# Q03 ChatGPT — "What vitamin C serum should I buy?"

Captured 2026-09-02, logged out, 5 fresh sessions, full-page.

## Runs

| Run | Top pick | Card | Other brands | Citations | Ad |
|---|---|---|---|---|---|
| 1 | **CeraVe** Skin Renewing Vit C | $28.99·Ulta | Vanicream; e.l.f. Bright Icon; SkinCeuticals C E Ferulic | **ELLE**; **Good Housekeeping**; **Glamour**; verywellhealth | Protocol Lab (on-category) |
| 2 | **Timeless 20%** | $27.99·Target | LRP Pure 12% ($46.99·Ulta); CeraVe ($29.99·CVS); e.l.f. ($16·Ulta) | **Healthline**; **Prevention**; Cleveland Clinic | Protocol Lab |
| 3 | **Timeless 20%** | **$35.95·Timeless Skin Care** | CeraVe; e.l.f.; SkinCeuticals (~$185) | Cleveland Clinic; **Good Housekeeping** | Protocol Lab |
| 4 | **CeraVe** Skin Renewing Vit C | $29.99·CVS | e.l.f. ($16·Ulta); Cetaphil ($25.49·CVS); Vichy ($34.99·CVS); Sunday Riley ($85·Nordstrom) | WebMD; Cleveland Clinic | Protocol Lab |
| 5 | **Timeless 20%** | $27.99·Target | LRP Pure 12%; e.l.f.; SkinCeuticals; BeautyStat ($62·Ulta); CeraVe | **Healthline**; **Allure**; **Good Housekeeping**; Cleveland Clinic | Protocol Lab |

## 1. First top-pick churn in the study

q01: CeraVe 5/5. q02: EltaMD 5/5. **q03: Timeless 3/5, CeraVe 2/5.**

The recommendation is not stable on this query. Two different products are "the one to buy"
depending on which run you happen to get. This is why 5 runs per query was the right call.

## 2. THE 100% BREAKS — three recommended brands are absent from Google's top 40

| ChatGPT brand | Runs | Google organic position |
|---|---|---|
| CeraVe | 5/5 | #2 ✅ |
| La Roche-Posay | 2/5 | #4 ✅ |
| Sunday Riley | 1/5 | #6 ✅ |
| Timeless | 3/5 | #7 ✅ |
| Vichy | 1/5 | #13 |
| Vanicream | 1/5 | #14 |
| Cetaphil | 1/5 | #31 |
| **e.l.f.** | **5/5** | **ABSENT from top 40** ❌ |
| **SkinCeuticals** | 3/5 | **ABSENT from top 40** ❌ |
| **BeautyStat** | 1/5 | **ABSENT from top 40** ❌ |

**Top 40: 7/10 = 70%. Top 10: 4/10 = 40%.**

**e.l.f. Bright Icon appears in 5 of 5 runs and is nowhere in Google's organic top 40.** It is
one of the most consistently recommended products in the entire study, and Google's shopping
shelf does not rank it.

Peec's "100% of products explained by the top 40 organic" **does not hold universally.** It
held on q01 and q02 and fails here. That is a real, publishable limit on a widely repeated claim
— and it is exactly why the replication was worth doing rather than citing.

## 3. What explains the divergence: the editorial layer overrides the commerce layer

q03's citations are almost entirely **consumer magazines**: ELLE, Good Housekeeping, Glamour,
Allure, Healthline, Prevention, WebMD.

SkinCeuticals C E Ferulic is the canonical "editor favorite" vitamin C serum in beauty press —
ChatGPT says so explicitly ("it remains a dermatologist/editor favorite", citing Good
Housekeeping) — and at ~$185 it does not rank on a shopping shelf sorted by commercial signals.
e.l.f. Bright Icon is a heavily-covered budget launch. Neither reaches Google's top 40; both
reach the recommendation.

### The pattern across three queries, and it is testable

| Query | Citation character | Brands named | Top-40 overlap | Behavior |
|---|---|---|---|---|
| q01 moisturizer | **Medical** (Cleveland Clinic, AAD, Nebraska Medicine, brand sites) | 3 | **100%** | Mirrors Google's top 4 |
| q02 sunscreen | **Prevention-led** | 6 | **100%** | Google's set, magazine's ranking |
| q03 vitamin C | **Magazine-saturated** (ELLE, GH, Glamour, Allure) | 10 | **70%** | Magazine brands Google doesn't rank |

**Hypothesis: the more the answer leans on magazine sources, the further it drifts from the
commerce shelf.** Three data points, consistent direction. Seven queries left to test it. If it
holds, that is the finding of the whole study, and it is a mechanism nobody has published.

## 4. Ads: 5 of 5, and always the same advertiser

**Protocol Lab, "#1 Vitamin C Serum — Award-winning, dermatologist-approved", in every single
run.** On-category here.

Note the cross-contamination: **Protocol Lab also appeared on q01, a *moisturizer* query**,
where a vitamin C serum ad was off-category. Same advertiser, bleeding across queries.

Running ad total: **10 of 15 runs (67%)**.

## 5. Price instability: the brand-direct link costs 28% more

**Timeless 20% Vitamin C + E Ferulic:**
- **$27.99 · Target** (runs 2, 5)
- **$35.95 · Timeless Skin Care** (run 3)

Same product, same query, same day. In run 3 the recommendation sends the buyer to the
manufacturer at **28% above the Target price shown in the other runs.**
