Category
220 articles in this category · page 4 of 19
A third of online research participants now use LLMs to answer open-ended questions, and an autonomous AI agent passed 99.8% of standard attention checks in a 2025 PNAS study. Here is what the evidence actually shows, why detection is losing, and what still works.
Koji Team · August 7, 2026 · 13 min read
Mid-market companies are too big for founder-led discovery and too small to staff a research team. Startups get free tiers, enterprises get site licences, and the middle gets sold the wrong one. What to run, who runs it, and how to buy at $10M to $1B in revenue.
Koji Team · August 6, 2026 · 11 min read
Your warehouse can tell you what changed, for whom, and by how much. It cannot tell you why, because motivation was never emitted as an event. A practical playbook for analytics engineers, analysts and data scientists on adding qualitative evidence to a quantitative practice, without hiring researchers.
Koji Team · August 6, 2026 · 11 min read
The most common objection to AI-moderated research is that people will not be honest with a bot. Three decades of survey mode-effect research says the opposite: respondents disclose more to an automated interviewer than to a human one. Here is the evidence, the mechanism, and where the advantage does not apply.
Koji Team · August 6, 2026 · 12 min read
Your community is the fastest customer signal you have and the most selected sample you own — four filters deep before anyone posts. Here is which questions a community can genuinely answer, which it structurally cannot, and how to correct for the difference.
Koji Team · August 5, 2026 · 11 min read
Deflection counts exits, not answers — Gartner finds only 14% of customer service issues are fully resolved in self-service. Support teams hold the highest-volume customer signal in the company and the least ability to act on it. Here is how CX teams run real research without adding headcount.
Koji Team · August 5, 2026 · 11 min read
Being the only researcher is not a staffing gap you wait out. At the modal 1:5 researcher-to-designer ratio and a median output of one study a month, each designer gets about 2.4 studies a year. Here is the operating model that fixes the math.
Koji Team · August 5, 2026 · 11 min read
Every term you will meet in a research plan, a vendor quote or a sample provider invoice — methods, sampling, bias, statistics, metrics, commercial jargon like CPI, LOI and IR, and the new AI research vocabulary. Plain definitions, one line each.
Koji Team · August 4, 2026 · 14 min read
73% of PMs now use AI weekly. 95% of enterprise AI pilots produce no measurable ROI. Here are 16 sourced product management statistics for 2026 — and what the gap between those two numbers means.
Koji Research Team · July 29, 2026 · 9 min read
HCP recruitment is the most expensive corner of market research: single-digit incidence, $3–$8 per minute honoraria, and respondents with no free time. Here is what actually works — and what to verify before you field.
Koji Team · July 27, 2026 · 9 min read
The problem with frequent survey-takers is not that they lie. It is that a tiny fraction of panelists produce a large share of all responses, and answering surveys repeatedly changes the people answering them.
Koji Team · July 27, 2026 · 9 min read
Three sourcing models sit behind almost every survey you will ever buy, and they produce meaningfully different respondents. Here is how each works, what each costs, and when the differences change your conclusions.
Koji Team · July 27, 2026 · 9 min read