claude research
Claude research for teams who need to hear how B2B purchases actually stall
Most B2B narratives describe a buyer who is tidy, linear, and already half convinced. Claude research stays qualitative: interviews with people who can approve, delay, or kill a purchase, observation of the touchpoints they use, and a written record of the patterns and friction that repeat. The output is a positioning synthesis in the buyer’s language. It is not a revenue forecast and not a representative sample.
4
Moves in the sequence
Interview, observe, name friction, then synthesize. A design of the method, not a market measurement.
1
Memo per cycle
One positioning note in language captured from buyers, with the limits of the evidence on the first page.
0
Representative claims
No confidence interval, no segment percentage, and no promised return on the work.
Why qualitative marketing research kept getting replaced by a chart
Serious B2B teams once learned buying from ride-alongs, loss reviews, and long conversations. Win-loss interviews asked what the other vendor did in the last mile. Voice-of-customer work, at its best, kept the words customers already used. Those programs were later asked to look like a board chart. Open answers became bars. A journey map became a poster. The hesitation that delayed a signature disappeared under a stage called consideration.
Surveys still help when the question is narrow and the list is real: which of three contract terms legal rejected last quarter, among accounts you can name. They go thin when the purchase is political. A security reviewer and a champion do not share one attitude a five-point scale can average.
Claude research returns to the earlier instruments without the costume. You run Claude customer interviews about one past decision. You add market observation of traces: the security form, the forwarded thread, the meeting that keeps moving. You look for purchase friction across roles, cluster it with pattern finding, and stop while the claim is still qualitative.
The last page is a positioning synthesis: who the offer is for, what it replaces, and which phrases to retire. Terms live in the glossary.
Three entry points
Methods that answer different questions
Interviews reconstruct a decision. Friction names where it slowed. Synthesis decides what the offer is allowed to say.
01
Customer interviews
A structured hour on the last real purchase: who acted, which alternative was good enough, and which sentence was used inside the account.
02
Purchase friction
Repeated stalls: security review, internal resale, procurement path, and the pilot that never left one team. Not a complaint list.
03
Positioning synthesis
A short memo that cuts. It keeps buyer language, drops workshop slogans, and states what the notes cannot prove.
Working sequence
A modern pass that still fits in a notebook
Recruit people who can refuse the purchase, stay on one past decision, and watch the touchpoints you already own before you code.
Observe the trace
What was forwarded, delayed, or rewritten before anyone asked a survey question.
Separate pattern from story
Repetition across roles is a pattern. A vivid quote alone is an anecdote.
Keep roles apart
A user, a champion, and a budget holder are not three samples of the same opinion.
Write the cut
End with a memo a skeptic can read without hunting for the hidden ROI slide.
Knowledge hub
Read deeper before you rewrite the offer
Decision-makers
How buyers, champions, reviewers, and blockers each distort the story, and what they can still tell you.
Glossary
Definitions for claude research, friction, pattern, touchpoint, and the claims this guidebook will not make.
Touchpoint mapping
A contact map built from artifacts that exist, not from a diagram of stages nobody recognized.
Pattern finding
How to cluster quotes, when to stop, and how to say “we heard this again” without calling it proof of a market.