Custom Signals

Custom Signals let you monitor data sources for specific changes on a regular schedule. You can:

• Monitor websites, social media platforms, and other digital sources.
• Track technology adoption, hiring profiles, and new feature rollouts.
• Monitor compliance changes (e.g., when a company adds GDPR compliance to their trust center).
• Set up RSS feeds for web mentions.
• Schedule automated runs using Clay's AI tool to create your own unique signals.

Creating a custom signal

1. To create a custom signal:
   • In a workbook: Click + Add → Custom signal.
   • In a table: Click Actions → Import → Custom signal.
2. Select a source to monitor.
   • Next to each source, you'll see the estimated Clay credit cost per result.
3. Configure your source.
   • Some sources require an account, while others use a Clay-provided account.
4. Set how frequently you want your signal to run.
5. Optionally, add enrichments (such as Slack notifications or Salesforce updates that trigger when the signal runs).
6. Click Save.

Editing a custom signal

1. Click the column title of the signal (signals will have a toggle next to their titles)
2. Click Edit column.
3. Adjust any of the configurations or the frequency to run and click Save.

Guide: Turning enrichments into signals

You may occasionally need to monitor changes in an enrichment. Below is a step-by-step guide on creating a signal for any enrichment. In this guide, we'll start with a list of companies and add enrichments to monitor.

Table 1: Setting up an enrichment

1. Start with a list of companies (Create a new table or in an existing table, click Actions → Import)
2. If you don't already have one, create the enrichment you want to turn into a signal.
3. Set up a scheduled run for your enrichment by clicking the ⚙️ → enable Re-run columns on a schedule.
4. Under Actions → click Send table data and include the enrichment in the columns that are sent. Send this data to a new table that we'll call "Run history".
   • Make sure the company name or company domain is part of the data sent to the new table, as these will be essential in the upcoming steps.
5. Toggle off Update existing rows on re-run to create new rows for each recurring run.

Table 2: Setting up a run history table

1. Go into the "Run history" table and pull out the company name/domain and the output of the enrichment as columns from the source.
2. Click the columns dropdown in the upper left corner of the table and unhide Created At.

Table 3: Identify the difference between runs

1. In a third table, which we'll call the "Lookup" table, click Add enrichment → Lookup Multiple Rows in Other Table.
2. Set up the lookup to check the "Run history" table and use the company name or domain as the identifier.
3. Set up a scheduled run for this lookup by clicking the ⚙️ → enable Re-run columns on a schedule.
   • This should run at the same schedule as the first table.
   • Note: You can't currently set a specific time for a scheduled run. Instead, it runs every 24 hours from when you first schedule it. To avoid conflicts, ensure a short delay between the scheduled times of the first and second table.
4. Click Add enrichment → Use AI. Generate a prompt that references the output from the Lookup Multiple Rows in Other Table and identifies the difference between the two most recent runs, using Created At.
   
   • This prompt will provide two outputs: any new information returned from the 2nd run that wasn't in the 1st run, and a True/False boolean indicating if there is any new information.‍
   • Note: If your data is structured or numerical, you can use a formula to detect changes. However, if there's significant variability between outputs, you'll likely want to use an AI action to holistically determine the difference between runs.
5. Take the result from Use AI and write it to a fourth table (the "Signal" table), using Actions → Send table data.
   
   • You can do this using run conditions.
     
     • For a numerical output, the run condition should be if the change is not 0.
       
     • For a text output (string), the run condition should be if the boolean generated from the AI action is True.

Table 4: Store the signal

1. Go into the "Signal" table and pull out the company name/domain and the output from the AI action that described the difference between runs.
2. Click the columns dropdown in the upper left corner of the table and unhide Created At. This date will serve as a timestamp showing when the signal was detected.