> ## Documentation Index
> Fetch the complete documentation index at: https://docs.dataville.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Tag CRM accounts by industry in Claude

> Upload a CRM export to Claude and get back a CSV file with an industry for every account.

export const accountsTemplate = (String.raw)`account_id,account_name
ACC-1001,The Home Depot
ACC-1002,Shopify
ACC-1003,Pfizer
ACC-1004,Delta Air Lines
ACC-1005,Northwind Payments LLC
`;

You can fill in the industry of every account in your CRM from each company's Wikipedia article. An article's categories and opening paragraph name the company's industry, and Claude can match them to the industry values your CRM allows.

In this guide, you'll upload an export of your accounts to Claude. Claude will look up each company in Wikipedia through Dataville and create a CSV file with an industry for each account, ready to import into your CRM.

## Before you start

You'll need:

* Claude on the web or desktop with Dataville added as a connector. See [Quickstart: Claude](/quickstart-claude) for the setup.
* An export of your CRM accounts, with an ID and a company name for each account.

## Tag your accounts

Once you've connected Dataville to Claude, tagging a list of accounts takes a few minutes.

<Steps>
  <Step title="Turn on file creation in Claude">
    Claude needs file creation turned on to give you a file to download. In Claude, go to **Settings → Capabilities** and turn on **Code execution and file creation**.

    File creation is off by default on the Free, Pro, and Max plans. On Team and Enterprise plans it's on by default, and an owner controls it in **Organization settings → Capabilities**. If you'd rather leave it off, ask Claude to reply with the CSV in the chat instead of creating a file.
  </Step>

  <Step title="Download the accounts template">
    The template has an `account_id` column and an `account_name` column, like a CRM export, and five example accounts.

    <a href={"data:text/csv;charset=utf-8," + encodeURIComponent(accountsTemplate)} download="accounts.csv" className="block">
      <Card title="accounts.csv" icon="file-csv">
        Download the template with the example accounts.
      </Card>
    </a>

    The file looks like this:

    ```csv theme={null}
    account_id,account_name
    ACC-1001,The Home Depot
    ACC-1002,Shopify
    ACC-1003,Pfizer
    ACC-1004,Delta Air Lines
    ACC-1005,Northwind Payments LLC
    ```

    Northwind Payments LLC is a made-up private company. To tag your own accounts, export them from your CRM with the same two columns, or replace the example rows.
  </Step>

  <Step title="Ask Claude to tag each account">
    Start a new chat in Claude and turn on the Dataville connector. Attach your `accounts.csv` file and send this prompt, with your CRM's industry values in place of the example list:

    ```text theme={null}
    Search Dataville's Wikipedia source for each account in accounts.csv and create a CSV file named accounts-with-industry.csv. Use only the results from Dataville. Keep the original columns and add these:

    - wikipedia_url: the link to the article. Leave it blank if the article isn't about the company.
    - wikidata_id: the article's Wikidata ID.
    - industry: one of Retail, Technology, Healthcare, Transportation, Financial services, or Other. Choose it from the article's categories and opening paragraph, and ignore categories about Wikipedia itself, stock listings, locations, or founding dates. Use Unknown if there's no article about the company.
    - industry_source: the category or phrase the industry is based on.
    ```

    Claude searches the Wikipedia source for each account, and its response lists a "Search a Dataville data source" step for each search.

    If Wikipedia is unavailable, Dataville answers from its stored copy and the search result says the data may be out of date. See [Response format](/response-format#stale-results) to learn more about stale results.
  </Step>

  <Step title="Review the industries">
    Claude replies with a summary and the file `accounts-with-industry.csv` for you to download. For the example accounts, the file has these rows, shown without the `account_id` and `wikipedia_url` columns:

    | account\_name | wikidata\_id | industry | industry\_source |
    | - | - | - | - |
    | The Home Depot | Q864407 | Retail | Home improvement retailers of the United States |
    | Shopify | Q7501150 | Technology | E-commerce software |
    | Pfizer | Q206921 | Healthcare | Pharmaceutical companies of the United States |
    | Delta Air Lines | Q188920 | Transportation | Airlines based in Georgia (U.S. state) |
    | Northwind Payments LLC | None | Unknown | None |

    Wikipedia's categories mix industries with maintenance tags and facts like stock listings, so Claude reads the opening paragraph too. Shopify's categories include "Retail companies established in 2006", but its article describes e-commerce software, so its industry is Technology.

    Wikipedia has no article about Northwind Payments LLC, so its industry is Unknown. Review Unknown rows by hand.

    Every spelling of a company's name returns the same Wikidata ID, so you can use it to spot duplicate accounts.
  </Step>

  <Step title="Import the industries into your CRM">
    Import `accounts-with-industry.csv` with your CRM's import tool, matching rows on `account_id`, to update each account's industry. Leave out the Unknown rows, or fill them in first.
  </Step>
</Steps>

## Next steps

To learn what else the Wikipedia source returns, or to look up companies in SEC EDGAR as well, see these pages.

<CardGroup cols={2}>
  <Card title="Data sources" icon="database" href="/data-sources">
    Every source in the catalog, and how to search each one.
  </Card>

  <Card title="Check business applications in Claude" icon="clipboard-check" href="/guides/underwrite-from-claude">
    Check companies against their SEC filings and get back a CSV file.
  </Card>
</CardGroup>


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.