My first stalled fermentation wasn't a temperature problem or a sanitation problem. It was arithmetic. I'd bought a liquid yeast pack that had been sitting in a shop fridge since who knows when, pitched it into a 1.064 wort, and then spent four days staring at a motionless airlock. The beer eventually came around, but it finished high and tasted faintly of solvent. What I'd done was ask a small, tired population of cells to do a job that needed roughly twice as many of them. A starter fixes that before brew day, and it costs almost nothing.
The cell count problem nobody prints on the pack
A yeast pack looks like a single serving, so it's easy to assume one pack equals one batch. The number underneath tells a different story. The American Homebrewers Association puts a standard liquid yeast package at roughly 100 billion cells, and describes that as enough for a five-gallon batch at about 1.048 original gravity — a modest pale ale, in other words, and only if the pack is fresh. Push the gravity up to a stout or a barleywine and the same pack is suddenly stretched thin. Liquid yeast also loses viability as it sits; a pack several months past its production date is carrying noticeably fewer live cells than the label implies, even when it swells up normally.
Underpitching doesn't usually announce itself as a total failure. It shows up as a long lag before anything happens, which hands wild microbes an open window, and then as a fermentation that runs hot and stressed once it finally gets going. The yeast compensates for its small numbers with extra budding cycles, and those growth phases are exactly when it throws off the most esters and fusel alcohols. That solvent note in my stalled batch wasn't contamination — it was overworked yeast. A stubbornly high finishing gravity often traces back to the same root cause.
Dry yeast changes the math completely. An 11.5-gram sachet carries several times the cell count of a liquid pack, which is why nobody builds starters for it — rehydrate as the manufacturer specifies and move on. Everything below applies to liquid yeast.
Building a one-liter starter with kitchen gear
The recipe is one ingredient and a scale. The AHA's ratio is one gram of dry malt extract for every ten milliliters of target volume, which for a one-liter starter means 100 grams of DME in a liter of water. That lands the gravity in the 1.030 to 1.040 band they recommend — strong enough to feed real growth, dilute enough that the yeast isn't fighting osmotic pressure the whole time.
Stir the DME into cool water first so it doesn't clump into a scorched lump on the bottom of the pan, then boil it for about ten minutes to sterilize it. Cover the pan with foil, drop it into an ice bath, and get it under 75°F before it touches yeast. Everything from that point forward — the jar or flask, the foil cap, the spoon, the outside of the yeast pack — gets the same sanitizer treatment you'd give bottling gear.
Cap the vessel with loose foil rather than an airlock. A starter is not trying to make beer; it's trying to make yeast, and yeast multiplies best with oxygen available. Foil keeps dust and insects out while letting the vessel breathe. Hold it around 72°F, per the AHA's guidance, running a few degrees warmer for ale strains and a few cooler for lager strains.
Stir plate, shaking, or neither
How you agitate the starter determines how much yeast you actually grow, and there are three tiers worth knowing. A magnetic stir plate is the top of the range and the AHA recommends one, because keeping the wort in constant motion continuously works oxygen into it and keeps the cells suspended in their food rather than settled in a layer at the bottom. Cheap plates exist, and plenty of brewers build their own from a computer fan and a hard-drive magnet.
The middle tier is intermittent shaking, and it works better than its reputation suggests. Swirl the flask hard every time you walk past it — half a dozen times across the day is plenty. You won't match a stir plate, but you'll comfortably beat leaving it alone, and it costs nothing.
The bottom tier is a static starter that just sits there. It still grows yeast, and it still beats pitching an old pack straight into your wort, but the gain is modest — give it a shake now and then.
Pitching the whole thing vs. decanting first
There are two ways to get the yeast out of the flask and into the beer, and they suit different situations. The convenient one is pitching at high krausen — the peak of the starter's own fermentation, which the AHA places at roughly 12 to 18 hours after you pitch the pack. The cells are actively dividing and ready to go. The one rule is thermal: get the starter within about 5 to 15°F of your wort before they meet, because dumping warm starter into cold wort or the reverse shocks the yeast and causes exactly the sluggish start you built it to avoid.
The tidier option is to cold crash instead. Put the finished starter in the fridge for a day, let the yeast pack down into a firm layer, and pour off the spent liquid before pitching the slurry. This matters more than it sounds like it should. A liter of starter going into five gallons is about five percent of your finished volume, and that liter is oxidized, unhopped, and tastes like nothing you want in a delicate beer. In a robust porter you'd never notice. In a light lager or a crisp pilsner you can. I decant for anything pale and clean, and pitch whole into anything dark or assertive.
The batches where I skip it entirely
Starters aren't a virtue in themselves, and there's no prize for building one you didn't need. I skip it for any dry yeast, full stop. I skip it for a fresh liquid pack going into a standard-gravity ale — anything at or below about 1.050 — because that's the case the pack is sized for. I skip it for small batches, since two and a half gallons effectively doubles the pitch rate of a single pack on its own.
I never skip it for a lager, because lager fermentation conventionally wants around double the cells per gallon that an ale does, and a single pack simply cannot cover that. I never skip it above roughly 1.060, where the extra sugar demands more cells and the rising alcohol makes the yeast's life harder as it goes. And I never skip it for a pack more than a few months past its production date, regardless of what I'm brewing, because the number on the box stopped being true a while ago.
How I decide now, in about ten seconds
The whole judgment collapses into three questions I ask standing in front of the fridge on the Sunday before brew day. Is it dry yeast? Then there's nothing to do. Is the liquid pack fresh and the target gravity under about 1.050 for an ale? Then pitch it straight. Anything else — an older pack, a big beer, or any lager — gets a one-liter starter started two days out, so it has time to hit high krausen and, if I'm decanting, a night in the fridge afterward.
The build itself is twenty minutes of actual work: 100 grams of DME, a liter of water, a ten-minute boil, an ice bath, sanitized foil, and a shake whenever I pass the counter. Two days later I've turned a questionable pack into a pitch I trust, and fermentation starts within twelve hours instead of leaving me refreshing forum threads at midnight. Read the AHA's full method once before your first attempt — How to Make a Yeast Starter — and after that it's muscle memory.