Early stage · research prototype

Customer conversations move before revenue does.

I turn what customers say online into leading indicators of consumer-brand revenue. Each brand gets its own signal, tested against its reported sales.

See the first result Get in touch
lead time Signal Revenue PAST TODAY AHEAD

01 · First result · Hermès leather goods

For Hermès, the signal was how the bag fits clients’ lives. Sales followed a quarter later.

Every brand has its own leading signal. Finding it, brand by brand, is the work.

10% 20% Sales 10.8% Signal ≈10% Q1 ’24 Q3 ’24 Q1 ’25 Q3 ’25 Q1 ’26 Q3 ’26
What clients said 3 months earlierWhat clients say What Hermès sold
~3 monthslead time before reported sales
0.93rank correlation with sales one quarter later
4 of 4times it disagreed with analysts, it was right

Research prototype: 417K public Reddit conversations, Aug 2023 – Aug 2026. Sales: Hermès leather goods at constant exchange rates, from company reports. Errors measured leaving each quarter out. Small sample: about 10 quarters, one brand, signal chosen in-sample. Independent analysis, not affiliated with Hermès.

02 · Against analysts’ forecasts

When analysts were wrong, the signal saw it.

The rule: a call only when the signal and the analysts differ by at least 1 point. Otherwise, no call.

ReportAnalystsSignalCallActual
Q1 202415%18.8%Above20.3%Right
Q4 202413%18.9%Above21.5%Right
Q1 20259%10.7%Above10.0%Right
Q3 202514%10.8%Below13.3%Right
Q2 202610.8%11.2%No call10.2%—
Q3 2026~10%9.8%No call10.8%—

Backtest, not live calls: each quarter left out in turn. Analysts’ forecasts for leather goods as reported in the press (Reuters/Visible Alpha, broker notes); Q3 2026 approximate.

03 · Who it’s for

About three months before the report, the signal already shows the turn.

Investors

Will the report beat analysts’ forecasts?

Buy or hedge before results, when the signal and the analysts disagree.

Q4 2024: analysts +13%, signal ≈ +19%, actual +21.5%. Shares rose ~4% on the day.

Analysts

Is my estimate too high or too low?

A second opinion from customer conversations, a quarter early. Adjust the forecast before the report.

Q3 2025: analysts +14%, signal ≈ +11%, actual +13.3%. The signal flagged the risk.

Brands

Is demand turning, and why?

Plan stock, prices and campaigns with time to act. See which topic in customer talk moves first.

For Hermès, talk about how the bag fits clients’ lives moved first. Every brand gets its own signal.

Examples come from the backtest, not live calls. Not investment advice.

04 · Next: prove it live

Every forecast, dated before the report.

1

Licensed data

Rebuild the Hermès signal on a licensed feed.

2

More brands

Find each brand’s own signal: luxury, apparel, footwear, beauty.

3

Live forecasts

A dated forecast before every report, scored here against analysts.

4

First brand pilot

The signal on a brand’s own metric, with its team.

ForecastSignalAnalystsActual
Q3 2026
Hermès leather · within 1 pp, no call
~10%~10%10.8%
FY 2026
Hermès leather · due Feb 2027
[__%][__%]Pending

05 · Founder

Built alone, from raw conversations to a tested signal.

Maryna Said · BSc Artificial Intelligence, JKU Linz · data and analytics focus on luxury and high fashion.

3 yearsof customer conversations collected, 2023–2026
417Kcomments analysed, 45K statements tagged by stage, topic and tone
1working signal, tested against a brand’s reported sales

Get in touch

Looking for data partners, early investors and pilot brands.

Maryna Said · Linz, Austria

saidmaryna@icloud.com LinkedIn

Independent research. Not affiliated with any brand mentioned. Not investment advice.