The problem is almost never a shortage of recipes
If you are standing in the kitchen at six o'clock with no idea what to make, the internet is not short of suggestions. That is exactly the problem. Open a recipe site and you are handed thousands of options, most of which need an ingredient you do not have, a technique you do not want tonight, or forty minutes you have already spent looking.
Deciding what to cook is a filtering problem, not a search problem. The useful move is to cut the options down before you start looking, not after. This guide is a ten-minute routine for doing that. It works on paper, and it works faster in an app, but the order of the steps is what matters.
Set a timer for ten minutes
This sounds like a gimmick. It is not. Dinner decisions expand to fill whatever time you give them, and the twenty-fifth recipe you look at is rarely better than the third. A visible limit turns an open-ended browse into a task with an end.
Ten minutes is enough to check what you have, apply two or three constraints, and commit. When the timer ends, cook whatever is currently in the lead. A good dinner you actually started at 18:10 beats a better one you were still researching at 18:45.
Answer three questions, in this order
The order matters, because each answer removes options that the next question would otherwise have to consider.
First: how much time and attention do you have?
Not how much time you wish you had. Be honest about tonight. Twenty minutes with a toddler in the room is a different constraint from ninety quiet minutes, and it eliminates most of the internet immediately.
Sort your answer into a rough bucket: under 20 minutes, 20 to 40, or an evening where cooking is the activity. Almost every meal format falls neatly into one of those.
Second: who is eating, and what is genuinely off the table?
Count the servings, then list the hard constraints — allergies, medical dietary needs, religious requirements, and the one vegetable someone will not touch. Hard constraints are not preferences, and they are worth writing down once rather than remembering each night.
Keep this list short. If you treat every mild preference as a rule, you will filter yourself down to nothing.
Third: what in the kitchen needs using first?
Now, and only now, look at the food. Check the fridge shelves, any opened packages, cooked leftovers, soft vegetables, and anything that was thawed. You are looking for the two or three items that will not be as good in three days.
Use dates as one clue among several — storage history, smell, texture, and package condition matter too. Do not taste food to decide whether it is safe, and do not ask an AI tool to make that judgment. When you are unsure, throw it out.
The output of this step should be a list of no more than three ingredients. If you write down twelve, you have made a new decision problem instead of solving the original one.
Choose the shape of the meal before the recipe
Here is the step most people skip. Instead of searching for a dish, pick a format. A format is a structure that tolerates substitution, and choosing one collapses thousands of recipes into a handful of decisions.
- Bowl: a base, something substantial, a vegetable, and a sauce.
- Soup or stew: forgiving with mixed vegetables, beans, grains, and small leftovers.
- Frittata or omelet: built for cooked vegetables, herbs, cheese, and odds and ends.
- Stir-fry: works when everything can be cut to a similar size.
- Pasta or grain skillet: makes a small amount of one ingredient feel like a full meal.
- Traybake: the least hands-on option when you have time but no attention.
- Toast, wraps, or flatbread: the honest answer when there is not enough of any one thing.
Match the format to your time bucket and to the condition of the food. A crisp vegetable can go in a salad; the same vegetable, slightly soft but still fine, belongs in a soup or a pan.
Give yourself three candidates, then stop
Search with your constraints already applied rather than starting from a blank query. Lead with the ingredient that is hardest to use up, add the format, and add one limit. "Spinach frittata 20 minutes" is a workable query. "Dinner ideas" is not.
Stop at three candidates and pick one. Three is enough to feel like a choice and few enough to compare. This is also the point where a tool that starts from your own ingredients saves the most time: CookFeed's pantry-first recipe suggestions begin with what you marked as available, so the results are already filtered by the constraint that eliminates the most options.
Treat any generated or unfamiliar recipe as a draft. Check quantities, allergens, and cooking temperatures before you rely on it.
Keep a short list of meals that need no decision
Every household should have five to eight meals that can be made from shelf-stable and freezer staples, on autopilot, in under thirty minutes. Write them down and keep the ingredients in stock.
This list is not about cooking well. It is insurance against the nights when you are too tired to run any process at all, including this one. Once it exists, "I don't know what to cook" stops being a crisis and becomes a shrug.
When you still cannot decide
Two fallbacks work reliably. First, cook the ingredient that needs using and build the smallest possible meal around it — one pan, one starch, done. Second, repeat a meal you made in the last two weeks. Repetition is not a failure; it is how most people who eat well actually eat.
If the same decision keeps costing you an hour every evening, the real fix is upstream. Deciding once a week is far cheaper than deciding seven times, and a plan built around the food you already own is the version most likely to survive contact with a real week — that is the subject of our guide on building a weekly meal plan from your pantry.
Tonight, though, the goal is smaller. Ten minutes, three questions, one format, one meal.