The signals satisficing misses

SpaceX filed. Anthropic is rumored close. OpenAI restructured toward it. The next eighteen months will see more consequential IPOs than any period since 2021, and the marketing teams at companies watching from the sidelines are asking the wrong question.

The wrong question is: how do we compete for attention against that noise? The right question is: what kind of company gets built when attention is this fragmented?

We've been watching journeys across a dozen B2B companies preparing for 2026 planning, and here's what we keep noticing: the ones growing steadily aren't optimizing harder. They're reading differently.

One example, anonymized but real. A mid-stage infrastructure company noticed their demo requests had flatlined despite traffic holding steady. The obvious read: conversion problem, fix the CTA, test the form length, maybe add social proof above the fold. Standard playbook. They ran those experiments for six weeks. Nothing moved.

When they looked at the journeys instead of the funnel, they saw something else. The people who eventually became customers weren't hitting the demo page first. They were reading the same three documentation pages, in the same order, usually across multiple sessions over two to three weeks. Then they'd come back, skip the marketing site entirely, and book through a direct link someone on their team had shared.

The demo form wasn't broken. It was irrelevant to the actual buying motion.

This is the kind of thing satisficing misses. Herbert Simon's term for how humans make decisions under uncertainty—we don't optimize, we find something good enough and stop. Most marketing analytics are built for satisficing. They give you a number that's good enough to act on, and you act. Traffic up? Good. Conversion down? Bad. Fix the bad number.

But the number doesn't tell you whether the people it's counting are the same people who eventually buy. It doesn't show you the three-week documentation arc. It doesn't reveal that your funnel is measuring the wrong motion entirely.

This matters more when attention is expensive and scattered. SpaceX and Anthropic and OpenAI will soak up enormous amounts of the conversation in their categories. Companies adjacent to those categories—infra, dev tools, security, anything AI-adjacent—will feel the compression. The temptation will be to spend more, optimize harder, run more experiments on the same tired surfaces.

We think the companies that pull ahead will do something different. They'll get patient about what they're actually seeing. They'll stop treating every session as independent and start reading them as chapters in a longer story. They'll notice that their best customers don't behave like their funnel assumes.

The interpretation layer is the part most tools skip. They collect the data, they visualize the data, they let you slice the data fourteen ways. But they don't help you read it. Reading means asking: what is this person actually trying to figure out? What does the sequence tell me that the snapshot doesn't? Where is the journey surprising, and what does that surprise mean?

A small counter-observation: some businesses really do have a conversion problem, and the funnel really is the right unit. Transactional e-commerce, high-volume low-touch SaaS, anything where the buying motion is compressed into a single session. For those, optimize away. But if your sales cycle is longer than a week and your product requires explanation, the funnel is probably lying to you about where the action is.

2026 is going to reward companies that know the difference.