Pick the same columns for every agency you talk to. Ask the same questions in the same order. Write every answer on the same sheet. If each agency gets a different set of questions, you are comparing sales talk, not terms.

Local option: New Plan Care gives families in the St. Louis area in-home help so a parent can stay home: companionship, help with bathing and dressing, meals, and overnight care. Call (314) 405-0887.

The columns

  • How a visit is billed: the unit, the minimum block, and where the clock starts.
  • Waiting time: whether idle time bills, and under what line.
  • Night, weekend, and holiday terms: a listed table, or a promise to call you later.
  • Travel: any charge to reach the house.
  • Notice: how much notice ends or changes a visit, in writing.
  • Replacement staff: who decides, and how fast you hear about it.
  • After-hours path: who answers when the office is closed.
  • Quote life: how long the written quote holds.
  • Public rates: whether the agency posts its prices on its own site.

Mark each cell with one of four marks

Use only these: written, verbal, unclear, blank. Written means it is in a document you can keep. Verbal means someone said it on a call. Unclear means you heard a sentence that could mean two things. Blank means the agency did not answer.

Blank is a finding. An agency that does not publish its rates, and does not answer on the first call, has told you how it will handle money after you sign. Write that down the same way you write the good answers.

What breaks a tie

When two agencies look close, the written answer wins. A warm voice on the phone does not replace a line in the quote. Ask each one to send its answers in writing, then compare the documents, not your memory of the calls.

Keep the sheet honest

Do not score by feel. Do not fill a blank with what you assume they meant. Keep the sheet with your notes from every call and every visit, and add to it as the answers come in.

The site's own method for rating agencies is laid out on How We Rate. Read it to see the same plain test applied in public, then run your own version on the agencies you are actually weighing.