TL;DR: Here are 20 AI prompts for customer journey mapping, organized stage by stage from awareness through renewal, each built around emotion and friction columns you can assemble into one consultant-ready deliverable. Every prompt assumes real inputs, such as interviews, tickets, or analytics; none of them invent how a customer probably feels at a stage no one asked them about.
What do these journey-mapping prompts assume you already have?
Customer journey prompts are only as good as the map they're built on: a stage-by-stage picture of what a customer experiences, feels, and struggles with while moving through a relationship with your product, from first hearing about it to renewing or leaving. The 20 prompts below assume you already have something to map: interview notes, support tickets, analytics, sales call recordings, or survey verbatims. None of them are designed to invent a journey from a topic alone. A journey map built without evidence attached to each stage isn't a lighter version of a real one; it's a set of assumptions with a professional-looking template wrapped around them.
This guide picks up where our companion post on gathering real interview data leaves off. That guide covers running interviews properly, including the consent rules for what you can paste into a model. This one assumes that data, or an equivalent from tickets and analytics, already exists, and turns it into a stage-by-stage deliverable.
The structure below follows six stages most journeys share: awareness, consideration, onboarding, usage, support, and renewal or advocacy. Not every product needs all six as separate stages. A low-touch consumer product might collapse onboarding into usage, while a long B2B sales cycle might need consideration split in two. Adjust the stage list to match a real transition your customers make, not a template built for someone else's product.
How do you map the awareness stage with AI prompts?
Awareness covers the moment a prospective customer first learns you exist and starts forming an impression, mostly outside anything you directly observe. These three prompts turn scattered acquisition data (ad performance, referral sources, first-touch survey answers) into a structured first stage.
1. Awareness stage draft from acquisition data
Use when you have channel and first-touch data and need it turned into a stage description.
Role: Journey mapping analyst drafting the awareness stage from acquisition
data, not from assumptions about how people "discover" products like yours.
Context
- Acquisition data: [paste channel breakdown, referral sources, ad data]
- First-touch survey answers: [paste, if you have them, e.g. "how did you
hear about us"]
- Product category: [what you sell, for context only]
Task
Draft the awareness stage of the journey map.
Rules
- Every touchpoint listed must trace to a row in the data provided. Do not
add a channel you assume matters if it isn't in the data.
- Where first-touch survey answers exist, use the customer's own phrasing
for how they describe finding you, not a category you invent.
- Flag any touchpoint with very little data behind it as low-confidence,
rather than describing it with the same certainty as a well-evidenced one.
- Do not fill in an emotion column here. Awareness data rarely includes
emotion, and prompt 2 below covers where that comes from.
Example of the standard I want
Weak: "Customers discover us through word of mouth and social media."
Strong: "38% of surveyed signups cited a specific colleague referral
(verbatim: 'a coworker showed me'); paid social drove 22% of
traffic but only 4% of surveyed signups, a gap worth a second
look at attribution."
Output
Touchpoints with their data source · confidence level per touchpoint ·
the attribution gap, if the data shows one.
2. Awareness-stage emotion from verbatims
Use when you have open-text survey answers or reviews mentioning the discovery moment.
Role: Analyst extracting emotion at the awareness stage from customers'
own words, not from what a marketer would guess they felt.
Context
- Verbatims: [paste open-text survey answers, review excerpts, or
first-touch quotes, each labelled with its source]
- Awareness touchpoints: [from prompt 1, or paste them]
Task
Extract emotion language tied to each touchpoint.
Rules
- Quote the verbatim exactly, with its source label, for every emotion you
report. An emotion with no attached quote doesn't go on the map.
- Distinguish emotion about your brand specifically from emotion about the
category in general ("I'm always skeptical of new tools" is not about you).
- Note where verbatims contradict each other, and report both sides rather
than picking the more flattering one.
- If a touchpoint has no verbatims at all, mark its emotion cell "no data"
rather than inferring one.
Example of the standard I want
Weak: "Customers feel curious and excited when they discover us."
Strong: "'Honestly I almost closed the tab, it looked like every other one
of these' (review, 3-star). Skepticism, not excitement, is the
more common first reaction in this sample."
Output
Emotion per touchpoint with source quotes · contradictions, both sides
shown · touchpoints marked "no data".
3. Awareness-stage competitive context
Use when you need to know what else a prospect is seeing at this stage, without inventing competitor claims.
Role: Analyst noting the competitive context of the awareness stage,
careful not to assert anything about a competitor you can't verify.
Context
- Where prospects say they compared you: [paste verbatims or sales notes
mentioning competitor names, if any]
- What you actually know about competitors: [only verified facts, or "none
verified"]
Task
Describe the competitive context at this stage.
Rules
- Only report a competitor comparison if a customer's own words mention
it. Do not add competitors you assume are relevant.
- Never state a competitor's price, rating, or feature as fact unless the
source explicitly gave it to you as verified. If unverified, say so.
- Note what the customer was using or considering as an alternative,
labelled by source, not as a general market landscape.
- Flag if there's simply not enough data to say anything here.
Example of the standard I want
Weak: "Prospects also consider Competitor X, which is cheaper."
Strong: "2 of 14 sales notes mention a competitor by name; neither notes a
price comparison. Not enough data to characterize competitive
positioning at this stage."
Output
Competitive mentions with sources · what's unverified and flagged as such
· whether the data supports a competitive-context claim at all.
How do you map consideration and evaluation?
Consideration is where a prospect actively compares options and decides whether to try yours, often leaving a trail in sales calls, demo requests, and pricing-page behavior. These four prompts turn that trail into a stage with real friction points attached.
4. Consideration friction from sales call notes
Use when you have call notes or demo recordings and need the recurring objections pulled out.
Role: Analyst extracting friction points from consideration-stage
conversations, distinguishing a recurring objection from a one-off comment.
Context
- Call notes or transcripts: [paste, labelled by call]
- Number of calls represented: [so frequency claims are honest]
Task
Extract friction points from this stage.
Rules
- Count how many calls mention each friction point, and report the count,
not just "several prospects mentioned."
- Quote the objection in the prospect's own words at least once per
friction point.
- Separate a pricing objection from a feature-gap objection from a trust
objection. They need different fixes and shouldn't be merged.
- Do not add a friction point you'd expect to exist if it isn't actually
in the notes.
Example of the standard I want
Weak: "Prospects have concerns about pricing."
Strong: "5 of 12 calls raised pricing specifically about the per-seat
model ('we don't want to pay per person, we share logins'), a
policy objection, not a price-level one."
Output
Friction points with call counts and quotes · objection type per friction
point.
5. Pricing-page and comparison-page behavior read
Use when you have analytics on the pricing or comparison pages and need it turned into a stage insight.
Role: Analyst reading pricing-page and comparison-page analytics for
consideration-stage behavior, careful not to overinterpret a metric.
Context
- Analytics: [paste pageviews, time on page, scroll depth, exit rate,
whatever you have]
- What changed recently on these pages, if anything: [context]
Task
Interpret this behavior for the journey map.
Rules
- State what the metric literally shows before interpreting it. A high
exit rate on the pricing page could mean confusion or could mean a fast,
confident "no". Say which the data can and can't distinguish.
- Do not claim a specific emotion from analytics alone. Analytics shows
behavior; pair it with verbatims before writing an emotion cell.
- Flag any metric that moved recently and note the change alongside it,
since journey maps get read as static and this context matters.
- Do not compare to industry benchmarks unless I gave them to you.
Example of the standard I want
Weak: "Users seem confused on the pricing page."
Strong: "62% scroll-depth drop right after the comparison table; behavior
alone doesn't say why. Pair with call-note objections before
writing this as 'table causes confusion.'"
Output
Literal behavior read · what it can and can't tell you alone · where to
pair it with other evidence before drawing a conclusion.
6. Decision-criteria extraction
Use when you need to know what prospects are actually weighing before they decide.
Role: Analyst extracting the real decision criteria prospects use, from
what they said, not from your own feature roadmap.
Context
- Call notes and won/lost reasons: [paste, labelled]
- Deals represented: [how many, won vs lost split]
Task
Extract the decision criteria this stage actually turns on.
Rules
- Rank criteria by how often they appear in the notes, not by how
important your team assumes they are.
- Separate criteria that appeared in won deals from those in lost deals. A
criterion that shows up in both isn't decisive on its own.
- Quote at least one instance per criterion.
- If the notes don't support ranking with confidence, say the sample is
too small rather than forcing a ranked list.
Example of the standard I want
Weak: "Prospects care most about ease of use."
Strong: "Integration with existing tools appears in 8 of 10 lost-deal
notes and only 2 of 10 won-deal notes, the strongest signal in
this sample, worth checking against a larger set before acting."
Output
Ranked criteria with counts and quotes · won vs lost split · a caution if
the sample is too thin to rank confidently.
7. Consideration-stage summary for the map
Use when you have the pieces from prompts 4-6 and need them assembled into one stage entry.
Role: Journey map analyst assembling the consideration stage from
already-extracted friction points, behavior reads, and decision criteria.
Context
- Friction points: [from prompt 4]
- Behavior reads: [from prompt 5]
- Decision criteria: [from prompt 6]
Task
Assemble the consideration stage entry for the map.
Rules
- Structure as: touchpoints, friction (with source and count), emotion
(only where a verbatim supports it), and the decision criteria this
stage turns on.
- Keep source labels visible in every field. A stakeholder should be able
to ask "where did this come from" about any cell.
- Note any field with genuinely thin evidence, rather than smoothing it to
match the confidence of the well-evidenced fields next to it.
- Do not add anything not already established in the three inputs.
Example of the standard I want
Weak: [a tidy stage summary with no visible sourcing]
Strong: [a stage entry where every friction and emotion cell carries a
(source, n=X) tag]
Output
The assembled consideration-stage entry, ready to slot into the full map.
How do you map onboarding and first use?
Onboarding is where a new customer either reaches a first real result or churns before ever seeing one, and it usually leaves the richest behavioral trail of any stage: activation events, support tickets, and drop-off points.
8. Onboarding drop-off analysis
Use when you have step-by-step funnel data and need the drop-off points turned into friction entries.
Role: Analyst reading an onboarding funnel for drop-off, careful to
distinguish where people leave from why.
Context
- Funnel data: [paste step-by-step completion and drop-off numbers]
- Any known context: [a recent change, a known confusing step]
Task
Identify the friction points in this funnel.
Rules
- Report the step with the largest percentage drop first, with the actual
numbers, not just "many users drop off here."
- State plainly that drop-off location isn't the same as drop-off cause,
and that cause needs a different evidence source (support tickets,
session recordings, or interviews).
- Note any step where a recent change coincides with a change in drop-off
rate.
- Do not guess at a UX cause for a drop-off unless a source below
supports it.
Example of the standard I want
Weak: "A lot of users drop off during setup."
Strong: "42% completion-to-completion drop between step 3 (connect
account) and step 4 (first import), the largest single-step loss
in the funnel. Cause needs ticket or session data."
Output
Ranked drop-off points with numbers · a note on cause needing separate
evidence · anything coinciding with a recent change.
9. Support ticket theme extraction for onboarding
Use when you have a batch of onboarding-stage support tickets and need the recurring issues out.
Role: Analyst extracting themes from onboarding support tickets, the way
you'd extract themes from interview transcripts.
Context
- Tickets: [paste subject lines and bodies, or a summarized export]
- Time period represented: [so frequency is honest]
- Total onboarding tickets in this period: [for a denominator]
Task
Extract themes from these tickets.
Rules
- Report each theme's ticket count against the total, not just a raw count
with no denominator.
- Quote representative ticket language for each theme.
- Distinguish a "how do I" theme (a comprehension gap, possibly a docs
fix) from a "this doesn't work" theme (a bug or design flaw). They need
different owners.
- Do not merge two different underlying issues because their ticket text
looks similar on the surface.
Example of the standard I want
Weak: "Many tickets are about the setup process."
Strong: "31 of 140 onboarding tickets this month (22%) are 'how do I
connect X', a comprehension gap, likely a docs or in-product copy
fix, not a bug."
Output
Themes with ticket counts against total · representative quotes ·
comprehension-gap vs functional-issue split.
10. First-value moment definition
Use when you need to pin down exactly what "aha" looks like for your product, from data rather than opinion.
Role: Analyst defining the first-value moment from behavioral data,
distinguishing a real value event from an easy-to-measure proxy for one.
Context
- Candidate events: [features or actions your team suspects matter]
- Retention or activation data tied to each candidate: [paste correlations
or cohort comparisons, if you have them]
- Qualitative signal: [interview or ticket mentions of "that's when it
clicked", if you have any]
Task
Assess which candidate is the real first-value moment.
Rules
- Weigh quantitative correlation with retention alongside qualitative
mentions of the moment clicking, since a metric can correlate with
retention without being the cause customers actually experience.
- Flag any candidate event that's easy to log but weakly tied to either
kind of evidence, and say so plainly rather than defaulting to it
because it's convenient to track.
- State your confidence level in the conclusion given the data provided.
- Do not declare a first-value moment with only one type of evidence if
the other type contradicts it.
Example of the standard I want
Weak: "Signup is the aha moment."
Strong: "Completing a second import correlates most strongly with 90-day
retention and matches 4 of 6 interview mentions of 'that's when I
got it', moderate confidence, worth confirming with more
interviews."
Output
Assessed first-value candidate with confidence level · evidence for and
against · what more would raise confidence.
11. Onboarding stage assembly
Use when you have drop-off data, ticket themes, and a first-value definition, and need them combined into the stage entry.
Role: Journey map analyst assembling the onboarding stage from
already-extracted drop-off, ticket themes, and the first-value definition.
Context
- Drop-off points: [from prompt 8]
- Ticket themes: [from prompt 9]
- First-value moment: [from prompt 10]
Task
Assemble the onboarding stage entry for the map.
Rules
- Structure as: touchpoints/steps, friction (with source and n), emotion
(from ticket language, quoted), and the first-value moment with its
confidence level.
- Keep the comprehension-gap vs functional-issue distinction from prompt 9
visible, since they route to different owners in a real deliverable.
- Note any step with no evidence attached as a gap to fill, not as a blank
to leave unexplained.
- Do not add anything not already established in the three inputs.
Example of the standard I want
Weak: [a tidy stage summary with no visible sourcing]
Strong: [a stage entry where the first-value moment carries its confidence
level and evidence type]
Output
The assembled onboarding-stage entry, ready to slot into the full map.
How do you map ongoing usage and support?
Usage and support cover the longest stretch of the relationship and the one most teams map thinnest, because the data is scattered across product analytics, support tools, and NPS or CSAT surveys that rarely get read together.
12. Feature usage pattern extraction
Use when you have product analytics and need the real usage pattern, not the one your roadmap assumes.
Role: Analyst reading product usage data for the journey map, reporting
what's actually used rather than what the product was designed around.
Context
- Usage data: [paste feature adoption, frequency, or session data]
- Segment, if relevant: [plan tier, tenure, or use case]
Task
Extract the usage pattern for this stage.
Rules
- Report adoption and frequency with real numbers, not "most users" or
"frequently."
- Note any feature with very low adoption relative to its prominence in
onboarding or marketing. That gap is itself a finding.
- If usage differs meaningfully by segment, report the segments
separately rather than one blended average that describes no one.
- Do not speculate on why a pattern exists here; that's the job of prompts
13 and 14 with different evidence.
Example of the standard I want
Weak: "Users mostly use the core features."
Strong: "Feature C, prominent in onboarding, is used by only 9% of active
accounts after 90 days, a gap worth investigating with support or
interview data."
Output
Usage pattern with real numbers · segment splits if relevant · gaps
between prominence and adoption, flagged.
13. Ongoing friction from support ticket trends
Use when you need to know what breaks the experience after onboarding, not during it.
Role: Analyst extracting post-onboarding friction from support ticket
trends over time.
Context
- Tickets by tenure or account age: [paste, or describe how you can
segment them]
- Time period: [for frequency context]
Task
Extract friction points specific to the usage stage.
Rules
- Separate tickets from newer accounts (still onboarding) from ones with
tenure, so onboarding issues don't contaminate the usage-stage read.
- Report trend direction if you have more than one period's data (rising,
falling, flat), not just a snapshot count.
- Quote representative ticket language per friction point.
- Flag any friction point that correlates with churn risk if you have
churn data to check it against.
Example of the standard I want
Weak: "Support tickets show ongoing issues."
Strong: "Tickets about report export among accounts 6+ months old rose
from 4% to 11% of ticket volume over two quarters, a rising
trend worth flagging to product."
Output
Friction points with tenure segmentation · trend direction · churn
correlation if checkable.
14. NPS or CSAT verbatim theming by stage
Use when you have open-text survey responses and need them tied to a specific point in the usage stage.
Role: Analyst theming satisfaction survey verbatims and tying them to
where in the usage stage they were collected.
Context
- Verbatims with scores: [paste, with the survey trigger point noted, e.g.
"sent after 30 days", "sent after a support ticket closed"]
Task
Theme the verbatims by what's driving the score.
Rules
- Report themes separately for high scores and low scores. Don't average
sentiment into one blended theme list.
- Quote verbatims for every theme.
- Note the survey's trigger point explicitly, since a score collected
right after a support interaction measures something different from one
collected on a fixed schedule.
- Flag any theme appearing in both high- and low-score responses, since
that usually means the same feature affects different customers
oppositely.
Example of the standard I want
Weak: "Overall sentiment is moderately positive."
Strong: "Among 8-10 scores: 'reliable', 'does what it says' (n=14). Among
0-6 scores: 'reliable but slow to load' (n=6), same feature,
different threshold for acceptable."
Output
Themes split by score band, with quotes · the trigger-point context · any
theme cutting across both bands.
15. Support interaction emotional temperature check
Use when you want to know how customers feel during a support interaction itself, not just whether it resolved.
Role: Analyst reading support transcript tone across a resolution,
distinguishing the emotion at contact from the emotion at close.
Context
- Support transcripts: [paste, labelled by ticket]
- Resolution outcome per ticket: [resolved, escalated, abandoned]
Task
Characterize the emotional arc of these interactions.
Rules
- Report emotion at the start of contact and at close separately, since a
ticket can start frustrated and end satisfied, or the reverse.
- Quote the transcript for both points per ticket.
- Note any ticket where the emotional arc doesn't match the stated
resolution outcome (marked "resolved" but tone at close still reads
frustrated).
- Do not average emotion across tickets into a single score. Report the
range and the mismatches.
Example of the standard I want
Weak: "Support interactions are generally positive."
Strong: "6 of 9 tickets marked 'resolved' show tone still frustrated at
close ('fine I guess, took long enough'). Resolution status
alone overstates satisfaction here."
Output
Start-vs-close emotion per ticket · mismatches between tone and stated
resolution · range across the sample.
16. Usage and support stage assembly
Use when you have the pieces from prompts 12-15 and need them combined into one stage entry.
Role: Journey map analyst assembling the usage and support stage from
already-extracted usage patterns, friction trends, survey themes, and
support tone.
Context
- Usage pattern: [from prompt 12]
- Friction trends: [from prompt 13]
- Survey themes: [from prompt 14]
- Support emotional arcs: [from prompt 15]
Task
Assemble the usage and support stage entry for the map.
Rules
- Structure as: touchpoints, friction (with source, n, and trend
direction), emotion (split by context, general usage vs mid-support),
and the adoption gaps from prompt 12.
- Keep the resolved-but-frustrated mismatches from prompt 15 visible; they
are usually the most actionable single finding in this stage.
- Note thin-evidence fields rather than smoothing them.
- Do not add anything not already established in the four inputs.
Example of the standard I want
Weak: [a tidy stage summary with no visible sourcing]
Strong: [a stage entry that separately shows general-usage emotion and
support-interaction emotion, since they're often different]
Output
The assembled usage-and-support stage entry, ready to slot into the full
map.
How do you map renewal, churn risk, and advocacy?
The final stage covers the decision point that determines whether the relationship continues, expands, or ends, plus the advocacy behavior (referrals, reviews, case studies) a strong version of this stage produces.
17. Churn-risk signal extraction
Use when you have account data and need the real leading indicators, not assumed ones.
Role: Analyst identifying churn-risk signals from account data, checking
correlation rather than assuming which signals matter.
Context
- Account data with outcomes: [paste usage, support, and billing history
for accounts that churned and accounts that renewed]
- Time window before outcome: [how far back you're looking]
Task
Identify signals that precede churn.
Rules
- Compare churned accounts against renewed accounts on the same metrics. A
pattern only in churned accounts, with no comparison group, isn't a
signal yet.
- Report how far in advance each signal appears, since a signal that only
shows up a week before churn is less actionable than one visible at 60
days.
- Do not assume declining usage is always the leading signal; check
support ticket tone and billing friction too.
- State sample size and flag if it's too small to generalize.
Example of the standard I want
Weak: "Churned accounts used the product less."
Strong: "Accounts that churned showed a 40%+ usage drop 45+ days before
cancellation in 11 of 15 cases; renewed accounts in the
comparison set showed this pattern in only 2 of 15, visible early
enough to act on."
Output
Signals with lead time and churned-vs-renewed comparison · sample size and
confidence.
18. Renewal conversation friction from account team notes
Use when you have CSM or account manager notes from renewal conversations.
Role: Analyst extracting friction from renewal-stage conversations,
separating a price objection from a value question.
Context
- Renewal call or email notes: [paste, labelled by account]
- Renewal outcome per account: [renewed, downgraded, churned]
Task
Extract friction points from these conversations.
Rules
- Separate "we can't justify the price" from "we're not sure we're still
getting value". The first is a pricing conversation, the second is a
product or success problem, and they need different owners.
- Quote the account's own framing for each friction point.
- Note any friction point appearing disproportionately among downgraded or
churned accounts versus renewed ones.
- Do not merge accounts with very different outcomes into one blended
read.
Example of the standard I want
Weak: "Some accounts had renewal concerns."
Strong: "4 of 7 downgraded accounts specifically cited 'we're using less
of it than we expected to', a value-realization gap, not a price
objection, in this subset."
Output
Friction points with account outcomes attached · pricing vs value-gap
split.
19. Advocacy behavior and case-study candidate extraction
Use when you want to identify where genuine advocacy is happening, not assume it exists everywhere.
Role: Analyst identifying real advocacy signals — referrals, public
reviews, expansion requests — rather than assuming satisfaction implies
advocacy.
Context
- Referral data: [paste, if tracked]
- Public reviews or mentions: [paste, with source]
- Expansion or upsell requests initiated by the customer: [paste, if
tracked]
Task
Extract advocacy signals for this stage.
Rules
- Only count a behavior as advocacy if it's an action (referred someone,
posted a review, asked to expand), not a sentiment you're inferring from
a high satisfaction score.
- Quote the review or referral context where available.
- Note if advocacy behavior clusters in a particular segment or use case,
since that's useful for case-study selection.
- Flag if the data provided is too thin to say anything about advocacy at
all.
Example of the standard I want
Weak: "Happy customers likely refer others."
Strong: "3 tracked referrals this quarter, all from accounts on the Team
plan with 6+ months tenure, a thin but specific pattern worth
watching, not yet a claim about who typically refers."
Output
Advocacy signals with sources · segment clustering, if any · a note if the
data is too thin to generalize.
20. Renewal and advocacy stage assembly
Use when you have the pieces from prompts 17-19 and need the final stage of the map.
Role: Journey map analyst assembling the renewal and advocacy stage from
churn signals, renewal friction, and advocacy behavior.
Context
- Churn-risk signals: [from prompt 17]
- Renewal friction: [from prompt 18]
- Advocacy signals: [from prompt 19]
Task
Assemble the renewal and advocacy stage entry for the map.
Rules
- Structure as: touchpoints, friction (pricing vs value-gap, sourced),
emotion (from account notes, quoted), and churn signals with lead time.
- Keep advocacy as its own row rather than folding it into "positive
emotion", since it's a distinct behavior with its own evidence.
- Note thin-evidence fields rather than smoothing them.
- Do not add anything not already established in the three inputs.
Example of the standard I want
Weak: [a tidy stage summary with no visible sourcing]
Strong: [a stage entry where churn signals carry their lead time and
advocacy sits in its own row with its own sources]
Output
The assembled renewal-and-advocacy stage entry, completing the map.
How do you assemble one journey map deliverable from these prompts?
A consultant-friendly journey map is a single table a client can read in one sitting, not six separate stage documents. Once you have all six stage assemblies (prompts 7, 11, 16, and 20, plus the awareness and consideration pieces), the final step is combining them into one table with consistent columns.
| Stage | Key touchpoint | Customer emotion | Friction (source) | Opportunity |
|---|---|---|---|---|
| Awareness | Colleague referral, paid social | Skeptical, "looked like every other tool" (review) | Attribution gap between paid spend and surveyed signups (n=118) | Test whether paid social should shift budget toward retargeting past referral clickers |
| Consideration | Comparison page, sales call | Cautious, comparing carefully | Per-seat pricing model objection (5/12 calls) | Clarify seat-sharing policy earlier in the sales conversation |
| Onboarding | Account connect, first import | Uncertain, "how do I connect X" (22% of tickets) | Largest single-step drop-off at import (42%) | Add inline guidance at the import step specifically |
| Usage | Core reporting feature | Mixed: reliable but "slow to load" at scale | Rising export-related tickets among 6+ month accounts | Investigate export performance for longer-tenured, higher-volume accounts |
| Support | Ticket resolution | Resolution status overstates satisfaction (6/9 "resolved" tickets still read frustrated) | Gap between ticket status and actual tone at close | Add a tone-aware follow-up check, not just a status field |
| Renewal | CSM renewal call | Value-realization doubt in downgraded accounts (4/7) | Distinct from a pricing objection | Build a lightweight value-recap step before renewal conversations start |
Every cell in a table like this should carry a source, exactly as shown in the friction column above. A client-facing version can move the source into a footnote or an appendix rather than the visible cell, but it should never disappear entirely. That is what separates a consultant deliverable a client can act on from a template filled in from best guesses.
What should back a journey map, and what shouldn't?
Every stage above names its evidence type on purpose: analytics for behavior, tickets and transcripts for friction, verbatims and interviews for emotion. A journey map assembled entirely in a facilitation workshop, sticky notes and a room of stakeholders guessing how customers feel, produces a map that's genuinely useful for building shared understanding inside a team, and genuinely risky if it then gets presented to a client or an executive as research. The two documents can look identical on the page. Only one of them can answer "how do you know" for every cell.
The honest fix isn't refusing to build a map without a full research budget. It's labeling confidence level per stage or per cell. A stage built from six months of ticket data deserves a different confidence label than one built from three internal guesses in a workshop. Say which is which, and update the labels as better evidence arrives.
How Prompt Architects fits this workflow
The 20 prompts above work in whatever AI tool you already use for this kind of analysis. What Prompt Architects adds is a place to keep the version that's actually reviewed: a prompt library where the six stage-assembly prompts, your emotion-sourcing rule, and your confidence-labeling convention live together, tagged to the client or product they're for, instead of being rebuilt in a chat window every quarter.
The Contexts feature stores your standing rules, such as which evidence sources count as "confirmed" versus "hypothesis", so every stage prompt inherits the same standard automatically. For consultants running this across several clients at once, our best AI prompt tools for consultants and analysts roundup and our agency client deliverables guide cover keeping that consistency across an entire practice, not just one map.
Prompt Architects is free to start, no credit card required. If the evidence behind your map is thin on the interview side specifically, our companion guide to 30 AI prompts for user research and interviews covers gathering that data properly, consent rules included, before you fill in another emotion column from a guess.
Stop rewriting prompts. Start shipping.
Works with ChatGPT, Claude, Gemini, Grok, Midjourney, Ideogram, Veo3 & Kling. 4.8★ on the Chrome Web Store.
Create An Account