Writing Great Signals

The quality of your pipeline depends on the quality of your signals. This guide covers how to think about signal design, with real examples you can use or adapt.

---

​

The Two Most Important Signal Types

​

Web Visitors

Your website is your best intent source. Companies visiting your pricing, product, or demo pages are actively evaluating. A Web Visit signal captures this automatically after you install the SDK. Start with high-intent pages:

• /pricing or /plans
• /demo or /book-a-demo
• /product or feature-specific pages
• /case-studies or /customers

Then broaden as needed:

• All website traffic filtered by session duration (e.g., 10+ seconds)
• Blog visits combined with multiple page views

​

Custom AI Signals

Custom AI Signals are Avina’s most powerful feature. Describe any buying behavior in plain language and the Signals Agent builds a detection pipeline for it. What makes them special:

• They read context, not just keywords
• They scan the web, news, job boards, social media, SEC filings, and more
• They catch signals that fixed-platform tools can’t

---

​

Examples of Great Signals

Browse the full Signals Library for 200+ ready-to-use signals. Here are some of the most effective patterns:

​

By use case

Catching active evaluation:

• “Contacts from target accounts who visited your pricing page at least once in the last 7 days” (Web Visit)
• “Companies posting job listings for data engineers that mention Snowflake or dbt” (Job Listing)

Spotting organizational change:

• “Companies hiring their first Chief Information Security Officer” (New Hire)
• “Companies with 3+ C-level executives updating profiles to ‘Former’ in the last 30 days” (Custom AI)

Detecting strategic shifts:

• “Companies actively migrating to the cloud, with engineering blogs mentioning ‘AWS migration’ or ‘cloud transformation’” (Custom AI)
• “Companies shifting to enterprise sales, with website pricing updates adding ‘Contact Sales’ to high-tier plans” (Custom AI)

Finding funding and growth:

• “B2B SaaS companies raising >$20M Series B from top-tier VCs in the last 3 months” (Custom AI)
• “Companies with headcount crossing 100 employees within a 6-month period” (Custom AI)

Tracking competitive dynamics:

• “Companies migrating from Salesforce to HubSpot, with job descriptions mentioning ‘HubSpot implementation’” (Job Listing)
• “Companies acquired by private equity firms in the last 3 months” (Custom AI)

​

By vertical

---

​

Signal Design Principles

​

Be specific about behavior, not demographics

Demographics go in your ICP (Settings > Personas). Signals should capture actions and events that indicate buying intent. Good: “Companies posting job listings for AI/ML engineers that mention ‘production deployment’” Bad: “SaaS companies with 100+ employees” (this is an ICP filter, not a signal)

​

Use qualification prompts for nuance

For signal types that support it, add a qualification prompt. This tells the Qualification Agent to evaluate each signal with AI before delivering it.

“Only surface companies that sell to enterprise customers and have raised at least Series A funding”

Use this for criteria that standard filters can’t express. If a filter already exists for it (industry, location, company size), use the filter instead.

​

Start narrow, then broaden

It’s easier to loosen a signal than to clean up a noisy one. Start with tight criteria, validate the results, then expand.

​

Preview before activating

Always click Preview before activating. Check that the accounts match your expectations. If the results are off, adjust filters before going live.

---

​

Combining Signals with Automations

A signal on its own just surfaces intent. Pair it with an automation to turn that intent into action: See Sample Plays for ready-to-use templates.