How CocktailHQ works

The recipes

CocktailHQ publishes 193 recipes: 142 cocktails, 11 low-alcohol drinks and 40 alcohol-free drinks. Each is written and reviewed by hand, with its measures in millilitres, its method, glass, garnish and the date it was last reviewed. A recipe is not added because a name is popular; it is added when there is something worth saying about how it is made.

Measures follow a widely made modern specification. Well-known variants are recorded on the recipe as variations, with what changes and why, rather than published as separate near-identical pages.

Strength, ABV and UK units

Each recipe's strength is an estimate, worked out from its measures and typical bottled strengths and recorded with the assumptions behind it and a confidence level — both shown on the recipe page under “Show calculation assumptions”. UK units are the millilitres of the drink multiplied by its ABV and divided by 1,000: the NHS definition, where one unit is 10ml of pure alcohol.

These are estimates. Bottled strengths vary by brand (a gin can be anywhere from 37.5% to over 47%), pours vary, and so does dilution. Figures are shown with that uncertainty rather than to false precision, and they are information, not a target — see Responsible Drinking.

Dilution

Chilling a drink with ice also waters it, and the two cannot be separated. CocktailHQ uses a typical range for each method, expressed as the share of the finished drink that is meltwater:

The stirred and shaken ranges follow standard bar teaching. The others are reasoned from those rather than measured: a built drink dilutes in the glass as you drink rather than before the first sip, a blended drink has its ice inside it, and a layered drink is barely mixed at all. Real dilution depends on ice size, temperature and how long you work, which is why the batching calculator lets you change the figure.

Taste, aroma and structure

Each recipe carries editor-assigned scores from 0 to 5 for sweetness, sourness, bitterness, strength, fruitiness, freshness, creaminess and smokiness. They are judgements on a consistent scale, not lab measurements, and they let drinks be compared with each other.

The deeper reading on each recipe page — which ingredient is the sour element, which carries the sugar, how the drink is diluted and served, where its aroma comes from — is worked out from the recipe itself. Where an ingredient's job in a recipe was stated by an editor, that is used; where it was not, it is inferred from what kind of ingredient it is and how much is used, and the page says so rather than presenting the inference with the same authority.

Alcohol-free and low-alcohol

Alcohol-free means the recipe contains no alcoholic ingredient. Some alcohol-free products — de-alcoholised wines, some alcohol-free spirits, kombucha — can contain a trace of alcohol, and recipes using them say so. Most aromatic bitters are strongly alcoholic even though only a dash is used; recipes that use them are not classed as alcohol-free.

Low-alcohol is assigned recipe by recipe after review, never to a whole family: a spritz is not low-alcohol just because it is a spritz. A drink needs a unit figure clearly below the catalogue norm and a composition that is genuinely built to be lighter. Everything else stays classed as a cocktail. Alcohol-free is never presented as a health claim.

Substitutions

A substitute is judged on three things: whether it can do the same job in that recipe (weighted most), how close its sweetness, acidity, bitterness and intensity are, and how it changes the drink's strength. Substitutes an editor has written for a recipe are always shown first. A swap that changes the drink enough to make it a different drink says so, and some ingredients are recorded as having no good substitute at all, which is a real answer.

Glassware and alternatives

Every recipe's glass is resolved to one of a fixed set of glass types with known shapes and capacities. Alternatives are suggested only where they suit the drink's volume and service — a stemmed glass for a drink served without ice, a tall glass for a long drink — and the page explains what changes: how warm it gets, how much aroma reaches you, how it looks.

Each recipe's glass and ice panel works out how much glass the finished drink needs from its measures and its method's typical dilution, and checks the size of your own glass if you enter it. The working assumptions:

• A drink with no ice needs a little of the glass left empty — about a tenth — so it can be carried without spilling and has room for aroma.
• Filled to the top with cubes or packed crushed ice, roughly half to three-fifths of a glass is left for the drink.
• A drink built in the glass dilutes from the ice already there, so its melt water is not added on top.
• One large cube takes up roughly 65–125ml, from a sphere to a 5cm cube.
• Glass capacities are brimful and vary by maker — your own glass, measured with water, is the figure to trust.

The ice in the glass follows from how the drink is served — none for a drink served up, one large cube for a short spirit-forward drink, cubes to the top for a long one, crushed ice where the drink is built to lengthen as you drink it — and the panel says what changing it would do.

Making ahead and freezing

Whether a drink can be made ahead is decided by what is in it. Anything fizzy, fresh citrus, fresh herbs and cut fruit, dairy and egg are added at service; everything else can be bottled ahead and kept cold. A drink with a shaken foam, dairy alongside citrus, a blended texture or crushed ice is made to order, though its ingredients can still be measured out in advance. CocktailHQ does not give precise shelf lives: how long a batch keeps depends on the bottle, the fridge and the ingredients you used.

In the Spec Lab, a batch is judged for the freezer by its strength before any water is added. At about 35% ABV or more it stays liquid in a −18°C home freezer; between 25% and 35% it turns slushy; below that it freezes. Sugar lowers the freezing point a little further, so these are the cautious figures. Anything fresh or fizzy is never suggested for the freezer.

Party quantities and fresh fruit

The party planner scales each recipe to the number of drinks you choose — it never suggests how much anyone should drink — and turns the juice and citrus garnishes into fruit to buy, always as a range:

• Juice per fruit, hand-juiced: lime 25–30ml, lemon 40–45ml, orange 60–80ml, grapefruit 120–180ml. Small, cold or hard fruit gives less; roll it and bring it to room temperature first.
• Garnish pieces per fruit: 6–8 wedges, 5–7 wheels or slices, 4–6 strips of peel.
• Peel the fruit you are going to juice before you juice it: twists and peels then cost no extra fruit. Wedges and wheels do.
• The larger number is the one to buy. Fresh juice is best used the day it is squeezed.

Recommendations and tools

The Finder, What Can I Make, the Recipe Doctor, the Spec Lab and the calculators are deterministic: the same input always gives the same answer, from rules that can be read and corrected. Nothing is generated by AI on request, and nothing is personalised beyond what you enter. The Recipe Doctor offers a fault only when the recipe can genuinely have it — across the catalogue it knows 9.3 faults per recipe on average, not every fault for every drink. Where a fault depends on something in the recipe, it says what — the ingredient, the measure or the method — so you can check the reasoning against the drink in front of you.

The Spec Lab's figures for volume, strength, dilution and sweetness are arithmetic on the measures you enter. Its notes on balance are inferences from patterns across classic recipes: they say what a drink is likely to taste like, not what it does. Only tasting tells you that.

What you enter into a tool stays in your browser or is used only to compute your answer. It is never published.

What stays on your device

Your bar inventory, your saved drinks and your own versions of recipes — changed measures, a garnish, a private 1–5 rating, the date you last made it and a note — are kept in this browser's storage. They are not sent to CocktailHQ, not included in analytics, not shown to anyone else and not indexed. There is no count of drinks made and no streak. Clearing your browser's site data removes them; each one can also be deleted where it is shown.

Analytics on CocktailHQ, where you have allowed it, records which pages and tools are used, and what was searched for with email addresses and long numbers removed. It never includes your bar, your notes, measures or recipes you enter into the Spec Lab, or prices you enter into the cost calculator — those are used only to work out your answer and are not stored.

History and attribution

Cocktail history is full of confident stories with little evidence behind them. The claims in the culture section are each graded as documented, widely reported, disputed or folklore, and say which. A recipe is placed in a historical era only when its dating is not in dispute; where it is, the recipe is given no era rather than a guess.

Images

A recipe shows a photograph only when an approved image of that drink exists. Images come from three places: licensed stock photography, used under its stated licence and credited wherever the licence requires it; images CocktailHQ's publisher produced; and AI-generated images, each recorded as AI-generated and checked by a person for the right drink, glass, garnish and ice before it is published. An AI-generated image is labelled as one beneath it. Where no approved image exists, the recipe shows a plain colour panel rather than a stand-in photograph of something else.

Diagrams and charts on CocktailHQ are drawn from the site's own data and are its own work — the research findings among them.

What CocktailHQ cannot tell you

It cannot tell you how a specific bottle will taste, measure your drink, or replace your own palate — a recipe is a starting point, and tasting as you go is always the final step. It does not give medical or nutritional advice. And it is edited by people, so it will sometimes be wrong: corrections are always welcome and acted on.