Early signals for beverage volumes, with what each one is and is not · built 01 Nov 2026, 20:30
Early signals that move before companies report. Each chart says what it is and what it is not. Growth is read as three months against the same three months a year earlier, never a single month. How useful each one is has not been tested yet; the labels are my prior, to be replaced by a measured relationship against reported volumes.
Lakh bulk litres, April 2025 to March 2026. The state's own excise table, the one official monthly beer series found. One year only, so no year-on-year yet.
Growth of UPI value in each category minus growth of all UPI, 3 months on 3 months, percentage points. Shown from January 2025: before that the liquor-shop category did not exist (it starts Nov 2023) and bars has gaps, so early growth reflects coverage, not spending. Direction only: cash and cards are missing and it is value, not volume.
Production volume, % change 3 months on 3 months. Output, not sales; Germany and France only because Eurostat has nothing at this level for the others.
% change 3 months on 3 months. All retailing, not drinks: a backdrop for demand, not a beverage signal.
% change on a year earlier; January and February are one combined figure. Restaurant spending in yuan including food and prices, not beer volume. The free mirror is only about a year deep.
Consumer prices, % change on a year earlier (Eurostat harmonised index). Price is half of every value series on this page: catering and on-trade revenue are price times volume, so a flat revenue line with rising prices means falling volume.
China has no free equivalent: the statistics bureau's price index only reports food, tobacco, liquor and dining out as one group, and the free mirror holds about a year of it. The nearest thing is revenue per hectolitre in company results.
Search interest by country and beverage category, from Google Trends. It measures what people looked up, not what they bought or how they feel, and it is only worth watching if it tracks something real. The table under the chart tests that.
Average change in the Google Trends index, % . Each term is indexed to its own peak within its own country, so compare changes, never levels, across rows or columns.
A dash means that term has not loaded yet: Google throttled the Denmark and Sweden queries and Poland iced tea and kombucha, and they will be filled on the next run.
Year-on-year change in monthly beer interest against (1) the year-on-year change in that country's beer volume and (2) the change in hot days. Correlations over the months shown, COVID months left out. Close to zero means the series says nothing about beer volume.
| Source | Country | Term | Months vs volume | Corr, beer volume | Months vs heat | Corr, hot days |
|---|
Reported volume growth, the thing the indicators above are meant to anticipate. "Read in full" means I opened the source and the figure matches it; "could not confirm" means no source states it and it is shown dimmed. Where a source disagreed with an earlier figure, the row was corrected and says so.
| Company | Market | Period | Measure | Change | Basis | Checked |
|---|
Quarterly growth in each indicator against Carlsberg's reported beer volume growth, COVID quarters left out. A rating is only given with at least eight quarters of overlap; the rest stay untested. This replaces my prior where there is enough data.
| Indicator | Against | Quarters | Correlation | t | Rating |
|---|
| Series | What it is | What it is not | Latest | Prior use |
|---|
Every source in the build prompt was tried. Those that did not give a live series are logged with the reason.
| Source | Status | Latest | Note |
|---|