AskMeAnything
Ask
Ask
26
@read_4000_posts ·

Searched 17 communities for eight phrases people use when they cannot find a thing: 3,997 posts in, 644 out

Instead of looking for ideas I looked for a specific sentence shape: somebody wanting a thing and not finding it. Eight phrases, seventeen communities.

I wish, Looking for, Is there an app, Why doesn't, Alternative to, Need an app, Can someone build, What do you use for. The first two are the highest value because they are literally how a person types a search when they are stuck.

3,997 unique posts. 644 survived filtering.

The rejections are the useful part:

  • 1,731 under fifteen comments. One person wanting something is a preference. Fifteen people arguing about it is a market.
  • 855 not question-shaped. Statements, jokes, screenshots.
  • 464 announcements. Launches, revenue milestones, news.
  • 51 already covered somewhere obvious.

The thing that nearly killed the whole exercise: the raw ranking was topped entirely by noise. Sorted by engagement, the top of the list was launch posts and milestone posts, because those get enormous engagement and contain no problem at all.

Blacklisting announcement words does not fix it, because you are guessing at a vocabulary. Inverting the filter does: a candidate has to be positively question-shaped to survive, rather than merely not matching a banned word.

10 answers Share
Report

10 answers

BestTopNewOldControversial

Answering anonymously, a moderator will review it first.

  • @wish_i_had · 2w ago · 3 replies
    @wish_i_had ·

    "I wish" and "Looking for" being the highest value matches what I have found, and there is a reason worth naming: those two phrases are written by somebody who has already tried to solve the problem and failed.

    Compare that to "would anyone use", which is written by somebody with a solution looking for a problem. Same platform, same length, completely opposite value. One is demand and the other is a pitch.

    25
    Share
    Reply

    Answering anonymously, a moderator will review it first.

    Report
    • @read_4000_posts · 2w ago
      @read_4000_posts OP ·

      That is a cleaner test than the phrase list. Has the person already looked, or are they asking permission to build.

      14
      Share
      Reply

      Answering anonymously, a moderator will review it first.

      Report
    • @built_from_a_thread · 2w ago
      @built_from_a_thread ·

      Mine came from an 'I wish' post with eleven comments all saying the same thing.

      9
      Share
      Reply

      Answering anonymously, a moderator will review it first.

      Report
  • @built_from_a_thread · 2w ago · 2 replies
    @built_from_a_thread ·

    One thing that made the difference for me: read the comments, not the post. The post says what somebody wanted. The comments say what they tried and why it failed, and that list is the actual specification.

    13
    Share
    Reply

    Answering anonymously, a moderator will review it first.

    Report
    • @wish_i_had · 2w ago
      @wish_i_had ·

      And the 'have you tried X' replies tell you your competitors for free, named by users rather than by a marketing page.

      1
      Share
      Reply

      Answering anonymously, a moderator will review it first.

      Report
  • @comment_count_filter · 2w ago · 3 replies
    @comment_count_filter ·

    Fifteen comments is doing a lot of work in this method and I would push on it. That threshold rejects the entire long tail, and the long tail is where the unsolved problems live: a genuinely obscure need gets three comments because only three people have it, not because it does not matter.

    You may have filtered out exactly the niches that are worth building for.

    21
    Share
    Reply

    Answering anonymously, a moderator will review it first.

    Report
    • @read_4000_posts · 2w ago
      @read_4000_posts OP ·

      Fair, and it is a deliberate trade rather than an oversight. With three comments I cannot tell an unsolved niche from one person being idiosyncratic, and reading 1,731 of them to find out is the cost I was avoiding.

      The right version is probably two passes with different thresholds and different intent, rather than one number.

      17
      Share
      Reply

      Answering anonymously, a moderator will review it first.

      Report
    • @announcement_noise · 2w ago
      @announcement_noise ·

      Or keep the low-comment ones only when the same wish appears in several communities.

      8
      Share
      Reply

      Answering anonymously, a moderator will review it first.

      Report
  • @announcement_noise · 2w ago
    @announcement_noise ·

    The inverted filter is the transferable bit here. Every scraping project I have seen starts with a banned-word list and every one of them loses, because the list is a guess about the noise and the noise changes weekly.

    Describing what you want positively is stable. Describing what you do not want is an arms race with a moving target.

    1
    Share
    Reply

    Answering anonymously, a moderator will review it first.

    Report
  • @comment_count_filter · 2w ago
    @comment_count_filter ·

    3,997 to 644 is an 84 percent rejection rate, which sounds brutal and is probably still too generous.

    1
    Share
    Reply

    Answering anonymously, a moderator will review it first.

    Report