3 builders who never spoke to each other

Publish the method and the holes in it

Research travels on trust, so state the choices behind it: which domains were picked and how, what was excluded, what the data cannot see, and where a label is your interpretation rather than a verified filter. One study named its desktop only limitation and the exclusions outright, and another flagged that eye-catching over-index numbers on tiny baselines were directional at best. Releasing the framework itself, so peers can run the same study, is what makes you the reference point rather than a one off.

The method

1. 01

   Publish the choices and the exclusions behind your research, including what you left out and why.

2. 02

   Say plainly when your audience definition is an affinity interpretation rather than a verified filter, and flag the numbers that are artefacts.

3. 03

   Get your own numbers from a clickstream provider when the platform will not answer, and publish the gaps in the method alongside them.

4. 04

   Release the method behind your customer research, not just the conclusions, so peers in your industry can run the same study and you become the reference point for it.

What it returned

• 41

  Rand Fishkin states that he picked the 41 domains editorially from the most visited 250 on Datos' panel, that the study is desktop only, that adult sites and irrelevant domains such as postal carriers were dropped, and that a multi-turn AI conversation was counted as one search just as a Google query with dozens of clicks was.

• 72

  SparkToro notes about 72 percent of this cohort sat in software, IT or internet companies with 48 percent in software specifically, but with no company-size cut the phrase at startups really meant indexing toward startup interests, and the eye-catching over-index numbers on tiny-baseline tools are directional only.

• 2018

  After Sundar Pichai's December 2018 answers to Congress, Rand Fishkin asked contacts at Jumpshot for figures drawn from over 10 million US devices, and stated plainly that the data excludes iOS, in-app searches and voice devices, so the real share going back to Google is higher than 12 percent.

• Hillman said the plan was to keep adding interview sources and then publish the research framework so other coworking spaces could repeat the study.

Sources

• Why do people love Indy Hall? We asked. They told.Alex Hillman · dangerouslyawesome.com · 2011-07-01
• How Much of Google's Search Traffic is Left for Anyone But Themselves? - SparkToroRand Fishkin · sparktoro.com · 2019-06-18
• How We'd Market to Software Developers at Startups - SparkToroSparkToro · sparktoro.com · A transparent note on the data · 2026-06-04
• New Research: Search Happens Everywhere; an Analysis of 41 Websites with Significant Search Activity - SparkToroRand Fishkin · sparktoro.com · Important Caveats & Methodology Notes · undated