Fractal Summary： 2024／4／29 Customer Value and Delivery Method

From /villagepump/フラクタル要約:2024/4/29 from FractalReader Operation Diary Fractal Summary: 2024/4/29

I have various thoughts I want to ponder, but for now, I want to focus on the exam, so I’ll put down the thoughts in my head here (/villagepump/blu3mo):

• I won’t write replies or anything on this page for a while.

• #quoria

• Retention Rate

  • If people who regularly read news articles or papers think of using “Fractal Summary” at their usual reading times, they will use it.
    • How to eliminate the barriers in the process of getting them to that point
  • PDF drop, URL drop, Chrome extension
    • Considering the development costs, I want to think about the priority.

• Using LLM

  • Currently, in terms of pricing, API costs seem to be the bottleneck, with API costs requiring 50-150% of what users can afford (estimated).
  • Investigate if tools like Llama can be used.
  • Hypothesis: “API costs will become negligible in a year”
    • Depending on whether this is true, the business model will change significantly.
    • For example, in six months, an open-source model like GPT4 may be available at 1/10 of the price.
    • If we assume this, API costs will no longer be a bottleneck in pricing.

• Enhancing the Product

  • Mobile compatibility
    • Automatically scroll other columns as well
  • Quality as a reader
    • Readability
    • It feels like a deep dive, but I’m interested.
  • Where else can we improve?
    • I feel like my perspective on “where we can improve the product” is narrow, so I’ll think about it again later.
  • Before enhancing the product, this is more important 663041ea79e1130000f61b89
    • If there is an MVP, resources should be focused on exploring how to provide value.

• A certain individual, N 662faffa79e113000069b414

  • (/villagepump/blu3mo) is interested in making this a viable business.
    • Allocate 10-20% of my time to prioritize value for customers and create something once
    • I want to practice creating things for users and building a business, and this seems like a good subject for that.
  • The individual, N, personally does not have a strong interest in creating a product that stands as a business.
    • Instead, they seem more interested in experimenting and researching better systems.
    • There is a possibility that the insights gained there could be applied to the products of the affiliated company.

• How can the product’s value be increased by providing it to customers in what form?

  • Superficially, it seems like the value would be in “individuals being able to read articles quickly and deeply -> making those individuals happy.”
    • Monetizing based on that customer value would involve “charging those individuals through a subscription,” for example.
  • However, there seems to be additional value that can be generated by the use of FractalReader.
    • I want to explore this aspect.
    • Understanding from 662faffa79e113000069b419 (/villagepump/blu3mo)
    • Discussion on Non-public Information Held by Companies
      • There might be people who see value in “information that is not read and is lying dormant being read.”
  • By the way, [/blu3mo/20230824 TeaRoom岩本さん#64e8c38979e1130000d2f708](https://scrapbox.io/blu3mo/20230824 TeaRoom岩本さん#64e8c38979e1130000d2f708) was also talking about that.
    • By taking a broad view of society and identifying “If X is here, this person will feel value,” you can design a mechanism for money to flow into X.
    • Even something that seems unlikely to sell could be successful if someone designs that mechanism effectively.- Is it possible to use methods like interviews, hearings, or market research to find things like this?
  • It seems unlikely that you can find it just by brainstorming alone.
  • Observing the real world is necessary.
  • Another approach is embedding hypotheses and measurements into the product, analyzing measurement logs, such as determining the percentage of people accessing up to this point, assuming that users will move from using feature A to feature B but end up not going to B, instead focusing on C, etc., to understand “actual behavior” as “data.”
  • Embedding them to be understood as data. Lean’s [BML loop].

• Does PMF refer to when this is found?

• The feeling of truly understanding the meaning of “common startup terms”

  • (Not sure if it’s the “true meaning,” but at least understanding has deepened by one level)

• Turn end. This page will not be opened until 4.