Example post · Noah Ullman

Products, Signals, and Jamming

Every good product is basically an antenna.

It receives a signal from the person using it. That signal is constant and information-rich: the customer pays, or they don't. They come back, or they don't. They tell their friends, or they don't. They complain about specific things, or they quietly leave. All of that is signal. And when the product team receives it clearly, they tune the product to match. Quality improves. Fit tightens. The product gets better because it's listening to the right broadcast.

This is the natural state of a healthy market. Customer broadcasts, product receives, product adapts. Two bodies in a clean signal loop. And the products that come out of this loop tend to be great, because there's nowhere for the signal to go except straight into the thing being built.

Restaurants work like this. You eat the food. You pay for the food. If it's bad you don't come back, and the restaurant either fixes it or closes. Consumer software works like this. You download the app. You use it. If it sucks you delete it. The team ships an update or they lose you. Warby Parker worked like this. They sold glasses directly to the person wearing them, cut out the middlemen, and built a product shaped entirely by the gravitational pull of their actual user. Price dropped, quality stayed high, fit improved.

Every great DTC brand of the last fifteen years was, at its core, a company that restored a clean signal path in a market where it had been jammed.

The signal is the product's teacher

Here's the thing most people miss about product quality. Quality is not a decision someone makes in a conference room. Quality is an emergent property of the signal loop. When the loop is tight and clean, quality converges toward the customer's actual needs almost automatically. When the loop is broken, quality drifts. It doesn't matter how talented the team is or how much money is in the system. A product tuned to a bad signal will produce a bad product. Every time.

You can describe the strength of this loop along a few dimensions:

Directness of payment: When the person using the product is the same person paying for it, the signal is strong. Every intermediary between use and payment degrades it.
Observability: When the customer can tell whether the product is good, they broadcast a clear signal. You know within hours whether your new phone is good. You may never know whether your insurance was priced correctly. Clear signal versus noise.
Switching cost: When the customer can leave easily, the signal carries consequences. The product has to earn its keep every cycle. When switching is hard, the signal still broadcasts but nobody has to listen. It's like screaming into a wall.

Markets where all three are strong produce the best products on earth. Markets where all three are weak produce products that make you want to throw things.

This has implications for how you grow, too. Every business uses accelerants: promotions, channel partners, distribution deals, enterprise contracts. These are fine. They can be great. But they are only safe when they accelerate the rate at which you receive product signal from your actual customer. A promotion that gets more of the right people using your product faster is an accelerant. A channel partner that puts you in front of customers who will tell you what's working and what isn't is an accelerant.

The danger is when the accelerant becomes the whole GTM strategy, because then you start tuning the product to the accelerant instead of the customer.

Take healthcare. A common early move for digital health companies is to sell through self-insured employers. It makes sense on paper: big contracts, guaranteed volume, someone else handles distribution. But the employer is not the patient. The employer's signal says "reduce claims cost, show engagement metrics, make HR's life easier." The patient's signal says "help me manage my A1C and stop feeling like garbage." Those are not the same signal. And the product team, receiving the employer's signal at full volume and the patient's signal through a wall, will tune to the employer. They always do. Six months later you have a product that generates beautiful dashboards for benefits managers and does almost nothing for the person who's actually sick.

Use accelerants to get to your customer faster. Never use them as a substitute for hearing your customer directly. The moment your distribution strategy introduces a new signal source with more economic authority than your end user, you have a signal attenuation problem. You just built it yourself.

Now add a third body

In a growing number of markets, the person using the product is not the person paying for it. Someone else sits in between. An insurer. An employer. A government program. A procurement department. This third party has its own priorities, its own economics, its own definition of what "good" means.

And the moment it enters the system, it jams the signal.

The customer is still broadcasting. They're still experiencing quality, or the lack of it. They still have preferences. They still know when something is wrong. But the third-party payer absorbs that signal, filters it through its own incentive structure, and rebroadcasts something completely different to the product. The product receives this new signal and, because that's where the money comes from, tunes to it.

Over time the product becomes exquisitely well-calibrated to a signal that has almost nothing to do with the person actually using it.

This is the core mechanism. It's not that the system becomes chaotic or unpredictable. It's actually extremely predictable. The product always drifts toward whoever is writing the check. The customer's signal gets attenuated. The payer's signal gets amplified. And quality, real quality, the kind the customer actually experiences, erodes steadily because nobody with economic authority is measuring it.

Healthcare is the ultimate jammed signal

The American healthcare system is the most expensive signal attenuation problem in human history.

The patient uses the product (clinical care). The insurer pays for it. The provider delivers it. Three parties, three incentive structures, three definitions of quality, none of them aligned.

The insurer defines quality as cost containment, claims processing efficiency, and compliance with administrative protocols. The provider defines quality as reimbursement optimization, which means billing codes, procedure volume, and documentation that satisfies the payer. The patient defines quality as: did someone listen to me, did they understand my situation, did I get better.

Guess whose signal gets attenuated.

The patient is still broadcasting. They're filling out satisfaction surveys that nobody reads. They're switching doctors and nobody asks why. They're getting worse and the system codes it as a new episode rather than a failure of the last one. The signal is there. It's just not reaching anyone with the authority or incentive to act on it.

Meanwhile the provider is receiving the payer's signal at full strength. Every reimbursement rule, every prior authorization requirement, every HEDIS metric is a clear, high-fidelity instruction from the payer about what the product should optimize for. And the product obeys. Of course it does. That's where the money comes from.

So you get a system where the number of hospital beds per thousand residents has dropped by nearly half since 1980 while the number of hospital employees per bed has almost doubled. More people doing more work on fewer patients, producing marginally better outcomes at dramatically higher cost. That's what a product looks like when it's been tuned to the wrong signal for forty years.

This shows up everywhere

Education has the same problem. Federal student lending created a three-body dynamic between students, institutions, and the government. The student is the user. The government is the payer (via guaranteed loans). The institution delivers the product.

Once federal money started flowing, institutions began tuning to the payer's signal: enrollment volume, which captures federal dollars. The student's signal, which would have said "teach me something useful at a reasonable price," got attenuated. Tuition has risen at nearly three times the rate of inflation. For-profit colleges, the most aggressive optimizers of the enrollment-funding loop, produced some of the worst educational outcomes imaginable. They were beautifully tuned products. Just tuned to the wrong signal.

Enterprise software is a milder version of the same thing. The buyer is a VP or CIO. The user is some person on the front lines who has to actually click the buttons. The product tunes to the buyer's signal: security features, compliance checklists, integration specs, things that look good in a procurement deck. The user's signal, which would say "please just make this fast and not make me want to die," gets filtered out.

This is why consumer software is generally beautiful and enterprise software is generally miserable. Same industry, same talent pool, same technology. Different signal paths.

Why these markets resist reform

Every few years someone tries to fix healthcare by adding a new layer of incentive alignment on top of the existing structure. Managed care. Pay for performance. Value-based contracts. Accountable care organizations. Each one introduces new metrics, new administrative infrastructure, new rules.

None of them work, and the reason is structural. You cannot fix a signal attenuation problem by adding more signal processing. Every new metric, every new compliance requirement, every new quality framework is just another filter between the customer and the product. The signal degrades further. The product tunes to the new filter. The customer's experience doesn't change.

The payer's signal is also self-reinforcing. As healthcare gets more expensive, people become more dependent on insurance. As insurance covers more, patients become less price-sensitive. As patients become less price-sensitive, prices rise. The payer's signal gets louder. The customer's signal gets quieter. The gap widens.

This is why throwing money at these markets doesn't help. The money flows through the same attenuated signal path and gets absorbed by the same incentive structure. More money in a jammed signal system just makes the jamming louder.

The fix is always the same

The most effective interventions in attenuated markets are not incremental reforms. They are structural reconfigurations that restore the direct signal path between customer and product.

Direct primary care does this. Patients pay a monthly membership directly to their physician. No insurer in the middle. The doctor's revenue comes from the patient. The patient evaluates quality through their direct experience. Signal path: clean. The product (care) immediately starts tuning to the customer again. Physicians in these models spend more time with patients, less time on admin, and both sides report higher satisfaction.
Product-led growth in software does this. When the end user can adopt software without a procurement process, the product has to earn its keep with the person clicking the buttons. Slack, Figma, Notion all grew by making the user the signal source. The product got good because it was listening to the right broadcast. It feels like more than a mere coincidence that once these companies started building out enterprise sales, their products started to suck.

The pattern is always the same. Find the attenuated signal. Identify who's jamming it. Remove them from the path, or build around them. Let the customer's signal reach the product at full strength.

The rest takes care of itself. A product that receives a clean signal from its customer will, over time, become a good product. A product that receives a jammed signal from a third-party payer will, over time, become a product that serves the payer at the customer's expense. This is as close to a law of nature as anything in business.

The quality of a product converges toward the needs of whoever is sending the signal it's tuned to. Choose your antenna carefully.