left arrow icon

Publish one original data point

Original data is one of the few things a new site can offer that others will cite. The evidence is two success stories: Alex Ellman's page of LLM prices drew 12,000 visitors and over 100 backlinks after a Hacker News post, and a consultant's survey page drew links from more than 350 sites over time. AI engines lean on authority: pages rich in statistics are cited more, well-linked sites far more.

Evidence C, limited evidenceSituational

Stage
Stage 3. Faint pull, earn the right to distributeHow do more of the right people find it?
Category
Search and content
Time
1 day to 2 weeksAwiser's estimate of the active work
Cash
No cash
Live calls
No live calls
First result
Within weeks
Risk
LowA few days of effort at most; the risk is a dataset nobody sees, or one that exposes customer data.
Needs
A number nobody publishes in one place: aggregated product usage, a survey or compiled public dataPermission to publish any customer data, in aggregateA list of people who cover the topic

When does it fit?

  • You hold a number nobody publishes in one place: Google asks whether content provides original information, and Alex Ellman built a one-page table of LLM API prices in a couple of hours.[1][7]
  • You will show the method: survey standards ask for enough detail on how the research was done to allow independent checking.[3]
  • The topic is tied to a live story: Clear Spark Digital surveyed remote and hybrid workers because of the wave of return-to-office mandates.[8]

When should you skip it?

  • You only have common knowledge to restate: Google describes such commodity content as adding little unique insight.[2]
  • You expect AI engines to cite a new site: SE Ranking found sites with over 32K referring domains 3.5 times more likely to be cited by ChatGPT than sites with up to 200.[5]

How do you run it?

  1. Pick a question people already ask

    Choose a number your buyers or journalists look for and cannot find in one place.

  2. Use the fastest data you have

    Aggregated product usage or compiled public data takes days. A survey takes longer, because getting enough answers is the hard part.

  3. Clear the data

    Aggregate usage data so no customer can be identified, and check that your terms allow publishing it.

  4. Write the method down

    Who ran it, when, the sample and how it was collected, on the same page as the result.

  5. Lead with the finding

    One page, the main number stated plainly near the top, with a chart and the table behind it.

  6. Tell the people who cover the topic

    Pitch it to journalists, newsletters and communities that discuss the subject, not just one post.

  7. Log links and mentions

    Track linking sites and mentions for months, not days.

What usually happens?

  • Links can come fast after a hit: Alex Ellman reports over 100 backlinks within a day, from sites that pull popular Hacker News posts, and a Domain Authority of 13 on Ahrefs.[7]
  • More often they come slowly: Clear Spark Digital reports that a survey of over 200 entrepreneurs is now linked by more than 350 websites, mostly passively, as journalists cited it.[8]
  • Statistics help in AI answers, in lab data: in the GEO study on arXiv, adding statistics, quotations or cited sources to pages taken from the top 5 search results raised one visibility measure by 30 to 40%.[4]
  • And in correlation: SE Ranking found pages with 19 or more data points averaged 5.4 ChatGPT citations, against 2.8 for pages with little data, across 129,000 domains.[5]

Who has done it, and what happened?

Each case is one team's own report: what happened to them, not a typical result.

  • Alex Ellman, solo founder of Price Per Token2025, Positive result

    Compiled LLM API prices by hand into a one-page table in a couple of hours and posted it to Hacker News.

    Alex Ellman reports 12,000 unique visitors and over 100 newsletter subscribers by the next day.[7]

    Compiled public data rather than product data, a single Hacker News hit, and links mostly from aggregator sites.

  • Alex Ellman, the next post2025, Negative result

    Posted a new free tool, an AI coding tracker, to Hacker News.

    Alex Ellman reports that the post did not get a single upvote.[7]

    A tool rather than a dataset, and one attempt.

  • Founder Reports, a business publication, through a data-content consultant2025, Positive result

    Surveyed remote and hybrid workers as return-to-office mandates spread, and published the results.

    Clear Spark Digital reports links from 100+ websites to the page published in December 2025.[8]

    Reported by the consultant who sells this service, with no sampling method given; a 1,000-person survey is far beyond a quick survey of your own users.

What does the evidence say against it?

  • AI summaries cite big platforms most: the Pew Research Center found Wikipedia, YouTube and Reddit the most cited sources in Google AI summaries.[6]
  • Every documented case is a success story: no report was found of a small company's data page that drew no links, so survivorship is likely.

How do you read the result?

What to count
Linking sites and mentions of the finding
When to read it
One month, then three months
How to decide
Publish a second dataset only if the first earned links or mentions from sites your buyers read. Give it months: in the Founder Reports case, most links arrived passively over time.[8]

What goes wrong?

  • Hiding the method: survey standards ask researchers to name the sponsor and the dates of data collection.[3]
  • Betting the launch on one post: Alex Ellman, whose data page drew 12,000 visitors, got no upvotes on a later Hacker News post.[7]

How was it graded?

Evidence
C. Limited evidenceTwo success cases, one independent and one reported by a consultant who sells the service, plus a benchmark and a correlational study on statistics in AI answers. No base rate exists for how often a small company's data page gets cited, and no failure case was found.
Verdict
SituationalWorth it when you hold a number others want and you will pitch it. A new site should expect links from people, not citations from AI engines, which favor well-linked domains.

How the grades and verdicts are decided

Sources

  1. 1
    Creating helpful, reliable, people-first content (opens in a new tab)Google Search Central. Undated. Platform rules.
    Does the content provide original information, reporting, research, or analysis?
    Read on October 20, 2026.
  2. 2
    Optimizing your website for generative AI features on Google Search (opens in a new tab)Google Search Central. Undated. Platform rules.
    Commodity content (for example, something like "7 Tips for First-Time Homebuyers") is often based on common knowledge, which could originate from anyone, and typically adds little unique insight for readers.
    Read on October 20, 2026.
  3. 3
    AAPOR Disclosure Standards (opens in a new tab)American Association for Public Opinion Research. Undated. Method.
    Accordingly good professional practice imposes the obligation upon all public opinion and survey researchers to disclose sufficient information about how the research was conducted to allow for independent review and verification of research claims, regardless of the methodology used in the research.
    Name the sponsor of the research and the party(ies) who conducted it.
    Disclose the dates of data collection (e.g., data collection from January 15 through March 10 of 2019).
    Read on October 20, 2026.
  4. 4
    GEO: Generative Engine Optimization (KDD 2024) (opens in a new tab)arXiv, Pranjal Aggarwal and others. June 28, 2024. Study.
    Each query is augmented with the cleaned text content of the top 5 search results from the Google search engine.
    Specifically, our top-performing methods, Cite Sources, Quotation Addition, and Statistics Addition, achieved a relative improvement of 30-40% on the Position-Adjusted Word Count metric and 15-30% on the Subjective Impression metric.
    Read on October 20, 2026.
  5. 5
    How to optimize for ChatGPT: LLMs.txt doesn't matter but brand mentions on Quora and Reddit do (opens in a new tab)SE Ranking, Yulia Deda. November 24, 2025. Study.
    Sites with over 32K referring domains are 3.5x more likely to be cited by ChatGPT than those with up to 200 referring domains.
    To explore this, we analyzed a large dataset of 129,000 unique domains spanning 216,524 pages , covering 20 different niches to ensure a diverse and representative sample.
    Those rich in statistics (19+ data points) average 5.4 citations , compared to 2.8 for minimal data.
    Read on October 20, 2026.
  6. 6
    Google users are less likely to click on links when an AI summary appears in the results (opens in a new tab)Pew Research Center, Athena Chapekis, Anna Lieb. July 22, 2025. Study.
    The most frequently cited sources in both Google AI summaries and standard search results are Wikipedia, YouTube and Reddit.
    Read on October 20, 2026.
  7. 7
    What Happened When I Topped Hacker News and Pulled in 12,000 Readers Overnight (opens in a new tab)HackerNoon, Alex Ellman. August 14, 2025. Founder report.
    It was quite simple, just a website with a single page that displayed a table of the current per dollar prices of the major LLM APIs.
    I made the website in a couple hours, collecting the prices for each token manually and entering them into a csv file.
    By the next day I had acquired over 100 backlinks from websites that pull popular posts from Hacker News, leading to a quick climb to a Domain Authority of 13 on Ahrefs.
    By the time the next day arrived I had over 100 newsletter subscribers and 12,000 unique visitors had come to the website.
    I have posted to Hack News since, about the coding tracker I made, and didn't get a single upvote.
    Read on October 20, 2026.
  8. 8
    Founder Reports Case Study (opens in a new tab)Clear Spark Digital. Undated. Vendor data, from a seller of what it measures.
    I surveyed 1,000 remote and hybrid workers to gather data on the current state of remote work, which is a particularly relevant topic given the wave of return-to-office mandates.
    I led a survey of over 200 entrepreneurs about their mental health experiences.
    It's now been linked to by more than 350 different websites , including Forbes , Entrepreneur , Inc. , Fast Company , and Fortune .
    The majority of those links came in passively over time as journalists found the page and cited it as a source in their reporting.
    The results were published in December 2025 on Founder Reports, and that page has already attracted links from 100+ different websites .
    Read on October 20, 2026.

Sources last checked on October 20, 2026. First published on October 20, 2026.

Awiser is the builder network where founders show their skills, find collaborators, and grow their projects with an AI copilot.