> ## Documentation Index
> Fetch the complete documentation index at: https://docs.hired.tools/llms.txt
> Use this file to discover all available pages before exploring further.

# The weekly review

> Ten minutes that tell you which part of your search is broken, rather than how many applications you sent.

Counts are comforting and useless. "Twelve applications, three screens" does not tell you
whether to send more, rewrite the resume, or stop applying to that kind of company.

The review starts with the diagnosis and works outwards from it.

> Run my weekly review.

That is [`pipeline_review`](/workflows#weekly-pipeline-review). Here is what it does and why the
order matters.

## It starts with the diagnosis

`diagnose_search` does not report numbers — it names the step of the funnel that is losing
people:

| What it sees                     | What that means                                        |
| -------------------------------- | ------------------------------------------------------ |
| No responses at all              | A resume or targeting problem. Volume will not fix it. |
| Responses that die at the screen | A story problem. What you say in the first call.       |
| Interviews that do not convert   | Something else again, and worth naming separately.     |

It also returns per-step conversion, median days spent in each stage, weekly volume for
the last six weeks, applications that have gone quiet, and **the response rate of each
resume** — which is how you find out that the one you have been sending for six weeks is
not working.

Progress is measured by the furthest stage an application ever reached, so a rejection
after a final round counts as having got that far.

<Note>
  When there is not enough data yet, it says so plainly and the review stays short. A
  trend invented from four applications is worse than no trend.
</Note>

## Then the rest of the picture

The review reads the shape of the search, what is due within seven days, what has actually
moved, the open tasks, and the companies you are talking to — noting any without a website
on file, since that is what puts logos on the board.

## What you should get back

<Steps>
  <Step title="Two lines on where the search stands">
    Not a table. Two lines.
  </Step>

  <Step title="Anything stalled">
    Applied more than ten days ago with no movement, or a follow-up date that has passed.
    Named by company each time.
  </Step>

  <Step title="A prioritised list for the week">
    Most important first, each tied to a specific company.
  </Step>

  <Step title="The follow-up messages, drafted">
    Written from the timeline, so each one refers to what was actually said. A follow-up
    that mentions the thing the recruiter told you is a follow-up that gets answered.
  </Step>

  <Step title="Tasks created, with due dates">
    So the review produces work rather than a document about work.
  </Step>
</Steps>

## Ask for the bad news plainly

The review is instructed to lead with what the diagnosis found, and to be direct about it.
If the problem is the resume or the targeting, a list of follow-ups is the wrong answer —
it treats volume as the fix for something volume does not fix.

> Thirteen of your fourteen applications have had no response in two weeks. That is a
> signal about the resume or the targeting, not about you.

is more useful than a tidy status table, and it is what you should expect back.

That verdict needs ten applications behind it. Under ten, the diagnosis says so — "too
early to tell you anything useful", because around ten is where the numbers start meaning
something rather than describing luck — and the review stays short rather than reading a
trend into four cards.

## Closing things out

The most common thing a review surfaces is applications that ended and were never marked
as ended. Do it in one go:

> Close out everything I never heard back from — they're all ghosted.

Batch moves write a timeline entry and a follow-up reset for each one, exactly as if you
had moved them individually, so the funnel history stays intact. Ids that no longer exist
are skipped rather than failing the whole batch.

<Warning>
  Use `GHOSTED` for silence. Filing a non-response as a rejection tells `diagnose_search`
  that people are deciding against you, which is a different problem with different advice.
</Warning>

## And file the week

Once the review is done, the other half of Friday is getting the week back into the brain:

> Here's what I did this week: …

[`log_my_week`](/workflows#log-what-happened-this-week) sorts a ramble into the right places — role brain
dumps, highlights, timeline entries, stage moves, tasks and notes — then tells you what it
filed where and asks about anything ambiguous.

Ten minutes. It is the difference between a career record that is current and one that
describes the job you had two years ago.