E-commerce
July 1, 2026
"The bot sent me a block of text, I didn't read any of it." "Too much text on mobile, I closed the chat." "I just wanted to know if I can do a return, not get a full lecture." Three tickets where the length of chatbot responses degrades the experience.
The support for excessively long chatbot responses in e-commerce covers perceived verbosity, drowned key information, mobile abandonment, and requests for summaries. Distinct from bot misunderstanding (#879): here the customer has understood that the response was too long, not that it was off-topic.
This guide #887 deploys policy CHATLONG-SUP, flow CL-1 to CL-8, and matrix CHATLONG-MAP. Feeds into the future short responses guide (#888).
Summary
Why does chatbot verbosity generate tickets?
Verbose prompt, RAG that injects the entire help center, lack of token display limit: the bot drowns out the useful info. The agent redirects to the bot or copies and pastes the block of text instead of summarizing.
Five typical frictions of long responses
Wall of text: dense paragraph without structure
Unreadable on mobile: excessive scrolling on small screens
Buried info: yes/no answer drowned in the middle
Customer re-contact: reformulates because they didn't read
Widget abandonment: closes chat after the block of text
DTC retail example
DTC fashion, 8k chat tickets/month. After CHATLONG-MAP: chatlong_brief_resolution_rate 88 %, verbosity handoffs -35 %.
CHATLONG #887 vs CHATMIS #879, CHATRT #877, HANDOFF #12 and bot #888
Five UX bot contents, five distinct angles.
Quick matrix
#887 CHATLONG: manage tickets where the bot response is too long and unreadable on mobile
CHATMIS #879: bot did not understand separate from length
CHATRT #877: bot response delay separate from verbosity
CHATINT #881: interrupted conversation separate from block of text
HANDOFF #12: transfer rules after bot failure
#887 = too much text. #879 = wrong topic.
Promise #887
Policy CHATLONG-SUP, tree CHATLONG-GATE, 8 macros, summarize and resolve in brief, chatlong_brief_resolution_rate KPI.
Which chatlong_* typologies should be classified?
Action-oriented classifier: too_verbose ≠ mobile_pain ≠ wants_summary.
Eight CHATLONG-MAP typologies
chatlong_too_verbose: general complaint about wall of text bot
chatlong_mobile_pain: unreadable on smartphone
chatlong_lost_key_info: yes/no answer not found
chatlong_wants_summary: requests short version
chatlong_repeat_question: asked again because not read
chatlong_unread: admits to not reading
chatlong_handoff_request: wants a human after wall of text
chatlong_frustration: anger towards verbose bot tone
Policy CHATLONG-SUP: agent rules and escalation
The CHATLONG-SUP policy sets a short agent response even if the bot has been long.
Six CHATLONG-SUP rules
ACKNOWLEDGE-FIRST: CHATLONG-ACKNOWLEDGE macro validates length frustration
Summarize in 3 lines: CHATLONG-KEY-POINTS before any block of text
Resolve briefly: CHATLONG-RESOLVE-BRIEF direct customer service response
Do not copy the bot: do not paste the full agent transcript
Log UX: CHATLONG-LOG-UX feeds #888
Handoff if needed: CHATLONG-HANDOFF if the customer refuses the summary
Situation matrix (agent)
Simple question: KEY-POINTS + RESOLVE-BRIEF one sentence
Buried info: SUMMARIZE extract yes/no at the top
Mobile: KEY-POINTS short bullet points max 3
Misunderstood topic: handoff CHATMIS #879, not CHATLONG alone
Flow CL-1 to CL-8: standard resolution
Eight sequential steps, SLA P3 chatlong < 24 h, escalate to UX if too_verbose is recurring for the same intent.
Flow CL-1 to CL-8
CL-1 Triage: read complaint, tag chatlong_*, length or lack of understanding #879?
CL-2 Lookup: bot transcript, character length, initial question
CL-3 Extract: actual customer need, one underlying question
CL-4 Classify: chatlong_* via CHATLONG-MAP
CL-5 Execute: KEY-POINTS, RESOLVE-BRIEF, HANDOFF, LOG-UX
CL-6 Confirm: macro CHATLONG-DONE, summary of next step
CL-7 Test: customer has a short, useful answer
CL-8 Close: KPI chatlong_brief_resolution_rate + export #888
Eight CHATLONG-* macros ready to paste
Aligned macros: short summary and bot UX escalation.
CHATLONG-* Library
CHATLONG-ACKNOWLEDGE: "We understand that the chatbot's response was too long. Here is the main point."
CHATLONG-KEY-POINTS: "In short: {{point_1}}. {{point_2}}. {{optional_point_3}}."
CHATLONG-SUMMARIZE: "Direct answer: {{yes_no}}. {{one_sentence_detail}}."
CHATLONG-RESOLVE-BRIEF: "Regarding your {{subject}} request: {{short_solution}}."
CHATLONG-HANDOFF: "An agent is taking over in a short format. Reference: {{id}}."
CHATLONG-LOG-UX: "Verbosity feedback recorded to improve the chatbot. Intent: {{intent}}."
CHATLONG-ALTERNATIVE: "Summary also by email if preferred: {{channel}}."
CHATLONG-DONE: "Recap: {{summary}}. Action: {{resolution}}. UX report: {{yes_no}}."
CHATLONG-GATE tree and agent-ready conciseness register
Decision tree before copying the bot or ignoring the length complaint.
CHATLONG-GATE
Short-answer question? → SUMMARIZE + RESOLVE-BRIEF
Info drowned in a block of text? → KEY-POINTS yes/no in the first line
Customer wants a human? → HANDOFF short format #12
Same verbose intent 5+ times? → LOG-UX + escalate #888
Wrong bot topic? → handoff CHATMIS #879
Minimum Internal Register
Document helpdesk: max agent lines per typology, access to length transcript, LOG UX procedure #888. Train agents: verbosity ≠ lack of understanding #879 ≠ slowness #877.
KPI, QA and handoff to bot #888
Measuring CHATLONG detects uncorrected verbose intents produced.
Four CHATLONG KPIs
chatlong_brief_resolution_rate: customer satisfied with brief summary / total
chatlong_agent_paste_bot_rate: % of agents pasting the bot text block, target is low
chatlong_ux_logged_rate: % of too_verbose with LOG-UX tracked
chatlong_repeat_7d: same length complaint within 7 days
Handoff #888
Export weekly CHATLONG-MAP: chatlong_too_verbose and chatlong_lost_key_info are priorities. Guardrail CHATLONG-BREVITY-LOOP: each LOG-UX feeds max_tokens prompts #888.
Edge cases: complex topic, legal, client requests detail
Three cases outside the standard flow.
Complex multi-step return after-sales policy
KEY-POINTS 3 bullets + help hub link. No incomplete RESOLVE-BRIEF if there is a risk of error.
Customer voluntarily requests details
ALTERNATIVE full article link. No CHATLONG if the customer wants more text.
Bot response is long but correct
ACKNOWLEDGE + SUMMARIZE. LOG-UX anyway if there is a recurring complaint for the same intent.
Agent training: 20 minutes CHATLONG
Module: Systematic KEY-POINTS, never paste bot, LOG-UX, distinguish #879 #877, brief #888.
Exercises
Ticket A: feedback block 400 words → SUMMARIZE yes 30 d + RESOLVE-BRIEF
Ticket B: unreadable mobile → KEY-POINTS 3 short bullet points
Ticket C: long off-topic bot → handoff CHATMIS #879 not CHATLONG alone
How Qstomy structures CHATLONG in your stack
Qstomy route chatlong_*, displays agent transcript length, KEY-POINTS macros and LOG export to #888 brevity.
Three building blocks
Routing: intent response_too_long vs misunderstanding vs sav_issue
Verbosity panel: sync intent characters macros CHATLONG-*
UX loop #888: weekly product aggregation of chatlong_too_verbose
Scenario: retail DTC, verbose return intent. KEY-POINTS agents, #888 max_tokens reduced. chatlong_brief_resolution_rate goes from 71% to 90% in 4 weeks.
FAQ and CHATLONG deployment checklist
FAQ
Can the agent send the full bot text?
No. KEY-POINTS + RESOLVE-BRIEF. Hub link if details wanted.
Difference from #879?
#887 = too long. #879 = poor topic understanding.
Difference from #888?
#887 = agents managing verbosity tickets. #888 = bot writing short on the widget side.
7-day Checklist
D1: CHATLONG-SUP + CHATLONG-MAP + length transcript access
D2: 8 helpdesk macros
D3: routing matrix #879 #877 #12
D4: 20 min agent training
D5: chatlong_* tags + KPIs
D6: SUMMARIZE vs HANDOFF vs CHATMIS test
D7: weekly export to backlog #888
Interlinking
Enzo
July 1, 2026
