processes as a competitive advantage.

Now: This is the sleeper. Companies that develop superior processes for using AI compound advantages that competitors can't copy by buying the same API.

The process of how you integrate AI into your org — not just whether you do — becomes the differentiator.

The process isn't the tool — it's how you evaluate outputs, how you design human-in-the-loop workflows, how you systematize prompt engineering, how you build feedback loops between AI output and business metrics.

But here's the thing about process power: it's relative to your peers.

You have to stay the most sophisticated team at a given use case.

If your competitor is iterating on their AI workflows faster than you, your process power erodes.

Toyota didn't just build a great production system once — they kept refining it while competitors tried to catch up. Same applies here. The compounding only works if you keep compounding.

So What Do You Do With This Information?

The companies that win through this shift won't be the ones with the best features. They'll be the ones with the most durable advantages in a world where features are now cheap.

Audit your business against these seven. Be honest about which ones you actually have vs. which ones you think you have.

Build so you’ll can thrive in this period, not just survive.

Good luck out there,

Dan

P.S - If your product does $2M ARR+ but your MRR, churn, or ARPU numbers aren’t strong enough, reach out. I help successful product scale revenue. Book a free strategy call here

Free Resources

Growth Ceiling Calculator 📊

MDE Calculator🧪

Newsletter Archive 📰