Building Retargeting Audiences

This guide walks you through creating audience segments for digital advertising retargeting and activating them to ad platforms like Google Ads, Facebook/Meta, and LinkedIn. You’ll learn how to define traits for ad targeting, build audiences with the right conditions, connect ad platform destinations, and monitor match rates.

When to Use This

Retargeting audiences are the right approach when you need to:

• Re-engage website visitors who didn’t convert
• Target users who abandoned their cart with product-specific ads
• Build lookalike seed audiences from your best customers
• Suppress existing customers from acquisition campaigns
• Run A/B tests across different audience segments with ad spend

Prerequisites

• A configured warehouse connected to your data warehouse
• A schema with a User entity type (or equivalent)
• At least one ad platform account:
  • Google Ads (Customer Match enabled)
  • Facebook/Meta Business Manager
  • LinkedIn Campaign Manager

Step 1: Create Traits for Ad Targeting

Effective retargeting starts with the right signals. Build traits that capture the behavioral and demographic data your ad campaigns need.

Behavioral Traits

These traits capture recent customer actions and intent signals:

Recent Website Visitors (SQL Trait):

SELECT
  user_id,
  CASE
    WHEN MAX(page_view_timestamp) >= DATEADD('day', -7, CURRENT_DATE) THEN true
    ELSE false
  END AS visited_last_7_days
FROM website_events
WHERE event_type = 'page_view'
GROUP BY user_id

Cart Abandoners (SQL Trait):

SELECT
  user_id,
  true AS abandoned_cart
FROM cart_events
WHERE event_type = 'add_to_cart'
  AND user_id NOT IN (
    SELECT user_id
    FROM order_events
    WHERE event_type = 'purchase'
      AND event_timestamp > cart_events.event_timestamp
  )
  AND event_timestamp >= DATEADD('day', -30, CURRENT_DATE)
GROUP BY user_id

Product Category Interest (SQL Trait):

SELECT
  user_id,
  MODE(product_category) AS top_product_category
FROM product_views
WHERE view_timestamp >= DATEADD('day', -30, CURRENT_DATE)
GROUP BY user_id

Value-Based Traits

These traits help you bid more effectively and build lookalike seeds:

Customer Lifetime Value (Aggregation Trait):

• Source table: orders
• Function: Sum
• Column: order_total
• Group by: user_id

Purchase Frequency (Aggregation Trait):

• Source table: orders
• Function: Count
• Group by: user_id

Days Since Last Purchase (SQL Trait):

SELECT
  user_id,
  DATEDIFF('day', MAX(order_date), CURRENT_DATE) AS days_since_last_purchase
FROM orders
GROUP BY user_id

Creating the Traits

For each trait:

1. Navigate to Segment → Traits
2. Click Create Trait
3. Select the entity type (e.g., “User”)
4. Choose the trait type (SQL, Aggregation, or Formula)
5. Enter the definition
6. Set the schedule — for retargeting traits, daily evaluation is usually sufficient; use hourly for time-sensitive signals like cart abandonment
7. Click Save

Step 2: Build Audiences with Targeting Conditions

Combine your traits into audiences that represent distinct targeting strategies.

Audience: Cart Abandoners (Last 7 Days)

Entity Type: User
WHERE
  abandoned_cart = true
  AND visited_last_7_days = true
  AND has_purchase_history = false

This targets users who recently added items to their cart, visited your site in the past week, but haven’t purchased. These are high-intent users who may convert with a reminder.

Audience: Lapsed High-Value Customers

Entity Type: User
WHERE
  lifetime_value > 500
  AND days_since_last_purchase > 60
  AND days_since_last_purchase <= 180

These are your best customers who’ve gone quiet. Re-engagement ads with personalized offers can bring them back before they churn completely.

Audience: Lookalike Seed — Top 10% Customers

Entity Type: User
WHERE
  lifetime_value_percentile >= 90
  AND total_orders >= 3
  AND email IS NOT NULL

A high-quality seed audience for lookalike/similar audience generation on ad platforms. The tighter the seed, the better the lookalike quality.

Audience: Suppression — Existing Customers

Entity Type: User
WHERE
  total_orders >= 1
  AND last_order_date >= DATEADD('day', -90, CURRENT_DATE)

Sync this as a suppression list to exclude recent customers from acquisition campaigns. This prevents wasting ad spend on people who’ve already converted.

Building Each Audience

1. Navigate to Segment → Audiences
2. Click Create Audience
3. Select the entity type
4. Add conditions using the filter builder
5. Click Estimate to gauge the audience size — ad platforms typically require a minimum audience size (e.g., 1,000 for Google Customer Match)
6. Click Preview to spot-check members
7. Name the audience and click Save

Step 3: Connect Ad Platform Destinations

Set up connections to the ad platforms where you want to activate your audiences.

Google Ads

1. Navigate to Destinations → Add Destination → Google Ads
2. Click Connect with Google and authorize with an account that has access to the Google Ads account
3. Select the Google Ads Customer ID (the 10-digit account number)
4. Click Save

Requirements: Customer Match must be enabled on your Google Ads account. This requires meeting Google’s policy requirements for customer data usage.

Facebook / Meta

1. Navigate to Destinations → Add Destination → Facebook Custom Audiences
2. Click Connect with Facebook and authorize with a Business Manager admin account
3. Select the Ad Account ID
4. Click Save

Requirements: Your Facebook Business Manager must be verified and your ad account in good standing to use Custom Audiences.

LinkedIn

1. Navigate to Destinations → Add Destination → LinkedIn Ads
2. Click Connect with LinkedIn and authorize with a Campaign Manager account
3. Select the LinkedIn Ad Account
4. Click Save

Requirements: LinkedIn Matched Audiences requires a minimum audience size of 300 matched members.

Step 4: Configure Customer Match / Custom Audience Syncs

Create audience syncs that push your segments to each ad platform.

1. Navigate to Segment → Audience Syncs
2. Click Create Audience Sync
3. Select the audience (e.g., “Cart Abandoners - Last 7 Days”)
4. Select the destination (e.g., Google Ads)
5. Choose the sync mode:
   • Mirror — Recommended for most retargeting audiences. Keeps the ad platform audience exactly in sync with your warehouse segment.
   • Additive — Use for growing seed audiences or when you want to accumulate members over time.
6. Map identifier fields:

SignalSmith automatically hashes PII (email, phone) before sending to ad platforms that require it. You do not need to pre-hash these fields.

7. Set the sync schedule:
   • Cart abandoners: Every 6 hours (time-sensitive)
   • Lookalike seeds: Daily (less time-sensitive)
   • Suppression lists: Every 6 hours (prevent wasted spend)
8. Click Save & Run

Step 5: Monitor Match Rates

Match rate — the percentage of your audience members that the ad platform can identify and target — is the key metric for retargeting effectiveness.

Checking Match Rates

1. Go to the Sync Runs tab on your audience sync
2. Review the run summary:
   • Members Sent — Total records pushed to the ad platform
   • Members Matched — Records the platform matched to real users (when reported by the platform)
   • Match Rate — Members Matched / Members Sent × 100

Typical Match Rates by Platform

Improving Match Rates

• Include multiple identifiers — Send email, phone, and mobile ad ID when available. Platforms use all identifiers to increase matches.
• Use primary email addresses — Work email addresses match better on LinkedIn; personal email addresses match better on Facebook.
• Clean your data — Standardize phone numbers (E.164 format), trim whitespace from emails, and remove obviously invalid addresses.
• Increase audience size — Larger audiences tend to have higher absolute match counts, even if the percentage stays similar.

Step 6: A/B Test with Audience Splits

Use SignalSmith’s split feature to divide an audience into random groups for controlled ad experiments.

Creating a Split

1. Navigate to Segment → Splits
2. Click Create Split
3. Select the audience to split (e.g., “Lapsed High-Value Customers”)
4. Define the split percentages:
   • Test group (80%) — Receives retargeting ads
   • Holdout group (20%) — No ads, serves as control
5. Name each group descriptively (e.g., “Retargeting - Test”, “Retargeting - Holdout”)
6. Click Save

Activating Split Groups

1. Create separate audience syncs for each split group
2. Sync the test group to your ad platform destination
3. Keep the holdout group unsynced (or sync it as a suppression list)
4. After the campaign runs, compare conversion rates between the test and holdout groups to measure incremental lift

Measuring Incrementality

The holdout group’s conversion rate represents the baseline (organic conversions that would have happened without ads). The difference is the true incremental impact of your retargeting campaign.

Best Practices

• Refresh frequently — Retargeting audiences go stale quickly. Cart abandoners from 30 days ago are much less likely to convert than those from 3 days ago. Set aggressive time windows and frequent sync schedules.
• Layer suppression — Always suppress recent converters from retargeting campaigns. Sync a “Recent Purchasers” audience as an exclusion list to every ad platform.
• Match your creative to your audience — Cart abandoners should see the products they browsed. High-value lapsed customers should see personalized win-back offers. Use trait values to inform your ad creative strategy.
• Respect frequency caps — Retargeting can become annoying. Use ad platform frequency caps to limit how often the same person sees your ads.
• Monitor and iterate — Review match rates and campaign performance weekly. Low match rates may indicate data quality issues. Low conversion rates may indicate targeting that’s too broad.

Next Steps

• Sync CRM Audiences — Extend activation to your CRM
• Splits documentation — Deep dive into audience split configuration
• Priority Lists — Resolve overlap when customers qualify for multiple audiences
• Google Ads destination — Platform-specific configuration
• Facebook destination — Platform-specific configuration
• LinkedIn destination — Platform-specific configuration