What is a learning surface in marketing?

The number of places the market can tell you something true in a given week: sales calls you sit in on, stated lost-deal reasons, security review and RFP questions, support tickets, panels and briefings, replies to outbound, and analyst calls. A wide surface produces frequent small corrections from outside the building.

Why does planning not substitute for market contact?

Planning compounds your assumptions while contact corrects them, and both feel like progress from the inside. The common failure is not a bad plan but a plan whose central assumption was never exposed to anyone who could disprove it, then executed well for two quarters.

How do you widen a learning surface without hiring?

Attach yourself to conversations that already happen: sit in on two sales calls a week, change the lost-deal field from a dropdown to a sentence and read them yourself, ask one real question in every room you are already in, publish the argument before you are certain of it, and give every campaign one falsifiable claim.

What does not count as a learning surface?

More channels running the same untested assumption, more dashboards, more advice from people who operated under different conditions, and tools that summarize calls on your behalf. Aggregates hide the sentence that would have changed your mind.

← Writing

OperatingSep 2026 · 5 min read

Maximize your learning surface.

Disruption doesn't hand you certainty, it hands you contact in the wild. How fast you learn is set by how many places the market can correct you, not by how carefully you planned.

Paula Fontana

Founder & CEO, eudai

The short version

When the market is changing quickly you don't have historical data. You have contact with change, and usually not much.
Your learning surface is how many places the market can tell you something in a given week.
Most teams run 1 or 2, read them late, and hear about them secondhand.
Creating more touchpoints costs almost nothing. Most of it is sitting in on conversations that already happen.
More channels, more dashboards and more advice don't count.

Whenever change happens, whether that's a new product, a market being reordered, a regulation landing, or a category nobody has named yet, the information you want doesn't exist. You can't research your way to it, and the plan you write in month 1 is a set of assumptions with dates on it.

What you can change is how many places the market gets to correct you. That's your learning surface, and most teams leave it alone while they rewrite the plan.

What counts

A surface is anywhere a real buyer tells you something without being asked to imagine anything.

Sales calls you sit in on. Not the recap.
Why a deal was lost, in the buyer's words, not a CRM dropdown.
The questions that keep coming back in security reviews and RFPs. Those are your objections.
Support tickets and onboarding calls.
Panels, roundtables and briefings.
Replies to outbound, including the rude ones.
Analyst and press calls, where you hear your framing come back with the weak parts missing.

Two of these running, both read a month late, is common.

You can't speed up the learning. You can widen the surface.

Two diagrams. On the left, a single line runs from a box labeled you to one contact point on the market, labeled 1 surface, a report once a quarter. On the right, nine lines fan out from the same box to nine contact points across the market, labeled 9 surfaces, corrections every week. — Same market, same month. The difference is how many places it can correct you.

Planning isn't a substitute

Planning stacks up your assumptions. Contact with your market knocks them down. From the inside both feel productive, so teams under pressure pick the one they can do easily by themselves.

The problem usually isn't a bad plan. It's a plan whose main assumption nobody outside the company ever saw, executed well for 2 quarters. That's the argument for evidence before spend: not to slow you down, but to get the cheap correction in before the expensive part starts.

Widening it without hiring

Almost none of this is new activity. You're attaching yourself to conversations that already happen, and writing down what you heard in the words you heard it.

Sit in on 2 sales calls a week. Live, and not as a participant.
Make the lost-deal field a sentence instead of a dropdown, then read them yourself on Fridays.
Ask 1 real question in every room you're already in, and let the answer be inconvenient.
Publish the argument before you're sure of it.
Keep a file of the exact words buyers use, separate from the words you use.
Give every campaign 1 claim that can be proven wrong.

Publishing is the one teams skip. A piece that takes a position gets corrected by people who'd never take your call, which is part of why launching the same thing again keeps working.

When your surface is too narrow

Every piece of evidence traces back to 1 channel, 1 customer or 1 enthusiastic advocate.
Your framing hasn't changed in 2 quarters, and nothing has disproved it either.
You can't remember the last thing a buyer said that surprised you.
Arguments about the market get settled by whoever is most senior.
You spend more time on the wording than on hearing how it resonates with actual customers.

That last one gives it away. When the surface is narrow, copy is the only thing left to argue about.

What doesn't count

Widening isn't adding. Four more channels pushing the same untested assumption gives you four more ways to be wrong at scale. That's the volume trap again.

More dashboards. Averages hide the one sentence that would have changed your mind.
More advice. People who ran the play under different conditions aren't market contact.
More tools listening for you. A summary of a call isn't the call.
Going back to what used to work, which is contact with a market that moved on.

Both feel like activity. Only one tells you something you didn't already think of. The trade between running what works and testing what might is old and well studied, and explore and exploit is a decent way to challenge it: some of the budget has to go on finding out.

How you know it's working

A wide surface doesn't feel like confidence. It feels like being corrected often, in small ways, by people who don't work for you.

You change a line in the pitch because a buyer said something, not because a colleague preferred it.
You can name the week you dropped a claim, and why.
Buyers start using your framing, in their own words, without being prompted.
The surprising input shows up from more than 1 direction. That's when it's a pattern instead of an anecdote.

None of this makes a disrupted market settle down faster. It changes what you do while it moves: collecting corrections instead of defending the plan. That's the difference between a message that's slow and a message that's wrong.

Part of our work on category creation in security.