I recently completed the HDI Customer Experience Foundationscourse with other IT leaders. Many expressed frustration as current financial constraints cannot accommodate another software purchase to help collect and analyze customer feedback and sentiment.
But don’t despair! With a good dataset and a strong prompt in an AI platform, you can quickly identify the top drivers of dissatisfaction, emerging issues and alist of fixes to prioritize.
Step 1: Gather feedback
Start with tools that are free or may already be licensed in your environment.Many of the ITSM platforms have a built-in customer satisfaction survey tool. Additionally, you may explore other free survey platforms,such as:Survey Monkey,Microsoft Forms or Google Forms.
Step 2: Prepare your data
Export the survey data into a spreadsheet for review. If you do not have access to asecure AI platform with data protection, remove any personal or sensitive data from your dataset before proceeding.You may want to standardize categories, remove null values and convert data values into common terms for clarity. For example, if the data uses a numerical rating for satisfaction, convert the values to words such as satisfied, dissatisfied, etc. Ensure the spreadsheet columns have descriptive headings.
Step 3: Login to your AI platform
Upload your prepared file within the AI platform and craft your prompt. Prompts should follow a basic framework. Start by grounding the prompt with a persona, including the role or position they should assume and the environment of the data. Explain the data and what you are looking for in as much detail as possible. Then specify the output you are seeking. As anyone who uses AI knows, the better your prompt, the better your results.
Taking the example of our dataset, you might consider starting with the following prompt:
“Act as a senior strategist/consultant in the [industry] with deep experience in service desk + CX survey analytics. Be direct, evidence-driven, and practical.
I am providing a dataset (last6 months) containing customer survey results. Use the column headers as provided. If a needed field is missing, state that explicitly and proceed with what is available.
Privacy: Do not output any personal or sensitive information. If you use customer quotes, remove identifiers (names, emails, phone numbers) and paraphrase when necessary.
Output in two clearly labeled sections:
A) Analyst notes (bulleted):
B) Executive briefing (paste-ready for Word, ≤600 words):
-
Recommended next steps as decisions (include: decision, expected impact, effort level, owner role, and the metric to track).
Your tone should be confident, concise, no hedging. Write as a trusted advisor, not a report generator.”
Step 4: Review the results
Once you receive the results, review the insights and cross-check the data. Consider validating the output by running the analysis in a second tool and sanity-checking against a small sample of raw comments and tickets. Then, turn the top themes into a measurable action plan.
Even in a constrained budget environment, you have options. Use the tools you already have, apply AI responsibly and turn customer feedback into the operational improvements your users will feel.