Design an effective Quantify query

Start with Enterpret Agent to answer a feedback question or prepare a report. Describe the measure, scope, and period you need. For recurring reporting, daily summaries, or weekly digests, use scheduled Agent automations; add connectors for external context or supported actions. Use the Quantify controls below when you need to build or inspect a specific chart or calculation.

Start with the decision you want the analysis to support, then choose the population, measure, grouping, and dates that answer it. This guide helps you design a query; for the interface walkthrough, see Build your first Quantify analysis.

1. Write one answerable question

For example: “Which themes appear most often in feedback from our chosen support source during the last completed month?” This defines a population, a period, a grouping, and a count. Avoid combining volume, customer reach, and business impact into one undefined measure.

Decide whether you need to find the largest groups, compare populations, examine a trend, or inspect individual records. Feedback patterns can guide investigation; they do not by themselves prove a cause or the effect of a product change.

2. Choose the grouping and the measured object

Plot chooses what the result is grouped by. Measured as chooses the object, field, and calculation used to measure those groups, from the options available for your query.

• Use a theme or other available taxonomy field to group feedback by topic.

• Use Count of feedback records for feedback volume. A count of linked accounts or users answers a different question and depends on the available relationships.

• Use Percentage for a defined share, then inspect its denominator. Use numeric calculations such as Mean only when the selected field and its data support the question.

Do not assume that counting feedback measures the number of unique customers, or that a selected group's share always uses every filtered record as its denominator. See Calculations on Quantify.

3. Define the population with Filter

Choose Add Filter and select the fields and values available in your workspace. For the example above, choose the intended support source. Add a product-area or customer-segment condition only when it belongs in the question and the corresponding field is available.

Check the AND/OR grouping. “Support source AND product area” differs from “Support source OR product area.” Missing metadata or incomplete user/account links can affect which records qualify.

For a Percentage query, review the denominator after changing numerator filters. Its filters can be configured separately, and whether it follows a filter change depends on the current configuration. Confirm the actual population rather than assuming the numerator and denominator changed together.

4. Set Duration deliberately

Choose the period using Duration. Use explicit dates for a completed calendar month or quarter, and review the displayed boundaries. A rolling 30-day period is not always a calendar month.

For a comparison, decide whether the periods need equal lengths or a different normalization. Check source coverage and incomplete recent data before attributing a difference to customer behavior.

5. Add a breakdown or comparison when it answers the question

Use Breakdown By, when offered, to split each plotted group by a second field. For example, a theme analysis measured by feedback-record Count can be broken down by Source to show where each theme is reported. Keep the same population and dates.

Use Compare for two periods or filter populations, when supported. Breakdown By and Compare cannot be used together. Keep the measured object and calculation consistent, and inspect each side's scope and Percentage denominator. For filter comparisons, do not assume Filter B inherits the conditions in Filter A.

See Compare feedback in Quantify for the supported workflow.

6. Review the results and check the evidence

After the results load, check the measure, filters, period, and coverage before interpreting the chart.

• Open supporting records from linked values or View Feedback when offered. Confirm that the records match the question.

• Distinguish the chart's selected Top values from the full analysis population. Hiding a row does not remove its data from a Percentage denominator.

• Check overlapping categories before adding group counts or expecting percentages to total 100%.

• Interpret an empty result within the searched scope. Check filters, dates, source coverage, and missing values before concluding that no customer reported an issue.

Complete example designs

These are hypothetical setups. Use the corresponding fields and sources available in your workspace.

Continue with the relevant guide

• Quantify query reference for controls and query options.

• Read and interpret Quantify results for charts, record inspection, and anomaly flags.

• Save, share, and export Quantify analyses for saving the reviewed query and distributing it.