# Sample Audit — "Company B": Specialty Coffee Gear Brand (DTC)

**Hundred Euro Labs · AI Search Visibility SNAPSHOT · 2026-08-25**
Tier: **Snapshot** (Method v1.0 §5.1). Anonymized public sample — identity withheld; measurements real and reproducible.

> **How this was measured (honesty box).** Citation layer: live web-search run executed 2026-08-25 on one battery query representative of C1/C2 (category + comparison intent), top results logged. Assistant layer: unaided recall check across paraphrases by the audit agent (a frontier LLM), single pass per paraphrase, training-data bound. No multi-engine UI/API runs claimed; full protocol is the paid tier.

---

## 1. Subject profile (anonymized)

Direct-to-consumer **specialty coffee equipment brand**: manual grinders and electric grinders in the enthusiast/budget-to-mid segment ($50–500 street prices), strong Amazon presence, cult following in coffee communities. Competes for "best grinder" attention against two dominant US brands (one famous for design-forward electrics, one the default entry-electric recommendation) plus a premium hand-grinder rival.

## 2. Battery slice run

| Query (as run) | Class | What it tests |
|---|---|---|
| "best budget espresso grinder [Brand] vs [US rival] 2026" | C1/C2 | Where does the brand appear when buyers compare value picks? |

Assistant-recall checks: generic ("what grinder should I buy?"), budget-manual qualified, espresso-under-$200 qualified.

## 3. Findings

### 3.1 Presence — owns its niche slice, absent from the head

**Citation layer (10 distinct surfaces sampled): brand products appear in 8 of them (80%)** — an unusually strong showing:

| Surface type | Inclusion | Framing |
|---|---|---|
| National-newspaper product guide | ✅ | "**Best for manual use**" — flagship pick |
| Long-running enthusiast publication's budget guide | ✅ | Featured pick: "genuine mechanical upgrade" |
| Specialty-advice buyer's guide | ✅ | "**Best value (manual)**" pick, praised grind quality vs 2× price |
| Coffee-review site deep review | ✅ | Positive; niche caveats noted |
| Independent reviewer teardown | ✅ | Enthusiastic recommendation |
| Retailer editorial roundup | ✅ | Multiple models listed across price tiers |
| Budget-grinders roundup | ✅ | "Budget champion" framing |
| Design-rival's review page | ⚠️ | Mentioned only as pricier alternative to the rival |
| Barista-edu comparison | ❌ | Absent (electric-first comparison) |
| Multi-method buyer guide | ❌ | Absent |

**The pattern:** the brand *dominates* the **budget/manual/travel** sub-queries and is nearly absent from **unqualified "best coffee grinder" and electric-espresso head terms**, where two US rivals take every default slot. Assistant recalls mirror this exactly: unaided "what grinder should I buy?" answers lead with the rivals; the brand appears when the question contains a budget or manual qualifier.

### 3.2 Narrative — sentiment is a genuine asset

Where present, third-party characterizations are strongly positive — rare enough to note explicitly: *"punches absurdly above its weight"*, *"rivals electrics twice the price"*, *"budget champion"*, *"brought the receipts with the C5"*. Estimated Sentiment Index from sampled clauses: **≈ +0.6**.

Two recurring structural criticisms create avoidable risk:

1. **Lineup confusion:** a long-running enthusiast publication notes (paraphrased; wording lightly altered to preserve anonymity) that the brand releases confusingly similar models at a rapid pace, so parsing the lineup requires patience. Model naming (numbered entry/flagship lines × espresso-focused variants with near-identical names) is genuinely hard to map to buyer needs — even careful reviewers misstate which model has which adjustment system.
2. **Design trade-offs stated bluntly:** internal (vs external) adjustment dial on the current line; "travel tool first, daily driver second."

### 3.3 Accuracy — mostly clean, one live hazard

Specs (burr sizes, click counts, micron steps) were consistent across surfaces checked; **AEC ≈ 0–1**. The hazard is the lineup confusion itself: assistants asked to distinguish the brand's own models are prone to blending generations — a self-inflicted accuracy risk that grows with each new SKU.

### 3.4 Snapshot scorecard (directional)

| Metric | Value | Read |
|---|---|---|
| Presence Rate (niche-qualified slice) | ~80% | Excellent within its lane |
| Presence in unqualified head terms | ~20–30% | The growth ceiling |
| Share of Voice vs 2 US rivals | #3 in head terms | Rivals own defaults |
| Sentiment Index | ≈ +0.6 | Best-in-sample narrative we have audited |
| Accuracy Errors | ~0–1 (lineup-blending risk) | Fixable with clarity work |

## 4. Prioritized fix plan (evidence-linked)

| Prio | Action | Workstream | Evidence |
|---|---|---|---|
| P0 | Publish one canonical **model-lineup hub** ("which grinder for which brewer") with per-model structured data (specs, burr, adjustment, ideal use) — make the catalog describable without guesswork | Structured clarity | §3.3 lineup hazard, §3.2 criticism 1 |
| P0 | `llms.txt` + spec-sheet consistency sweep across retail channels so assistant answers stop blending generations | Structured clarity | §3.3 |
| P1 | Placement program into the electric-inclusive "best espresso grinder under $200" roundups where absent — the entry-level electric slot is the single largest reachable audience gap | Cited-source placement | §3.1 ❌ rows |
| P1 | Turn the national-guide "Best for manual use" status into owned content and retailer co-marketing; quote it wherever buyers compare | Content placement | §3.1 ✅ row 1 |
| P1 | Brewing-guide content targeting problem/solution queries (dialing in, travel setups) where community trust already exists | Content placement | Method §7.3 |
| P2 | Wikidata entity enrichment; assess eligibility of founder/company story for knowledge-base presence | Knowledge-base entities | Method §7.5 |

**What success looks like in 14 days (verification re-run):** lineup hub live and indexed; zero model-confusion errors in re-run spot checks; ≥1 outreach thread opened with an absent roundup. Not promised: head-term placement changes — those require sustained programs and are never guaranteed.

---

*Snapshot limitations apply (Method §10): one citation-layer query, single-date runs, one assistant surface. Company identity withheld; source list available on request.*