# Phantom Story — Perception Baseline

**T0 · Revised issue · Captured August 18, 2026**

| | |
|---|---|
| Prepared for | Mathew and the Phantom Story team |
| Subject | Phantom Story · phantomstory.com |
| Category | Content marketing automation · B2B SaaS |
| Engagement | First measurement. This is the reference point for everything that follows. |
| Site State snapshot | August 18, 2026 (re-frozen; the composite is unchanged from the August 16 capture) |
| Measurement surface | Homepage and robots.txt |
| Issued | August 25, 2026 |
| Revision note | This issue extends the original baseline with six instruments that had not yet run: off-site authority, competitor authority, demand sizing, Google AI Overview treatment, an AI-mentions corpus read, and a question-level answer audit across four engines. Each is labeled with its method and pull date. |

---

## 1. The perception picture

Perception is the asset. A business does not get chosen because of what it is; it gets chosen because of how it is read. Three systems do the reading, and they are co-equal: search engines, AI systems, and human buyers. Evident measures all three against the same site state and rolls them into one standalone score.

**Evident Perception Index: 50**

| Lens | Score | What it reads |
|---|---|---|
| Search | 53 | How search infrastructure indexes, classifies and trusts the site |
| AI | 49 | How machine readers parse, extract and represent the business |
| Human | 48 | How a buyer experiences the site and decides whether to trust it |

Measured coverage: 66.1% of applicable factor weight, 363 usable factor units of 548 applicable, every unit measured fresh on August 18. Nothing in this scorecard is carried forward.

### Trajectory

This remains the first measurement. The baseline was first computed from an August 16 site state and re-frozen on August 18 with the same composite: 50 then, 50 now. The site did not change between the two captures, and the factor definitions behind both are identical, so future movement can be attributed to the site rather than to the instrument.

The three lens scores sit within five points of each other. The structure worth acting on is one level down.

### Where AI answers actually name Phantom Story

Ten questions a buyer might ask, put to four answer engines on August 18 (40 answers in total). Two of the ten questions never name the company; those are the ones that matter most, because they simulate a buyer who has not heard of Phantom Story.

| Cohort | Named in the answer | Cited as a source |
|---|---|---|
| All 40 answers | 80% | 60% |
| Questions that name the brand (8 × 4 engines) | 88% | 62% |
| Discovery questions, brand not named (2 × 4 engines) | 50% | 50% |

| Engine | Named | Cited | Discovery questions |
|---|---|---|---|
| ChatGPT | 9 of 10 | 9 of 10 | named in 2 of 2, once at rank 1 |
| Claude | 9 of 10 | 9 of 10 | named in 2 of 2 |
| Gemini | 7 of 10 | 6 of 10 | named in 0 of 2 |
| Perplexity | 7 of 10 | 0 of 10 | named in 0 of 2 |

Where the company is named in a ranked answer, it places well: seven of eleven ranked mentions sit at position 1, average position 2.7.

Read the structure precisely. Presence is no longer confined to one engine, which the original capture suggested; it is now two-tier. ChatGPT and Claude name and cite Phantom Story almost everywhere, including on discovery questions. Gemini and Perplexity name it only when the question already contains the name, and on discovery questions both return nothing. Perplexity never cites phantomstory.com as a source in any of its ten answers, which means its mentions rest entirely on third-party material.

Two independent instruments corroborate the discovery-side gap, both pulled August 25. Google's AI Overview on the category's buying queries names four competitors with prices and does not name Phantom Story. A corpus of recorded AI-surface mentions returns zero entries for the domain and the brand name. Presence in live chat answers is real; presence on the recorded, citable AI surfaces has not started.

**Method note.** Named means the engine's answer text names the company. Cited means phantomstory.com appears among the sources the answer itself lists. The audit panel is ten questions; the original capture used a different tracked set, so cohort rates are the durable comparison basis from this issue forward.

---

## 2. Context

Phantom Story sells autonomous blog deployment on custom domains, positioned to capture attention inside AI assistants. The site states the proposition plainly: standalone blogs spun up in under five minutes, at $500 per month, framed as the way to win Agentic Engine Optimization. The buyer is a B2B SaaS team that wants presence inside AI answers without operating its own content infrastructure.

That proposition is what makes this baseline unusual. Phantom Story sells machine visibility. This report measures whether Phantom Story's own front door is machine visible, and now also measures the field it competes in.

**Persona panel: not available.** No generated persona set exists yet, so Human-lens results are site-level reads, not persona-segmented ones.

---

## 3. Three risks

Presented in Search, AI, Human order. Each is checked against factor-level evidence, confirmed on the live site, and now corroborated by an off-site instrument where one exists.

### Search: no instructions for crawlers, and almost no authority behind the domain

Schema and rich-result readiness scores 17. E-E-A-T and authority signals score 19. Verified live on August 24: `robots.txt` still returns 404, `sitemap.xml` still returns 404, and the homepage still carries no canonical link element.

The off-site read makes the authority half concrete for the first time. Instrument: independent backlink index, pulled August 25. The domain has 36 referring domains, 24 of them nofollow, against a competitive field whose weakest member has 2,526. Link growth was near zero from February through June, then jumped in July and August (213 and 334 new backlinks), but the new links are overwhelmingly directory-style listings: 338 of the 490 referring links come from `.tools` directory domains. Volume is arriving; authority is not.

### AI: nothing structured for machines, and no presence on the recorded AI surfaces

Structured data for AI scores 4. Fourteen of its seventeen factors score exactly zero, and they share one root cause: the homepage serves zero `application/ld+json` blocks, so every schema-dependent factor beneath that reads absent. Agent interface and protocol legibility scores 0. The remedy is one block: organisation identity, the offering, authorship, and a question-and-answer section. Five premises-style fields (opening hours, geographic coordinates, name-address-phone, payment coverage, service area) do not apply to a software subscription and should not be implemented even though the measurement counts them.

The new instruments show what this costs. Google's AI Overview answers the category's own buying question, "best AEO tools", by naming HubSpot AEO at $50 per month, Profound at $99, Semrush's toolkit at $99, and Geoptie at $49, with citations to HubSpot, Geoptie, Reddit, G2 and YouTube. Phantom Story appears nowhere in the Overview, its citations, or the first ten organic results. The AI-mentions corpus (Google AI Overview plus ChatGPT mention database) holds zero recorded mentions of the domain or the brand.

### Human: a strong story with nothing to verify it against

Social proof depth scores 20. Trust architecture scores 21. The site displays seventeen customer logos, including recognisable ones, and the live check confirms there are still no named testimonials, no case studies, no team or about substance, no privacy policy, and no security indicators behind them.

A logo wall asserts that customers exist. It does not let a skeptical buyer confirm it, and it gives a machine nothing to extract. The same asset that scores 88 on emotional intelligence and 84 on distinctiveness scores 20 on proof.

---

## 4. Strategy

### 4.1 The gap worth acting on first

| What human readers find | | What machine readers find | |
|---|---|---|---|
| Emotional Intelligence | 88 | Agent Interface & Protocol Legibility | 0 |
| Distinctiveness | 84 | Structured Data for AI | 4 |
| Narrative Arc | 72 | Schema & Rich Results | 17 |
| Clarity & Comprehension | 71 | E-E-A-T & Authority | 19 |

Both sides are fully measured on the same August 18 site state, emotional intelligence at ten of ten high confidence, structured data at seventeen of seventeen measured. A person who lands on phantomstory.com gets a clear, distinctive, well-argued story. A machine that lands on the same page gets an unlabelled document, and the wider machine ecosystem (answer engines' recorded surfaces, Google's Overview, the link graph) barely knows the company exists.

This gap is entirely fixable, it is cheap, and the company already owns the capability to fix it at scale. That was true at the first capture and remains true; nothing on the list below had shipped as of August 24.

### 4.2 Positioning and white space

The competitive field is no longer hypothetical. The ten companies that answer engines actually place next to Phantom Story, confirmed and now tracked, are Profound, HubSpot, Clearscope, Jasper, AirOps, Conductor, Semrush, Animalz, Peec AI and Siege Media. Read what they sell and a pattern appears: almost all of them measure, analyse or advise on AI visibility. Profound, Peec and the suite toolkits track how AI talks about a brand; Animalz and Siege Media are agencies that write for it.

Phantom Story is the only company in this field whose product is execution: it deploys the publications rather than measuring them. That is the white space. The priced shelf Google's Overview presents is $49 to $99 per month for tracking; Phantom Story's $500 per month buys the thing the trackers only report on. None of the current messaging makes that distinction, and no machine-readable surface exists to carry it.

The distinctiveness score of 84 says the narrative asset to express this already exists. What is missing is corroboration and category placement, not craft.

### 4.3 Messaging

Three moves, in order of leverage:

1. Convert the logo wall into evidence. One named customer, one number, one before-and-after. The seventeen logos become credible the moment any one of them carries a verifiable outcome.
2. Attribute the content. Authorship, publication dates and experience signals all measure absent. For a company whose product is publishing, an unattributed site is a contradiction the buyer notices.
3. Say what happens next. Next-step clarity and post-action clarity both measure weak. The "Get Started" path does not tell a buyer what they are agreeing to.

### 4.4 Content, sized against real demand

Instrument: independent keyword-demand panel, United States, pulled August 25. The category has a measurable shelf, and its difficulty profile is unusually favourable at the entry points:

| Query | Searches/month | Difficulty | CPC | Note |
|---|---|---|---|---|
| generative engine optimization | 4,400 | 54 | $34.79 | The category head term |
| answer engine optimization | 2,400 | 36 | $30.77 | +26% year over year |
| ai visibility tools | 1,600 | 16 | $49.25 | Low difficulty, high commercial intent |
| llm seo | 880 | 8 | $28.58 | Lowest difficulty on the shelf |
| aeo tools | 720 | 18 | $36.85 | Buying-intent query |
| geo tools | 720 | 21 | $14.80 | +177% year over year |
| automated blog content | 10 | — | $8.33 | The mechanism vocabulary; no demand |

The last row matters as much as the first. Buyers search the category language (GEO, AEO, AI visibility), not the mechanism language (automated blogs). Content and schema should speak the demand vocabulary, with the mechanism as the differentiator inside it, not the label on the door.

The two highest-frequency on-page content gaps remain a question-and-answer block (FAQ coverage, content and schema all absent across two lenses) and an about-and-team surface.

### 4.5 Campaign concepts

Three, each grounded in a measured gap:

**Fix the front door in public.** Phantom Story sells machine legibility. Publishing the before-and-after of making its own site machine legible, with these measured numbers, is both a credible proof asset and a demonstration of the product thesis. This baseline is the "before".

**Publish outcome, not volume.** The site's own dashboard copy says: "Every phantom blog reports back to one dashboard. Watch impressions, clicks, and share of voice climb across the assistants your buyers actually ask." That is outcome data. Turning it into a small number of named, dated customer outcomes addresses the weakest measured area in the baseline and differentiates against a field that competes on measurement.

**Get onto the shelf that answers the category.** Every surface that Google's AI Overview cites on the category's buying queries is a list someone wrote: HubSpot's tool guide, Geoptie's comparison, a Reddit thread, the G2 category, a Yotpo roundup. Competitors author these gatekeeper pages themselves (Profound's own "best AEO platforms" post ranks sixth; AirOps and Scrunch publish theirs). Phantom Story appears on none of them, and the mentions corpus records zero third-party AI-surface presence. A deliberate earned-placement program — the G2 category, two or three of the cited listicles, the Reddit conversation, and a published comparison of its own — is the shortest path from "present in chat answers" to "present on the surfaces those answers cite".

### 4.6 Standing watch

**Not available this cycle.** No monitoring subscription and no signal clusters exist yet, so there is no competitive-move feed or answer-panel trend to report. Activating monitoring remains an action, and it is the precondition for the change measurement in section 9.

### 4.7 Site architecture

**Not available this cycle.** Measurement covered the homepage and the robots.txt path. No page inventory or competitor page intersection can be derived from a single measured page. Section 8 records how this bounds the baseline.

---

## 5. Action ledger

The original issue of this baseline listed eight actions. Every one was re-checked against the live site on August 24. None has shipped, so the ledger's first two books are short and the additions from the new instruments make eleven in total.

**Shipped and verified: nothing yet.** robots.txt and sitemap.xml still return 404, the homepage still has no JSON-LD and no canonical element, and no proof, attribution, policy or contact surface has appeared.

**Carried from the first issue, still detected (1–8):**

| # | Action | Addresses | Severity |
|---|---|---|---|
| 1 | Publish one JSON-LD block: Organization identity, the offering, authorship, FAQPage. Omit premises fields (hours, geo, NAP, payment coverage, service area). | The zero-scored structured-data family plus Search schema and entity identity | Critical and high |
| 2 | Add `robots.txt` and `sitemap.xml`. Both currently 404. | Agent interface (scored 0), crawl and indexability | Critical |
| 3 | Add a canonical link element to the homepage. | Canonical correctness across both Search and AI lenses | Critical |
| 4 | Make the contact path machine-readable and unambiguous: a real contact surface plus `ContactPoint`. | Contact accessibility, path clarity, friction | Critical and high |
| 5 | Convert the logo wall into verifiable proof: at least one named customer with a dated, quantified outcome. | Social proof depth (20), E-E-A-T (19), claim substantiation | High |
| 6 | Publish an FAQ block, as content and as schema. | FAQ coverage, content and schema, two lenses | High |
| 7 | Attribute the site: authorship, dates, an about-and-team surface with named people. | Attribution, experience signals, team identification | High |
| 8 | Publish a privacy policy and visible security indicators. | Privacy policy (absent), security indicators, risk reduction | High |

**New this cycle, from the new instruments (9–11):**

| # | Action | Grounded in |
|---|---|---|
| 9 | Run an earned-placement program on the category's gatekeeper surfaces: the G2 AEO category, the cited tool listicles, and the Reddit conversation the Overview cites. Publish an own-domain comparison page as part of it. | AI Overview treatment and mentions-corpus zero (§3, §4.5) |
| 10 | Build the demand-shelf content set against the low-difficulty entries (llm seo, ai visibility tools, aeo tools), speaking the category vocabulary. | Demand sizing (§4.4) |
| 11 | Standardise the public name. The site titles itself "Phantomstory" in one word while the brand is spoken as "Phantom Story" in two; answer engines treat the two forms inconsistently. Pick one canonical rendering, use it in the title, schema and profiles, and connect variants with `sameAs`. | Answer-audit name-form inconsistency; entity clarity (56) |

Actions 1 through 4 remain one working session and address all seven critical findings. Frontier factors from the first issue (the partial structured-data reads at 25) should move furthest once action 1 lands.

---

## 6. Lens deep dives

### 6.1 Search lens · 53

| Category | Score | Factors |
|---|---|---|
| On-Page SEO Fundamentals | 76 | 5 |
| Competitive Intelligence | 74 | 3 ⚠ |
| Technical SEO Advanced | 66 | 8 |
| Indexability & Crawlability | 65 | 9 |
| Core Web Vitals & Performance | 64 | 3 ⚠ |
| Entity Recognition & Knowledge Graph | 59 | 9 |
| Site Authority & Trust Signals | 56 | 11 |
| Content Quality Signals | 55 | 21 |
| User Engagement & Click Quality | 50 | 3 ⚠ |
| SERP Optimization | 48 | 9 |
| Search Policy & Content Safety | 33 | 12 |
| E-E-A-T & Authority Signals | 19 | 20 |
| Schema & Rich Results | 17 | 13 |

The mid-range is technically sound and adequately written. The two heaviest categories in the lens are also its two weakest, and the off-site instrument now grounds the authority half of that verdict.

**The authority ledger.** Instrument: independent backlink index (rank, referring domains, spam score), single source, pulled August 25. Rank runs 0 to 1,000, higher is stronger.

| Company | Authority rank | Referring domains | Spam score |
|---|---|---|---|
| HubSpot | 681 | 394,622 | 4 |
| Semrush | 611 | 139,575 | 9 |
| Jasper | 530 | 34,974 | 7 |
| Peec AI | 449 | 2,558 | 21 |
| Conductor | 440 | 15,385 | 16 |
| Profound | 378 | 3,914 | 5 |
| Clearscope | 371 | 6,874 | 6 |
| Siege Media | 350 | 6,312 | 10 |
| AirOps | 339 | 3,741 | 11 |
| Animalz | 323 | 2,526 | 6 |
| **Phantom Story** | **204** | **36** | **2** |

Phantom Story holds the lowest authority rank and 1.4% of the referring domains of the next-lowest company. The profile is clean (spam score 2, the best in the field) and young rather than damaged: link velocity was flat until July, and the July–August wave (+213, +334 backlinks) is directory listings rather than editorial coverage. The E-E-A-T score of 19 is not a template artifact; the off-site graph agrees with it.

**Google AI Overview treatment.** Instrument: live SERP reads on two category queries, United States, August 25. The Overview is present on both, holds the top position on both, names competitors with prices on the buying query, and does not name Phantom Story on either. Its cited sources (HubSpot, Geoptie, Reddit, G2, Yotpo, YouTube) define the earned-media queue in section 4.5.

**Rankings context.** Every organic result at depth ten on the buying query is a listicle, review directory or vendor guide; no standalone vendor homepage ranks. Entry to this shelf is through gatekeeper content, which is why action 9 is a placement program rather than a homepage optimisation.

### 6.2 AI lens · 49

| Category | Score | Factors |
|---|---|---|
| Technology Context | 77 | 10 |
| AI Brand Presence | 77 | 11 |
| Citation Packaging | 68 | 9 |
| Content Extractability | 66 | 23 |
| Multimedia AI Accessibility | 60 | 9 |
| AI Access & Permissions | 59 | 10 |
| Entity Clarity | 56 | 10 |
| Credibility Signals | 34 | 12 |
| Temporal Signals | 32 | 7 |
| Compliance & Regulatory | 13 | 7 |
| Structured Data for AI | 4 | 17 |
| Agent Interface & Protocol Legibility | 0 | 2 ⚠ |

Content extractability at 66 next to structured data at 4 repeats the baseline's central pattern: the prose is machine-parseable, the labelling that tells a machine what the prose means is absent.

**Answer-engine presence** is measured fresh in this snapshot and reported cohort-split in section 1. The category score of 77 reflects strong branded recall with real discovery presence on two engines; the two-tier structure and the recorded-surface absence are the working reality behind the number.

Compliance and regulatory at 13 is reported at category level only; its factors score low without recording individual deficiencies, so no specific claim is drawn from it.

### 6.3 Human lens · 48

| Category | Score | Factors |
|---|---|---|
| Emotional Intelligence | 88 | 10 |
| Distinctiveness | 84 | 11 |
| Narrative Arc | 72 | 12 |
| Clarity & Comprehension | 71 | 11 |
| Audience Journey Support | 61 | 8 |
| Visual & Emotional Hierarchy | 57 | 10 |
| First Impressions | 51 | 7 |
| Conversion Facilitation | 51 | 10 |
| Skeptic Resilience | 51 | 6 |
| HCI Fundamentals | 39 | 12 |
| Content Quality & Consistency | 25 | 1 ⚠ |
| Trust Architecture | 21 | 7 |
| Social Proof Depth | 20 | 5 |

The top of the table is the strongest result in the baseline and well measured. The bottom is where the buyer stalls: a narrative in the seventies and eighties sitting on trust and proof scores in the twenties.

⚠ Content quality and consistency rests on a single factor and is starred as thin collection, not narrated as a finding.

**Review record: not measured.** No business-profile or first-party review collection has run. There is no rating, count or velocity in this baseline.

---

## 7. Competitive landscape

**The field is now confirmed and tracked.** The ten companies above were selected by one rule: they are the names answer engines actually placed next to Phantom Story in the audited answer set, taken in order of how often the engines named them, each verified live on August 24. This replaces the provisional list in the original issue, which mixed sources and included companies the engines no longer surface.

**What is measured today:** the answer-space field itself (who gets named, section 1), and each company's off-site authority (section 6.1). **What is in progress:** full per-competitor factor scoring on the Evident library was launched for all ten on August 25 and lands in the platform as the runs complete; scored EPI comparisons follow in the next issue rather than being estimated here.

**Share of answers.** In the 40 audited answers, the most-named companies alongside Phantom Story were Profound (18 mentions), HubSpot (12), Clearscope (9) and Jasper (9); Phantom Story itself was named in 32. The measurement field is the one Phantom Story's buyers actually see, and it is dominated by companies that measure AI visibility rather than execute against it, which is the positioning opening in section 4.2.

**Entity verification.** The measured entity resolves cleanly to phantomstory.com and no name-twin condition exists. One consistency issue is real and client-actionable: the site's own one-word rendering "Phantomstory" versus the spoken two-word "Phantom Story" (action 11).

---

## 8. The full scorecard

Thirty-eight categories, all measured against the same frozen August 18 site state.

| | |
|---|---|
| Applicable factor units | 548 |
| Usable factor units | 363 |
| Weighted coverage | 66.1% |
| Newly refreshed | 363 (100% of usable) |
| Never measured | 94 |
| Expired or unavailable | 91 |
| Measured surface | 2 URLs (homepage, robots.txt) |

Two coverage facts shape this baseline. First, it is a single-page measurement: absences are absences on the homepage, and deeper content was not measured. Second, one third of applicable factor weight is unmeasured; the composite of 50 is a real score over a disclosed subset.

Sharpest measured deficits, by name: Agent Interface & Protocol Legibility (0), Structured Data for AI (4), Compliance & Regulatory (13), Schema & Rich Results (17), E-E-A-T & Authority Signals (19), Social Proof Depth (20), Trust Architecture (21), Temporal Signals (32), Search Policy & Content Safety (33), Credibility Signals (34).

Measured strengths, by name: Emotional Intelligence (88), Distinctiveness (84), AI Brand Presence (77), Technology Context (77), On-Page SEO Fundamentals (76), Narrative Arc (72), Clarity & Comprehension (71).

Thin-collection categories, starred rather than narrated: Content Quality & Consistency (1 factor), Agent Interface & Protocol Legibility (2), and the three-factor Search categories (Competitive Intelligence, Core Web Vitals, User Engagement).

---

## 9. Cadence and what T1 should measure

**Re-measure trigger.** Actions 1 through 4 change what machines can read. Once they ship, a re-measure will show movement in Structured Data for AI, Agent Interface & Protocol Legibility, Schema & Rich Results and Entity Clarity. Those four categories remain the cleanest test of whether a shipped change produces measured perception movement.

**Comparable from this issue forward.** The cohort-split answer rates (section 1), the authority ledger (section 6.1), the demand shelf (section 4.4), and the AI Overview treatment are all dated, method-labeled instruments that repeat cheaply. The competitor factor scores land in the platform as their runs complete and give T1 a scored leaderboard.

**Not collected at all.** Personas, business-profile and first-party reviews, monitoring, and journey evidence. If change on any of these matters, the instrument has to run before the change ships.

**Comparison keys.** The factor-definition revision behind this scorecard is identical to the original issue's, so instrument drift is ruled out for the next comparison. The machine-readable representation carries the snapshot identifiers a later measurement should diff against.

Suggested cadence: activate monitoring now, ship actions 1–4, re-measure, and treat that as T1.

---

## Use this report with an agent

This report exists in three representations of the same underlying measurement.

| Representation | Path | For |
|---|---|---|
| Human | `index.html` | Reading |
| Agent | `report.json` | Programmatic consumption |
| LLM context | `report.md` | Dropping into a model context |

```bash
curl -s https://<report-host>/report.json > phantomstory-t0.json
```

```bash
jq '.perception_gaps.cross_lens_divergence.pairs[0:5]' phantomstory-t0.json
```

The JSON is the measurement object: categories with per-factor scores, confidence, freshness and measurement ids; findings with severity and admission state; the EPI with full coverage accounting; derived perception gaps with their rules attached; the instrument declaration; and the provenance receipt.

---

*Evident measures how search systems, AI systems and human buyers read a business, and rolls the three into one standalone Perception Index. Baseline captured August 18, 2026 from a frozen Site State snapshot measured on phantomstory.com. Live verification of structured data, canonical, robots, sitemap and proof surfaces performed August 24, 2026. Off-site authority, demand, AI Overview and mentions-corpus instruments pulled August 25, 2026 (independent market telemetry; method and date labeled per section).*