Luxury · Benchmarking brand health across digital and editorial signals
During Milan Fashion Week, Gucci, Prada, and Versace each posted runway shows within 48 hours of each other. Within days, brand trackingbrand trackingRegular measurement of brand health metrics (awareness, image, preference, and purchase intent) over time, so shifts can be detected and linked to marketing activity.View full definition → dashboards showed Gucci "winning" share of voiceshare of voiceYour brand's share of total advertising or conversation volume in your category, measured against competitors over a defined period.View full definition → by a wide margin. Sounds decisive. Except the spike was driven almost entirely by a single controversial front-row seating arrangement that went viral on X, not by product desirability. Versace's smaller but denser spike came from fashion trade press (WWD, Business of Fashion) praising tailoring construction, a signal that correlates far better with next-season sell-through.
This is the core problem in luxury brand measurement: volume of attention and quality of attention are not the same thing, and most dashboards conflate them. This lesson is about telling them apart using data.
Why brand health data is hard in luxury
Luxury brand valuebrand valueThe commercial value your brand adds beyond functional product attributes: the price premium, preference and loyalty it generates.View full definition → lives in perception (desirability, exclusivity, craftsmanship reputation) more than in transaction data alone. Unlike mass retail, where clickstream and basket data tell most of the story, luxury houses must triangulate signals across editorial media, social platforms, search behavior, and resale markets to know if a brand is actually gaining desirability or just generating noise.
Three data families matter most:
1. Media and earned coverage data
- Share of voice (SOV): the percentage of total category media mentions captured by one brand, across a defined time window and source set.
- Earned media value (EMV): a modeled dollar-equivalent value assigned to organic mentions, based on estimated reachreachThe number of unique people exposed to your message in a given period. Unlike impressions, reach counts each person once, no matter how often they see it.View full definition → and engagement, as if the exposure had been paid mediapaid mediaVisitors arriving via paid ads or sponsored placements, where you pay a platform to display your message rather than earning visits organically.View full definition →. Vendors: Launchmetrics, Meltwater, Cision.
2. Search and interest data
- Search interest indices (e.g., Google Trends, a free tool indexing relative search volume on a 0 to 100 scale rather than absolute query counts).
- Branded vs. non-branded search share: whether people search "Prada re-nylon bag" (branded, high intent) vs. "waterproof designer tote" (category, lower intent).
3. Social and community data
- Engagement rateEngagement rateThe ratio of interactions (likes, comments, shares) to reach for a given piece of content, used to gauge how well audiences respond relative to how many people saw it.View full definition → (likes, comments, shares divided by followers or impressionsimpressionsThe total number of times an ad or piece of content is displayed, regardless of clicks. Each display counts as one impression, even to the same person.View full definition →).
- Sentiment classification (positive, neutral, negative mentions), typically produced by natural language processing (NLP) models applied to text.
- Influencer and creator tiering: mega (1M+ followers), macro (100k to 1M), micro (10k to 100k), each with different credibility-to-reach tradeoffs.
The metric everyone misuses: Earned Media ValueEarned Media ValueUnpaid media exposure such as press coverage, word-of-mouth, social shares and customer reviews generated organically rather than bought or self-published.View full definition →
EMV is popular because it produces a single dollar figure boards understand. It is also the metric most prone to inflation.
EMV is typically calculated as:
EMV = (estimated impressions) × (engagement rate benchmark) × (media value multiplier)The problem: the "media value multiplier" is a vendor-specific assumption, not an observed market price. Two vendors analyzing the identical Instagram post can produce EMV estimates that differ by 2 to 3x. There is no regulatory standard (no equivalent of GAAP, Generally Accepted Accounting Principles, for earned media). This is why EMV should be used only as a directional, within-vendor, quarter-over-quarter comparison, never as an absolute cross-brand truth, and never mixed across vendors in the same report.
Worked example (illustrative, not sourced from any single real campaign):
Suppose a runway show generates:
- 4,000 media placements
- Estimated total impressions: 180 million
- Applied benchmark engagement rate: 0.35%
- Vendor multiplier: $0.18 per engaged impression
Calculation:
180,000,000 × 0.0035 = 630,000 engaged impressions
630,000 × $0.18 = $113,400 estimated EMV for that placement set.
Now compare that to actual paid media spend for the same period. If paid spend was $2 million and EMV is $113,400, EMV is not "free marketing worth millions," it is a modest supplementary signal. Boards sometimes get this backwards.
Data quality and governance: what to check before trusting a dashboard
Before presenting any brand health number externally or to leadership, run this checklist:
- Source transparency: Does the vendor disclose which outlets, platforms, and languages are included? Trade press (WWD) versus tabloid mentions should not be weighted identically.
- Deduplication: Is the same article or repost counted once or many times across syndication?
- Bot and fake engagement filtering: Instagram and TikTok engagement pools are known to include inauthentic accounts. Reputable vendors disclose bot-filtering methodology; if they do not, treat the number as unverified.
- Time-window consistency: Comparing a 7-day post-show spike for one brand to a 30-day window for a competitor invalidates the comparison.
- Currency and geography normalization: EMV and search data must specify market (US, EU, China) since platform usage and press ecosystems differ sharply (Google Trends is near-irrelevant in China, where Baidu and Weibo dominate).
- Sentiment model auditing: NLP sentiment models trained on general English text often misclassify fashion irony, sarcasm, or niche slang. Spot-check a sample manually each quarter.
A useful public reference for media measurement standards is the AMEC Integrated Evaluation Framework, a free, non-vendor-captured methodology for auditing communications measurement.
Benchmarks: what "good" looks like
As-of-2025 industry estimates (directional, not precise, vendor-sourced figures vary):
- Top-tier luxury houses (Chanel, Dior, Louis Vuitton) typically capture an estimated 15 to 25% category share of voice during major fashion weeks, per Launchmetrics' published Fashion Week rankings.
- Engagement rate benchmarks on Instagram for luxury fashion brands run an estimated 0.5 to 1.5% of followers per post (industry benchmark reports, e.g., Rival IQ), notably lower than smaller niche or streetwear brands, which can exceed 3 to 5%, because larger follower bases dilute engagement rate mathematically.
- Search interest spikes around runway shows typically decay within 2 to 4 weeks back to baseline unless reinforced by product drops or celebrity placement, a pattern visible directly in Google Trends data.
The right question isn't "who had the biggest spike" but "whose search and sentiment held up after the news cycle moved on." Sustained branded search 30 days post-show is a far better desirability proxy than day-one SOV.
Knowledge check
1. In the Milan Fashion Week example, Gucci had a much larger share of voice spike than Versace, but analysts questioned whether this reflected true brand health. What was the core issue?
2. Why is brand health measurement particularly challenging for luxury houses compared to mass retail brands?
3. A brand's Google Trends score for its name rises from 40 to 80 over a quarter. What is the most accurate interpretation of this change?
4. Select ALL correct answers about why 'volume of attention' and 'quality of attention' should be treated as distinct in luxury brand tracking.
Select all the correct answers.
5. Select ALL correct answers about branded vs. non-branded search share as a brand health signal.
Select all the correct answers.
Building a simple cross-brand scorecard
A defensible brand health scorecard for three competing houses should weight sources by evidentiary strength, not just volume:
| Signal | Weight rationale |
|---|---|
| Trade press sentiment (BoF, WWD) | High: editorial credibility, professional judgment |
| Branded search sustain (30-day) | High: reveals real consumer intent, hard to fake |
| Consumer social engagement rate | Medium: useful but bot-vulnerable |
| Raw SOV / mention count | Low: easily inflated by controversy or bot networks |
| EMV (single vendor, consistent) | Low-medium: directional trend only |
This mirrors a broader data discipline: weight signals by resistance to manipulation and by correlation with the outcome you actually care about (in this case, desirability and eventual purchase intent), not by ease of collection.
Key Takeaways
- Volume metrics (raw mentions, SOV, EMV) are easy to inflate through controversy or bot activity; treat them as noise-prone unless paired with a quality signal.
- EMV has no standardized methodology across vendors: use it only for within-vendor, time-series comparison, never as an absolute cross-brand dollar figure.
- Sustained branded search interest (2 to 4 weeks post-event) is a stronger desirability proxy than day-one attention spikes, and it's freely observable via Google Trends.
- Always audit data quality before trusting a dashboard: source transparency, deduplication, bot filtering, consistent time windows, and geography normalization.
- Build scorecards that weight signals by resistance to manipulation and correlation with real outcomes (editorial credibility, sustained search), not by raw ease of collection.
Frequently asked questions
What does brand health actually measure for a luxury house?
Brand health in luxury measures perceived desirability, exclusivity and craftsmanship reputation, not just transactions. Because that perception doesn't show up in basket data, houses triangulate four signal families: editorial and earned media coverage, search interest, social engagement and sentiment, and resale market behaviour. A single source, especially social volume, gives a distorted picture.
How is Earned Media Value calculated, and why do two vendors give different numbers?
EMV is typically estimated impressions × a benchmark engagement rate × a media value multiplier. The multiplier is a vendor assumption, not an observed market price, so two providers analysing the same Instagram post can produce figures differing by 2 to 3x. There is no standard-setting body for earned media comparable to GAAP in accounting, which is why EMV should only be compared within one vendor, quarter over quarter.
Is share of voice a reliable indicator after a runway show?
No, share of voice measures volume of attention, not quality. A spike can come almost entirely from a viral controversy, such as a front-row seating row, with no link to product desirability, while a smaller spike driven by trade press praise (WWD, Business of Fashion) correlates better with next-season sell-through. Use SOV only alongside a quality signal such as editorial sentiment or sustained branded search.
Which metric best predicts real desirability after a fashion week?
Sustained branded search interest 30 days after the show is the strongest freely observable proxy. Search spikes around runway shows normally decay back to baseline within 2 to 4 weeks unless reinforced by a product drop or celebrity placement, so what survives the news cycle reveals genuine intent. Branded queries ("Prada re-nylon bag") also signal far higher intent than category queries ("waterproof designer tote").
What should I check before trusting a brand tracking dashboard?
Six points: source transparency (which outlets, platforms and languages are included, and are trade press and tabloids weighted differently), deduplication of syndicated articles, disclosed bot and fake engagement filtering, consistent time windows across brands, geography normalisation, and manual spot-checks of the NLP sentiment model. Comparing a 7-day post-show window for one brand to a 30-day window for a competitor invalidates the whole comparison. The AMEC Integrated Evaluation Framework is a free, vendor-neutral reference for auditing communications measurement.
Why does Google Trends fail for tracking brand interest in China?
Google has negligible usage in China, where Baidu dominates search and Weibo carries much of the conversation, so Google Trends returns near-meaningless data for that market. Any brand health comparison must specify its geography (US, EU, China) because press ecosystems and platform usage differ sharply. Running one market's methodology across all regions produces false gaps between brands with different regional footprints.