What Makes SaaS Win Now: AI, Analytics, and Economics That Scale

AI is making SaaS easier to build.

It is also making SaaS harder to differentiate.

That tension runs through this edition.

Instead of asking who has the most AI features, I think there are three better questions to ask:

Those questions get closer to what will separate durable SaaS products from the ones that simply look modern today.

This week, we look at what AI-first really means, why analytics so often fails to become a competitive advantage, and how pricing can quietly undermine a platform as it scales.

Let’s get into it.

Delete the AI from your product. What breaks?

If nothing does, you’ve shipped AI features but you haven’t built an AI-first product. That distinction matters more than the label suggests.

Most software products have been built on two pillars:

Those two pillars produced nearly every successful SaaS product of the last twenty years, and neither one is going anywhere.

An AI-first company has three, and the third differs in kind, not degree.

AI brings intelligence and, by nature, is different from data or software. It participates in decisions once reserved for people.

Uber's pricing illustrates this well because the mechanism is visible to anyone who has ordered a ride twice in one day. The price dynamically changes.

Uber initially used algorithmic software to determine the ride rate. Some years ago, they moved to a hybrid model, so a model sets the rate in real time, and the algorithm weighs conditions and approves it. 

As the AI model improves, Uber will trust the AI to make the decision. When it prices badly, that outcome feeds back, and the system adjusts. 

Here’s the difference: Deterministic software has no equivalent capacity. It repeats the same error until a person catches it and rewrites the logic.

Once intelligence sits inside the system, the design question changes shape. It stops being whether to add AI and becomes how much authority AI should hold in a given decision.

Should it suggest and leave the call to a person? Approve or reject within boundaries you define? Act autonomously? Different features deserve different answers.

Plenty of companies have AI features their customers use daily and would notice if they vanished.

Doesn't that count?

Run it against the definition.

A summarization feature reads what is already there and tells you faster.

A chat interface changes how you ask a question, not what happens after you ask it. Useful, both of them.

But every decision sits exactly where it sat before, with a person. Nothing has been delegated. Nothing in the system can be wrong in a way it can learn from.

That is a two-pillar product with a faster interface, and calling it AI-first is not a harmless exaggeration. It costs you the ability to see how far you actually have to go.

AI-first means something narrower.

A company willing to let AI participate in decisions for its customers, but not in how it runs itself, hasn't finished the thought.

Key Takeaways

Analytics can create real customer value without ever becoming a reason to buy. Here is why embedded analytics so often helps after the sale but fails to influence the decision before it.

Many SaaS companies invest heavily in analytics because customers use it to understand performance, make decisions, and work more effectively.

But usefulness is not the same as differentiation.

For analytics to create competitive advantage, buyers have to see it as a reason to choose the product over an alternative. Most SaaS companies never make that transition.

Buyers rarely remember an entire feature list. They reduce an evaluation to a few clear reasons why one product is better.

“Good reporting” is not much of a reason.

“This product helps us identify problems earlier and act faster” is.

Analytics matters competitively only when it becomes part of the buyer's explanation for choosing the product.

Dashboards, charts, filters, and reports are useful, but buyers now expect them.

When every product presents analytics in roughly the same way, the experience becomes easy to appreciate and hard to remember.

Analytics becomes an advantage only when it enables something competitors cannot easily replicate.

Many demos focus on what the analytics displays rather than what it changes.

The stronger story is not another dashboard. It is how the product helps users detect an issue, understand why it happened, decide what to do next, or take action faster.

Show information and buyers see reporting.

Show what becomes possible because of the analytics and buyers see differentiation.

If buyers first hear the analytics story during the demo, they may already see it as supporting functionality.

Differentiated analytics should appear much earlier, across the website, discovery conversation, sales narrative, and positioning.

The demo should prove the story, not introduce it.

Competitive advantage is built across multiple interactions.

If marketing describes analytics as strategic while sales treats it as reporting, buyers receive two different stories.

The strongest companies reinforce the same reason to believe across marketing, sales, demos, and product experience.

Three things have to happen:

First: Build something genuinely different.
Basic reporting is rarely enough. The analytics must enable capabilities buyers cannot easily get elsewhere.

Second: Make the difference usable.
A capability only matters when customers can experience it inside the product.

Third: Make the story impossible to miss.
Marketing, sales, demos, and product conversations should all reinforce the same differentiated value.

Miss one of those, and analytics may still create customer value without creating buying advantage.

The mistake is assuming customer value automatically becomes competitive advantage.

It does not.

Analytics becomes an advantage when buyers can clearly explain how it makes the product better than the alternatives, and why that difference matters to their business.

Key Takeaways

Many embedded analytics purchasing decisions focus on feature comparisons and initial licensing costs. While those factors matter, they overlook a critical question:

Will the economics still work when the business scales?

The biggest cost risk often comes from pricing model misalignment. A platform that looks affordable during procurement can become significantly more expensive once analytics is rolled out across tenants and users.

As adoption expands, pricing models such as per-user licensing, usage-based pricing, consumption thresholds, and tiered licensing can become disconnected from the way SaaS businesses generate revenue.

Hidden costs often appear through add-ons, service requirements, or pricing tiers that only become relevant as adoption grows. By the time they become visible, they can be difficult to avoid.

Organizations should evaluate embedded analytics the same way they evaluate any strategic investment: by understanding how costs behave at scale.

Questions worth asking include:

The best analytics economics are predictable. When costs scale in a way that supports business growth, analytics becomes a driver of expansion rather than a source of margin pressure.

Key takeaways

The next generation of SaaS will not be defined by how much AI a company can add.

It will be defined by what that AI is trusted to do, whether analytics gives buyers a reason to choose the product, and whether the economics still work as adoption grows.

AI tests the intelligence. Analytics tests the differentiation. Economics tests the model.

The strongest SaaS companies will connect all three.

That is a much higher bar than simply adding AI, but it is also where the real advantage will come from.

See you in the next edition,

Arman Eshraghi
The Embedded Intelligence Brief
Connect on LinkedIn

P.S. We’re looking for SaaS product and engineering leaders to join our October virtual panel with Dresner Advisory on Building SaaS with AI. If you have practical lessons to share, reply to this email and we’ll take it from there.