Taste is easiest to discuss in the abstract and hardest to inspect in the system.

That is why teams should audit it.

Not as a branding exercise. Not as a creative workshop. As an operating diagnostic. Where is judgment strong? Where is it hidden? Where is it stale? Where is it over-centralized? Where does mediocre work pass because nobody can explain what better means?

A taste audit asks whether the organization can reliably tell what is worth making, what is good, what is mediocre, what should be cut, and what deserves more effort.

Start with the work.

Pick five recurring work surfaces that matter: product specs, customer communications, AI outputs, hiring packets, strategy memos, technical design docs, sales collateral, support workflows, onboarding flows, launch plans, performance reviews. Do not audit everything. Pick the places where quality has consequence.

For each surface, ask the first question: what does good look like?

If the answer is vague, taste is not yet operational. “Clear,” “strategic,” “high quality,” “customer-centric,” “enterprise-grade,” and “on brand” are not enough. Show examples. Name attributes. Identify failure modes. Explain tradeoffs.

The second question: what does unacceptable look like even if it is polished?

This is where the audit gets useful. The danger zone is not obviously bad work. It is plausible work. The customer email that sounds professional but avoids ownership. The AI summary that is organized but unsupported. The strategy memo that names priorities but not choices. The product spec that lists requirements but hides the customer anxiety. The hiring note that praises a candidate without evidence. The technical proposal that works now but creates ownership debt.

If the team cannot identify plausible failures, it will accept them.

The third question: who currently catches quality problems?

List the names. If the same few names appear everywhere, taste is centralized. That may be necessary temporarily, but it is a bottleneck and a risk. The goal is not to remove senior judgment. The goal is to transfer enough of it that the organization is not waiting for inspection.

The fourth question: when are taste judgments applied?

Early taste creates leverage. Late taste creates rework. If the bar appears only in final approval, the review system is poorly designed. Add earlier gates: brief alignment, exemplar review, risk tiering, draft critique, source checks, decision criteria.

The fifth question: are corrections becoming standards?

Look at repeated feedback from the last month. Did it become an example, rubric, checklist, owner expectation, review rule, or training artifact? Or did it disappear into chat comments and meeting memory?

Every repeated correction should become a teachable standard.

The sixth question: which standards are observable?

A standard is not what leaders prefer. It is what the organization consistently enforces. Can you see the standard in artifacts and behavior? Decision memos name tradeoffs. Customer updates say what happened and what happens next. AI outputs cite sources when used for decisions. Hiring packets include evidence against the scorecard. Launch reviews include support readiness. Technical decisions name reversibility and ownership.

If the standard cannot be observed, it cannot be managed.

The seventh question: where is taste stale?

Find examples the company still treats as excellent. Are they still excellent? Against current customer expectations? Current tool capabilities? Current competitive baselines? Current strategy? Current scale?

Stale taste often hides in sacred artifacts: the old positioning that once worked, the old process that once created speed, the old technical rule born from a past scar, the old design principle that has become visual habit. Keep the principle if it still holds. Retire the expression if reality moved.

The eighth question: where is the organization over-impressed?

Every company has surfaces where polish gets too much credit. Decks. Prototypes. Confident analysis. Charismatic interviews. AI-generated fluency. Beautiful dashboards. Strong taste asks what is underneath the polish. Evidence, consequence, ownership, fit, reversibility, customer truth.

The ninth question: where is the organization under-sensitive?

Some quality failures are quiet until late. Ambiguous ownership. Weak source discipline. Slightly vague customer language. Small technical shortcuts. Unclear decision rights. Poor handoffs. These do not always trigger immediate pain. They compound.

Taste should catch compounding cost early.

The tenth question: where is taste being used as a veto instead of a system?

A veto may be necessary. Some work should be stopped. But if the same leader keeps vetoing the same type of work, the system is failing. Convert the veto into examples, standards, earlier review, and ownership.

The audit should include at least one real artifact per surface, not only a conversation about the surface. Pull the actual launch brief, actual AI-generated summary, actual hiring packet, actual technical design, actual customer email, or actual executive memo. Taste improves when the organization looks at work that had consequences, not hypothetical standards everyone can agree with.

The output of a taste audit should be practical. For each work surface, produce:

  • the current quality bar;
  • three strong examples;
  • three plausible failures;
  • the owner of the standard;
  • the review point where taste should appear;
  • the most common drift pattern;
  • one change to make the bar more legible.

This can fit on a page per domain.

For AI workflows, add a specific acceptance audit:

  • What is the workflow used for?
  • What is the risk tier?
  • What sources does it depend on?
  • What outputs require human review?
  • What examples define good, flawed, and unacceptable?
  • What failure modes are tracked?
  • Who owns updates to the rubric?
  • What happens when reviewers disagree?

This prevents the organization from confusing generation with quality.

The taste audit should end with one uncomfortable conversation: which work are we currently accepting that we would reject if our standards were explicit?

That question reveals the real bar.

Taste is not supposed to remain a private gift. It should become a shared operating capability. The audit is how you find where that has not happened yet.

Taste sees, standards protect, craft executes. If the company cannot see clearly, standards will protect the wrong things and craft will polish the wrong work.

Audit the sight first.