PowerUp Coaching — Living Knowledge Base

Insight: Audit From the Avatar's Eyes — Persona-Bound Quality Review

6 min read

“I’ll say, you know what — you’re a knowledge entrepreneur, you’re a solo, you’re trying to compete in the world of AI. Now audit this from the point of view of this avatar. Tell me what’s useful, what’s interesting, what’s practical, what’s not, what’s motivating, what’s not.” — Lou, 2026-05-21

Session context: 2026-05-21_Mastermind — In the same exchange that produced the Universal Audit Phrase, Lou described a second audit move that works on a different axis. The Universal Audit asks “what’s wrong with the work.” The Avatar Audit asks “what’s wrong with the work for the person it’s meant for.” The two are complementary.

Core Idea

A piece of work can be internally correct (no errors, no contradictions) and still completely fail the person it was made for. The Universal Audit catches the first failure mode. The Avatar Audit catches the second. It assigns the AI a specific persona and asks it to evaluate the work as that persona would — what they would find useful, what they would skip, what would actually move them, what would feel like noise.

The mechanism is simple and consistent. You give the model a one-paragraph avatar — role, constraints, situation, pressures, what they care about. You then point at the work and ask: if you were this person reading this, what would land and what would not? The model shifts evaluation criteria. Sentences that read “fine” from a generic editor’s perspective read “irrelevant” from the avatar’s. Frameworks that sound smart in the abstract get rejected as “too conceptual to implement this week.”

The non-obvious move is to make the avatar constrained, not aspirational. Marketing-style avatars (“the ambitious knowledge entrepreneur who values growth”) let the model interpret broadly and the audit comes back generic. Useful avatars are tight: “a solo coach, two years in, charging $2K-5K engagements, just experimenting with AI, has 30 minutes per day for new tools, has been burned by hype before.” The tighter the avatar, the sharper the audit. The AI now has a test subject, not a target market.

This is the active-voice version of Insight - Ground AI in Your ICH Before Asking It to Build Anything (grounding in the Ideal Client Handbook). Grounding prevents the AI from drifting into generic content during generation. Avatar Audit catches the same drift during review — including in work that was generated correctly but reads wrong for the audience.

Why This Matters for Knowledge Entrepreneurs

The single most common failure mode in expert content is insider correctness with outsider unreadability. The work is true. The work is well-structured. The work assumes a level of context the reader does not have, uses examples that signal credibility to peers but feel inaccessible to clients, or recommends actions the avatar literally cannot take given their resource constraints. The author cannot see this, because the author is not the avatar. The AI cannot see this either, unless you make it inhabit the avatar explicitly.

Avatar Audit is the cheapest possible feedback loop for this failure mode. It costs one inference call. It produces a critique that approximates what a real test reader would say, scoped to the persona you are actually trying to reach. And it forces you to name the avatar explicitly — which is a form of clarity that compounds: once you have a written avatar that produces useful audit feedback, you can reuse it on every piece of content you produce, and you can refine it as you learn what really resonates with the real-world version of that person.

The deeper leverage: pairing the Universal Audit and the Avatar Audit creates a two-pass review that catches both correctness failures and fit failures in one session. Universal Audit first (“is this internally sound?”). Avatar Audit second (“would this person find it useful?”). The two passes look for different things and rarely surface the same issues. Together they cover the gap that pure self-review almost always misses.

Practical Application

Build a reusable avatar block. Save it in your ~/voice/ or .claude/avatars/ folder as a markdown file. Format:

## Avatar: [Name]
- Role: [specific job + experience level + business stage]
- Constraints: [time, money, technical skill, existing commitments]
- What they care about: [3-5 specific outcomes]
- What they've been burned by: [past failures shaping current skepticism]
- How they currently make decisions: [actual heuristics, not aspirational ones]
- What "useful" means to them: [the concrete bar your work has to clear]

Run the audit with this exact prompt:

Audit this from the point of view of [avatar name, loaded from the file above]. Tell me what’s useful, what’s interesting, what’s practical, what’s not, what’s motivating, what’s not. Then tell me which sentences they would skip and which they would screenshot.

The “skip vs screenshot” clause is the load-bearing part — it forces a discriminating judgment rather than a generic summary.

Coaching question: “If the person I claim I am writing for actually read this — line by line — which lines would they stop at and which would they scroll past?”

Evolution Across Sessions

Builds on Insight - Ground AI in Your ICH Before Asking It to Build Anything and Insight - Map the Symptom Layer to Attract Before You Solve, both of which established avatar grounding as a generation-time discipline. The new development is using the same avatar as a review-time discriminator, paired explicitly with the Universal Audit Phrase to cover both correctness and audience-fit in a single workflow.

Source

  • 2026-05-21_Mastermind (Lou — described the avatar-audit move as the second half of the audit pair, immediately after the Universal Audit Phrase)

Graph View

Backlinks