Programmatic Ad Targeting: Audiences, Data, and Strategies

Your programmatic campaign is live, the DSP is spending the budget, and your cost per impression looks competitive. But three weeks in, you have 1.2 million impressions and eleven conversions. The problem is almost never the platform — it is the targeting. Programmatic ad targeting is the highest-leverage variable in programmatic performance, and most startup campaigns underinvest in it before launch.

This guide covers the data types, targeting strategies, and audience-building techniques that separate effective programmatic campaigns from expensive ones.

The Three Data Types That Power Programmatic Targeting

Every programmatic targeting decision uses one of three data types. Understanding which type powers each of your audience segments determines how accurate, scalable, and durable your targeting will be.

First-party data is data you collected directly from your own customers and prospects. This includes your CRM contact list, email subscribers, website visitors tracked by your DSP pixel, app users, and offline customer data. First-party data is the most accurate, the most privacy-compliant, and the only data type not subject to the deprecation of third-party cookies. It is also the most limited in scale — it is only as large as your existing customer and prospect base.

Second-party data is another company's first-party data, shared directly with you through a formal data partnership. Less common, but high quality when available. An example: a fintech startup sharing their customer email list with a complementary B2B software company so each can target the other's customers. Clean room environments make this increasingly feasible without sharing raw user data.

Third-party data is purchased audience segments from data aggregators like Lotame, Neustar, Nielsen, Bombora (for B2B intent), and LiveRamp. These segments contain behavioral, demographic, and intent signals compiled from browsing data, survey panels, purchase data, and other sources. Third-party data enables prospecting at scale but carries two significant limitations: accuracy degrades as the data ages, and privacy regulation is progressively restricting how it can be collected and used.

Targeting as the core of programmatic strategy means starting with your first-party data as the foundation and layering other data types on top, not relying on purchased segments as a substitute for knowing your audience.

Contextual vs Behavioral Targeting: When to Use Each

Behavioral targeting uses audience data — what people have done, searched for, purchased, or expressed interest in — to find them across the web regardless of what content they are currently reading. If someone has browsed SaaS HR software comparison pages in the last 30 days, behavioral targeting lets you reach them on any publisher where they appear.

Behavioral targeting is powerful for reaching specific customer profiles, but it depends on having that behavioral data available and reliable. As third-party cookies disappear from major browsers, behavioral targeting based on cross-site browsing data is becoming harder to execute outside of walled gardens.

Contextual targeting places your ads based on the content of the page being viewed rather than data about the user. Your ad appears on pages about project management software because the content signals relevant reader intent, not because you know anything about the specific person viewing the page.

Contextual targeting has gained significant relevance in the cookieless transition because it requires no personal data. Modern contextual targeting goes beyond keyword matching — AI-driven contextual platforms (Peer39, Grapeshot, Oracle Contextual) analyze full page content, sentiment, and topic relevance to match ads with the most appropriate editorial environment.

Use behavioral targeting for retargeting and high-intent prospecting where you have strong first-party audience signals. Use contextual targeting for reaching new audiences in brand-safe environments without relying on personal data — particularly effective for upper-funnel awareness campaigns.

Applying these targeting strategies to display requires combining both approaches: contextual for cold prospecting and behavioral for retargeting.

How to Build High-Performance Audience Segments

The most effective programmatic audiences are built from your own data, structured around intent signals rather than demographics alone.

Retargeting segments by intent stage: Do not create one "website visitors" retargeting audience. Segment by behavior: all visitors (broad, low intent), product page visitors (medium intent), pricing page visitors (high intent), demo page abandoners (highest intent). Each segment should receive different creative and different bid levels. Higher intent audiences justify higher CPMs because conversion rates are proportionally higher.

CRM matching for lookalike modeling: Upload your customer email list to your DSP's identity graph. The DSP matches emails to cookied or device-based profiles and creates a "seed" audience. Use this seed to find statistically similar people (lookalikes) across the open web. The quality of your lookalike directly reflects the quality of your CRM data — a clean list of your 500 best customers is more valuable than a noisy list of 5,000 mixed-quality contacts.

Intent-based B2B audiences: For B2B startups, Bombora's Company Surge intent data identifies companies actively researching topics related to your product. Combined with firmographic filters (company size, industry, revenue), you can build audiences of companies that match your ICP and are actively in a buying cycle. Which DSPs offer the best targeting options for B2B intent varies — Xandr/Microsoft has strong B2B audience capabilities via LinkedIn data.

Exclusion audiences: Define who you do not want to reach. Exclude current customers from prospecting campaigns (they are converting waste). Exclude visitors who converted in the last 30 days from retargeting. Exclude very short-session visitors (under 5 seconds) from your high-intent segments — they probably landed by accident.

Targeting Strategies That Work for Startup Budgets

Start with retargeting, not prospecting: Cold prospecting on a $5,000/month programmatic budget will produce thin data and limited conversions. Your highest-ROI programmatic spend is converting the visitors who already visited your site. Start there, build conversion data, and use that data to optimize prospecting.

Layer, do not replace: The most effective targeting uses multiple signals simultaneously. A contextual layer (fintech content) plus a behavioral layer (visited competitor sites) plus a first-party exclusion (existing customers excluded) produces a tighter audience than any single data type alone.

Use frequency caps as a targeting tool: Limiting impressions to 3–5 per user per day prevents frequency fatigue, but it also keeps your budget distributed across more unique users. If your frequency cap is too high, a small retargeting pool absorbs your entire budget. If too low, you lose the repetition required for brand recall.

Geographic and daypart refinement: If your conversion data shows 80% of conversions come from users in specific metro areas or during specific hours, tighten your targeting to focus spend where it performs. How bad targeting inflates fraud risk is one reason to keep audience definitions precise — broad targeting on open exchange means more exposure to fraudulent inventory.

What'S Changing: Cookieless Targeting in 2026

The deprecation of third-party cookies in Chrome — finalized for most users by late 2024 and now essentially complete — fundamentally changed how behavioral targeting works outside of walled gardens like Google and Meta.

The practical impact for programmatic ad targeting:

First-party data is now the primary targeting asset: Advertisers who invested in first-party data collection — consent-based email capture, server-side tracking, CRM enrichment — are experiencing minimal targeting disruption. Those who relied on third-party behavioral data are seeing audience scale drop by 30–50%.

Identity solutions are filling the gap: Unified ID 2.0 (The Trade Desk), LiveRamp's RampID, and other deterministic identity graphs match users via hashed email addresses rather than cookies. These work when users are logged in or have consented to email-based tracking. Adoption is growing but not universal.

Contextual targeting is experiencing a renaissance: Without behavioral data for cold prospecting, contextual placement has become the dominant method for reaching new audiences. Startups building contextual targeting capabilities now have an advantage over those waiting for identity solutions to mature.

Privacy Sandbox alternatives: Google's Privacy Sandbox APIs (Topics, Protected Audience) are now in active deployment. They allow audience-based buying without individual cross-site tracking, but the precision is lower than cookie-based behavioral targeting. The ecosystem is still adapting.

The targeting strategies that CTV audience targeting uses — device graph matching, household IP targeting — are less affected by the cookie deprecation than open web display targeting, which is one reason CTV has gained budget share from display.

Key Takeaways

  • First-party data (your CRM, site visitors, customers) is the most accurate targeting foundation and the only one not vulnerable to privacy regulation changes.
  • Segment retargeting audiences by intent stage — pricing page visitors need different messaging than general site visitors.
  • Contextual targeting has regained strategic importance as cookieless browsing becomes the norm for cold prospecting.
  • The best campaigns layer multiple targeting signals rather than relying on any single data type.
  • B2B startups should explore Bombora intent data and LinkedIn-sourced audience segments through DSPs that support them.
  • Cookieless transition is largely complete; if you have not built a first-party data strategy yet, it is now urgent.

FAQ

What is the most accurate type of programmatic targeting? Your own first-party retargeting data — specifically, users who completed a defined action on your site or app — is the most accurate targeting signal available. Behavioral third-party data is the least accurate because it is compiled indirectly and degrades rapidly. Contextual targeting sits in between: highly accurate for content relevance, but it tells you nothing about the individual user's intent or stage in the buying cycle.

How do lookalike audiences work in programmatic? You upload a seed audience (typically a CRM list or your best-converting customer segment) to your DSP. The DSP's identity graph finds profiles that share behavioral and demographic characteristics with your seed. The DSP then creates an expanded audience of people who "look like" your best customers but have not yet interacted with your brand. Performance depends heavily on the quality and specificity of your seed list.

Can I target by job title or company in programmatic display? Not directly from most open-web DSPs. Job title and company targeting is primarily a LinkedIn capability. However, B2B DSPs and platforms like Xandr/Microsoft Invest can apply LinkedIn-derived professional audience segments to open web inventory. Bombora intent data can identify companies researching specific topics, which combined with firmographic filtering (company size, industry) approximates job-level targeting without requiring individual professional profile data.

How often should I refresh my audience segments? Retargeting windows should be recency-based: 7-day windows for high-intent visitors (pricing, demo), 30-day windows for general site visitors, 90-day windows for upper-funnel content visitors. CRM-based lookalike seeds should be refreshed quarterly or whenever you have added a significant number of new customers. Third-party data segments typically auto-refresh on the provider's schedule, but verify that your provider updates segments at least monthly.