PETRICHOR.
Menu

Founder Question

Should I Focus on Distribution or Keep Improving the Product?

Updated July 19, 20266 min readBy

TL;DR

Focus on the bottleneck that is limiting learning. Improve the product when qualified users reach it, understand it, and still fail to adopt or retain. Improve distribution when the right users rarely encounter it or arrive with the wrong expectation. Most startups need a weekly system that tests both, with one declared constraint receiving most resources.

Key claims
  • Product and distribution should be diagnosed as connected learning loops.
  • Qualified retention problems point toward product, promise, or onboarding work.
  • Low qualified reach points toward distribution, positioning, or channel work.
  • More traffic can hide a weak product by increasing noisy feedback.
  • More product work can hide distribution fear by postponing market contact.

A startup can spend six months polishing a product nobody encounters. It can spend the same six months buying attention for a product nobody keeps.

The choice between product and distribution is not philosophical. It is a constraint diagnosis.

Distribution Mastery treats market contact as part of the product learning system. The right channel delivers users, context, objections, language, and evidence. The product turns that contact into value or exposes where the promise breaks.

Find the leak in the buyer path

Map five observable stages:

  1. a qualified buyer encounters the company;
  2. the buyer understands the problem and promise;
  3. the buyer starts the intended action;
  4. the user experiences the promised value;
  5. the user returns, pays, expands, or recommends.

Low performance at the first stage points toward reach. Failure between encounter and action points toward positioning, trust, offer design, or onboarding. Failure after real use points more strongly toward the product or a promise that attracted the wrong user.

Do not diagnose from total traffic. Segment by source, buyer fit, use case, and expectation. A channel full of unqualified visitors creates numbers without useful evidence.

Product work is the priority when qualified use fails

Prioritize the product when the right users can be recruited, understand the promise, and still do not reach value.

Look for repeated friction in activation, core-task completion, time to value, reliability, retention, and willingness to pay. Interview users around observed behavior. “Would you use this?” is weaker than watching where a real attempt stops.

Product work can include simplifying the promise or onboarding. A product that delivers value after ten steps has already lost the buyer who expected three. The constraint lives in the full experience, not just the feature set.

Distribution work is the priority when qualified reach is scarce

Prioritize distribution when a meaningful set of users retains, pays, recommends, or pulls the product into their work, but few similar people arrive.

The task is not “post more.” Identify where qualified buyers already learn, whom they trust, which trigger makes the problem urgent, and which proof reduces their risk. Choose a channel the team can repeat and improve.

Early distribution can be manual. Founder outreach, partnerships, events, communities, customer referrals, and highly specific content produce better learning than a broad campaign. Scale follows a working pattern.

Watch for the two avoidance loops

Product avoidance

The team increases traffic to avoid confronting weak retention. Each campaign creates a temporary spike, then the cohort fades. More reach magnifies the leak and makes acquisition look like the whole problem.

Pause scale. Recruit a narrow cohort, observe use, and repair the moment where expected value fails.

Distribution avoidance

The team keeps adding features after a small group already gets strong value. Product work feels concrete and controllable. Market exposure creates rejection, so the launch remains one improvement away.

Freeze nonessential scope. Put the current product in front of a defined buyer with a precise promise and record what happens.

Use a two-loop weekly operating system

Keep both loops visible:

Loop Weekly question Useful signal
Product Did qualified users reach and repeat value? Activation, retention, paid use, task success
Distribution Did more qualified buyers encounter the right promise? Qualified conversations, response, referral, conversion

Declare one primary constraint for the week. Give it most of the team's attention. Keep a smaller test alive on the other side so the diagnosis does not go stale.

At the review, ask what evidence would make the constraint move. If qualified acquisition improves and retention collapses, shift to product. If retention becomes strong and the pipeline stays empty, shift to distribution.

Evaluate channels by learning quality

A channel is valuable early when it brings the intended buyer, preserves the intended context, allows direct observation, and can be repeated. Raw reach matters less than the quality of the feedback it creates.

Mercury's discussion of building in public captures a useful channel truth: public work can create trust and relationships, but it can distract or attract the wrong attention. The same applies to every fashionable distribution tactic. The channel must fit the buyer and the company's proof.

Set the evidence threshold in advance

Define what a successful test must change before the sprint begins. A product test can target completion of the core task by eight of ten qualified users. A distribution test can target ten qualified conversations from one repeatable source. Use a threshold large enough to inform the next choice and small enough to reach quickly.

Record negative results with the audience, promise, channel, and product state. “Distribution failed” is too broad to teach the next experiment.

Make the decision reversible

Set a short diagnostic window, one primary bet, a budget, and a decision threshold. Avoid a permanent declaration that the company is “product-led” or “distribution-first.” Those labels can protect identity after the evidence changes.

The right answer is the next constraint. Improve the part of the system that currently prevents the company from learning whether qualified buyers can find, understand, use, and keep the product.

Stop treating every market reaction as evidence.

The Signal vs. Noise Audit applies the Signal Integrity Framework and produces a Signal Integrity Report, Noise Reduction Protocol, and Strategic Signal Dashboard. 2.5 hours. The team leaves with fewer inputs and a sharper decision.

See the Signal vs. Noise Audit →

Frequently asked

What is the simplest test for a product problem?

Bring a small number of well-qualified users to the product through direct outreach or a trusted channel. Confirm that they have the problem, understand the promise, and can use the product. If they still fail to activate, return, pay, or recommend it, the constraint likely sits in product value, onboarding, or expectation fit.

What is the simplest test for a distribution problem?

Find users who retain, pay, and describe the value clearly. Then inspect how many similar buyers encounter the product with the right context. If strong users exist but qualified reach stays scarce, the constraint likely sits in channel access, market language, authority, sales motion, or the consistency of distribution.

Can a startup work on product and distribution at the same time?

Yes, but one constraint should control the week's main bet. Keep a small product loop and a small distribution loop alive so learning continues on both sides. Put most time into the limiting constraint, define the signal that would change the diagnosis, and review it on a fixed cadence rather than switching from anxiety.

Does a great product distribute itself?

Rarely. Some products contain strong sharing, collaboration, marketplace, or word-of-mouth loops, but those are distribution mechanics built into the product. Buyers still need a reason to notice, trust, try, and repeat the story. Product quality makes distribution more productive. It does not remove the need for market contact.

When should a founder stop adding features?

Stop when feature work is not tied to a demonstrated adoption, retention, expansion, or customer-value constraint. A requested feature can feel safer than asking the market to choose. Require each meaningful addition to name the user behavior it should change and the evidence that will confirm the change. Otherwise, test distribution first.