✦ Task type · Research and checks
Podcast feedback, timestamped to where they drifted
Your analytics show a drop at eleven minutes. They cannot tell you whether people left bored, confused, or because the episode had already answered the question.
Podcast analytics are unusually blunt: a drop-off curve and nothing about cause. The curve tells you when, and the entire useful question is why.
What to put in the public brief
- Episode length, honestly
- The topic in general terms
- That they will listen and then write
- Whether headphones or a specific listening context matters
- Roughly how long in total
Keep the show identity private where you can, so responses are about the episode rather than the brand.
What to put in the private instructions
Ask where they would have stopped if not being paid
The single most useful question, and it needs permission to be answered honestly. Say plainly that "after four minutes" is a valid and well-paid answer.
Ask for three timestamps in their own words
One moment that held them, one that lost them, one they would quote. Chosen by them, not prompted by you.
Ask what they would tell a friend it was about
If that does not match your intent, the episode has a framing problem no amount of editing the middle will fix.
Ask about the first ninety seconds separately
It is where most of the loss happens and it is usually the part that was recorded last and edited least.
Proof worth requiring
Ask for
Three timestamps with a sentence each, the stopping point, the one-line description, and how they listened — headphones, car, speaker.
Do not ask for
A rating or a score. It compresses a drift into a number, and you will average the numbers across people who drifted at completely different places.
When this task type is the wrong tool
If you want to know whether people find the show at all, that is a discovery question and this measures the wrong thing entirely.
If the issue is audio quality rather than content, name that as the task instead — a listener asked about content will politely absorb a hum they would never tolerate unpaid.
Common questions
Should listeners be fans or strangers?
Strangers, for a first pass. Existing listeners have already accepted the format and will report on the content; strangers report on the format, which is what decides whether anyone becomes an existing listener.
What reward is fair?
Roughly a dollar a minute of listening, plus a few dollars for the written response. A forty minute episode is a real commitment and paying below that produces skimmed listening reported as full listening.
How do I know they actually listened?
Ask for three timestamps with a sentence each, chosen by them rather than by you. Someone who skimmed cannot produce three specific moments, and asking after the fact makes it much harder to fake than a summary would be.
Should I ask whether they liked it?
It is the least useful question available. Ask where their attention went, where they would have stopped if unpaid, and what they would tell a friend it was about. All three are answerable and all three are actionable.
Thinking of taking this task?
The same job, from the other side. Rewards are set by the poster, so treat these as the range this task type usually sits in rather than a promise.
- Usually pays
- About a dollar a listening minute
- You need
- Headphones or a quiet space
- Proof expected
- Three timestamps with a sentence each, plus your stopping point
Where you would have stopped if you were not being paid is the most valuable answer, and after four minutes is a perfectly good one.
Create a worker accountTest an episode
Set the reward and how many listeners you need. Workers reserve a spot; the episode stays private until it is theirs.
Post this task