Which job do you take first?

Paste the jobs you could take, then answer a handful of this-or-that questions. You never score anything — you just pick one — and the answers are assembled into a ranked work order.

After using this, you can now turn a list of possible jobs into an order you can act on, which used to mean staring at a whiteboard or scoring every job out of ten and getting twelve near-identical numbers back.

New to this? How do I decide which job to take first? covers what actually drives the order and how to run the comparison, on paper or here.

Your list

One job per line — a name and enough detail to tell them apart. Nothing is uploaded.

Your job list stays on this device. Nothing is uploaded. Usage counts contain only feature names — never job names, never anything about you — and are never sent anywhere.

Common questions

Why answer this-or-that questions instead of just scoring each job?

Because scoring a job on its own is a question nobody can answer consistently. Given twelve jobs and a 1-to-10 box, the seventh one gets scored against a standard that has already drifted from the first, and the result is a list of near-identical numbers. A head-to-head question is different: you are comparing two concrete jobs against each other, with every trade-off you care about — margin, travel, how the client pays, whether it fits the crew you have next week — folded into one judgement you can actually make. The ranking is assembled from those answers rather than asked for directly.

How many questions will it ask me?

About n times log2 of n, so roughly 22 questions for 8 jobs, 34 for 12, and 90 for 25. The counter above the buttons shows the estimate and how far through it you are. It is a ceiling rather than a promise — the sort stops asking about a group as soon as the answer is settled, so most runs finish a little early, and none run long.

What happens if I contradict myself?

You will still get a complete ranking, but it may not be the one you would have described. The method assumes that preferring A to B and B to C means you prefer A to C, and it relies on that to avoid asking about every possible pair — which is what keeps the question count to a few dozen instead of a few hundred. If your answers disagree with each other, the result reflects the order the questions happened to come in. If a finished ranking looks wrong to you, that disagreement is the useful signal: it usually means a job has a factor you had not admitted to yourself, not that the list is unrankable.

Does my job list leave my browser?

No. There is no account, no server, and no analytics vendor. Your list and your answers are held in this browser, and an unfinished ranking is saved to this browser's local storage so closing the tab does not lose your progress. Nothing is uploaded at any point.

What does the share link contain?

Your list of jobs, encoded after the # in the address — and not your answers or your ranking. A browser never sends the part after # to a server, so pasting that link into a text or an email is the only way it travels. Sending the list rather than the result is deliberate: it lets a partner or a foreman rank the same jobs independently and then compare, which is the conversation two people who disagree about what to take on actually need.

Is this only for contractors?

No — it ranks any list where the items are hard to score but easy to compare two at a time. Jobs and quotes are the case it was built for, because a trade with more work than weeks is the person who most often has to choose. It works the same way on a feature backlog, a shortlist of suppliers, or a stack of quotes you are deciding which to chase first.