~30 contextual inquiry sessions · real seller planning cycle
Outcome
Diagnosis fed a broader cross-functional redesign
When sellers don't rely on embedded AI, it's tempting to ask how to drive adoption. I asked a different question: what would make the capability worth trusting and using?
01 · The problem worth solving
An AI capability embedded in the Quota Attainment Plan and Account Planning experience wasn't being relied on. The reflexive question, "how do we drive adoption?", assumes the capability is fine and people just need pushing.
Low reliance can look like an enablement problem even when the underlying issue is trust, and those diagnoses call for very different responses. I set out to understand which conditions were actually breaking, by observing sellers inside their real planning workflow rather than asking abstract satisfaction questions.
Is this a knowledge problem, or a trust problem, and what's really breaking underneath?
Artifact · What the symptom was hiding
Surface symptom
Sellers weren't relying on the embedded AI
The visible signal, easy to read as an enablement gap.
What the symptom was hiding
01
AI output
Sometimes inaccurate or redundant
Coaching and summaries were seen as less reliable than a conversational assistant sellers already trusted.
02
Workflow
Planning was fragmented and out of flow
The official tools lagged the real planning cycle, so sellers dropped into spreadsheets and manual work.
03
Data
Gaps and reliability issues underneath
Missing visibility on some accounts degraded trust in everything built on top of it.
04
Value
Planning felt like an obligation, not a help
The experience wasn't useful enough in the actual selling motion to earn reliance.
02 · My role
I proactively identified the planning gap and led the contextual inquiry around the real seller planning cycle. I owned the field research, synthesis and diagnosis. The redesign that followed was a broader cross-functional effort; my research was one input into it, not something I owned end to end.
Where I placed the research
Planning activity begins
Observed the real work here
where real flow and formal flow diverged
Formal planning milestone
The insight wasn't only who I talked to; it was when. I placed the research in the window where the real planning motion had already started but the formal process hadn't yet caught up, which is exactly where the workarounds lived.
03 · The insight that reframed it
The issue wasn't simply whether sellers understood the AI. It was whether the experience had earned enough trust to become useful. Trust was breaking at several layers at once, so the question shifted from "how do we drive adoption?" to "what makes adoption rational in the first place?"
Trust in the data has to come before trust in AI built on top of it.
Adoption is an outcome. Trust has prerequisites.
A diagnostic sequence suggested by the research, not a causal model.
Data accuracy
is the underlying data reliable?
× ↓
Workflow fit
does it fit the real motion?
× ↓
Useful intervention
does it actually help?
↓
Trust
earned, not enabled
↓
Adoption
the result, not the input
04 · The decision it enabled
The work moved from "drive adoption" to fixing the prerequisites adoption depends on:
01
Foundational data accuracy
Fix the gaps and reliability issues so nothing built on top inherits broken inputs.
02
Planning indicators sellers can act on
Signals that reflect real risk, not indicators that can mislead the read.
03
Trusted insight inside the workflow
Bring the intelligence to where planning actually happens, rather than as a separate surface.
04
More proactive, in-context assistance
Help that shows up in the moment of work, once the foundations make it worth trusting.
05 · What happened next
Honest attribution
About five months later, a broader cross-functional redesign addressed several of the same frictions, including bringing territory intelligence closer to the daily workflow.
I present this as what followed, not as a personal result. My research was one input into that effort; I owned the diagnosis, not the end-to-end redesign.
06 · How I'd measure value
Success isn't "adoption went up." It's whether the prerequisites got fixed. I'd look for evidence that the underlying data became more reliable, the intervention fit the real workflow, sellers found the insight worth trusting, and, critically, that the same lens shaped how future AI capabilities were evaluated before launch:
Foundation readiness
Whether the data-accuracy failures sellers flagged actually get fixed.
Trust & usefulness
Whether sellers find the AI insight reliable and worth acting on, in follow-up inquiry.
Workflow fit
Whether the intelligence arrives where planning actually happens, not as a separate surface.
Decision adoption
Whether the "trust has prerequisites" lens changes how future AI capabilities get evaluated before launch.