[CX Enterprise Coworker] Check Your Experience Platform Data Quality with Coworker
[OVERVIEW]
You can check the quality of the data in your Adobe Experience Platform datasets with a plain-language prompt in Adobe CX Enterprise Coworker. It gives you a quick read on whether a dataset or a single field can be trusted, without writing SQL queries or digging through complex schema hierarchies. Coworker uses its Data Validation skill to run statistical and semantic checks on a sample of your data and returns the results in the same conversation. The skill is read-only, so it flags potential issues without changing your data, schemas, or mappings.
[IMPACT]
Use it for a fast quality check on a dataset or a field, for example after a new implementation or an implementation update, or as an ongoing check on a critical dataset to catch regressions early. The output is a results table with one row per field showing the share of valid, distinct, and null values, the most common values, the most common invalid values with an explanation, and a short note on the field's quality, followed by suggested next steps. For example, in a scenario where a field's values look wrong after a mapping change, inspecting its top values and invalid values can show where the problem is.
[LESSONS]
The word dataset in front of the name helps Coworker identify it correctly, so Validate the dataset Electronics Sample 1000 works better than Validate Electronics Sample 1000. The skill checks a sample of the dataset, typically the most recent 1,000 rows, so the results describe that sample and not every row. A dataset check covers up to five fields per request, and you can either name those fields or let Coworker choose them. Part of the invalid value detection relies on LLM-based inference, so it can occasionally miss subtle errors or flag borderline values, which makes the list of invalid values worth reviewing before you act on it. If you need more exhaustive checks or complex business logic, supplement the results with other tools such as Query Service or Data Prep validations.
[TRYABLE]
1. Sign in to Coworker and select New Chat.
2. Have the name or ID of the dataset you want to validate ready. If you want to check one specific field, have its name ready too.
3. Ask Coworker to validate the dataset, starting the name with the word dataset. For example: Validate the dataset Electronics Sample 1000. Coworker sends the request to its Data Validation skill, which analyzes a sample of the dataset and returns the results in the same conversation. A dataset check covers up to five fields, which you can name in the prompt or leave Coworker to choose.
4. To check a single field instead, name it in the prompt. For example: Validate the email field in the Customers_2024 dataset.
5. Review the results. Each validated field appears as a row in a table with its path, type, valid values, distinct values, null values, top five distinct values, top five invalid values with an explanation, and a short note on its quality. For a single field, Coworker also returns a chart, and you can switch between the chart and table views.
6. Use the Next Steps list under the results. It suggests follow-up prompts, such as validating another field or re-running the dataset.
7. To keep the results, select CSV to download the full table.
[SETUP]
Make sure you have access to Adobe CX Enterprise Coworker Chat, an Adobe Experience Platform dataset to validate, and the name or ID of that dataset. If you want to check one field, you also need that field's name. The skill is read-only and does not change your data, schemas, or mappings.
[SAMPLE_OUTPUT]
This recipe was sourced from Adobe's documentation. Full depth and context are available there: https://experienceleague.adobe.com/en/docs/cx-enterprise-ai/experience-cloud-ai/coworker/chat/use-cases/data-insights/data-validation-aep