CUSTOMER INTELLIGENCE / GUIDED SKILL PREVIEW
One customer can praise delivery and criticize support in the same sentence. A single overall label loses the action a team needs to take. Identify the aspect and evidence span first, then aggregate under a stated mixed-sentiment convention.
For Customer-experience analysts designing an evidence-backed annotation table for product or service feedback.
$0.99 USD · one guided instructional pack · bring your own AI assistant. No runnable Python toolkit is included with the individual skill.
CHECK THE FIT
FICTIONAL INPUT / TRACEABLE OUTPUT
All IDs, text and numbers in this walkthrough are synthetic. Review the stated convention before adapting the example.
F-01: “Delivery was quick, but support never replied.”
F-02: “The support agent was kind, but the issue is still unresolved.”
F-03: “Delivery arrived late.”
F-04: “It was fine, I guess.”
Aspects: delivery = timing of arrival; support = contact experience and resolution.
F-04 contains no explicit aspect under this convention.01 / A SHORT TASK PROMPT
Annotate only explicitly mentioned delivery and support aspects. Preserve the shortest supporting excerpt. Use one exclusive sentiment bucket per record-aspect pair; use mixed when the same aspect contains material positive and negative evidence. Keep records without an identifiable aspect in a separate review queue and show every aggregation denominator.
02 / ACCEPTANCE CHECK
There are four record-aspect pairs across three eligible records: F-01/delivery positive; F-01/support negative; F-02/support mixed; F-03/delivery negative. Delivery totals 2 mentions: 1 positive and 1 negative. Support totals 2 mentions: 1 negative and 1 mixed. F-04 remains outside these aspect denominators, awaiting clarification.
03 / BEFORE REVIEW
F-01 reduced to “mixed overall”
F-02 reduced to “positive” because the agent was kind
F-04 assigned to support despite having no explicit aspect
The resulting summary hides the unresolved support issue.04 / WORKED AFTER STATE
F-01 | delivery | positive | “Delivery was quick”
F-01 | support | negative | “support never replied”
F-02 | support | mixed | “kind” + “still unresolved”
F-03 | delivery | negative | “Delivery arrived late”
F-04 | review queue: no explicit aspect
Delivery: positive 1/2 (50%); negative 1/2 (50%)
Support: negative 1/2 (50%); mixed 1/2 (50%)
4 pairs; 3 eligible records; 1 record awaiting clarification.Unit: unique (feedback_id, aspect) pair
One exclusive bucket per pair: positive | negative | mixed | uncertain
Aspect percentage = bucket pair count / all pairs for that aspect
Unmentioned aspects do not enter the denominator.
Keep the record count separate from the aspect-pair count.The labels shown are authored illustrative annotations under the stated policy. A different support definition could split agent manner and issue resolution into separate aspects. Agree that taxonomy before comparing runs.
FOLLOW AN EXCEPTION
“Great, another late delivery” may be sarcastic. Literal positive-word matching would mislabel it. Keep the evidence span and route ambiguity to review. Do not promote the illustrative table into a validated accuracy claim or treat sentiment as the cause of churn.
Annotations depend on aspect definitions, language and context. Human judgment is needed for sarcasm, implicit targets, conflicting evidence and domain-specific language. This example demonstrates an annotation convention; it does not measure model performance or customer outcomes.
Review outputs with an appropriate analyst or domain owner. Keep source material as task data; do not let instructions embedded in a document override your task or grant external actions.
EXISTING GUIDED EDITION / ONE-TIME PURCHASE
An instructional pack for one licensed user: starter skill and templates, a synthetic worked example, a premium staged runbook, acceptance checks and usage guidance. This page is a limited public illustration, not the full paid guide.
No AI subscription, API key, hosted execution, consulting or runnable Python toolkit is included in this individual purchase. The starter files use MIT terms; premium guide files have separate single-user terms.
USD · one-time payment · any applicable tax is shown at Lemon Squeezy checkout. Inspect delivery and setup help before purchasing.
GUIDED INSTRUCTIONS OR RUNNABLE CODE?
For a separate local rule-based workflow with source evidence and a review queue, inspect the Feedback Intelligence Toolkit. It is included in the Customer intelligence category pack or complete collection. The individual Aspect Sentiment guided skill does not contain that Python toolkit.
Inspect that toolkit’s input, sample report and exact purchase options ↗