Cookieless Contextual Targeting: How to Reach Audiences Without Third-Party Cookies
Third-party cookies did not disappear in one day, but they no longer reach a large part of the web. Contextual targeting works in every browser because it only needs to know what a site is about. This guide shows where cookies stand, which signals remain and how to build a cookieless plan on domain data.
Google dropped its removal plan in 2025. Users can still block cookies in settings.
Third-party cookies have been blocked by default since 2020.
Cookies are partitioned so they cannot follow users across sites.
Apps must ask permission before tracking users across other companies' apps.
Where third-party cookies stand in 2026
For years the industry planned for the end of cookies in Chrome. That plan changed. The table gives the status as we understand it in October 2026; treat it as a summary and check each browser's own documentation before making technical decisions.
| Environment | Third-party cookies by default | Cross-site tracking possible? | What it means for targeting |
|---|---|---|---|
| Chrome (desktop and Android) | Allowed | Mostly yes | Cookie-based audiences still work, but users can switch them off. Incognito windows block them by default. |
| Safari (macOS and iOS) | Blocked | No | Behavioral audiences built on cookies do not reach these users. Context is the main option. |
| Firefox | Partitioned per site | Largely no | A cookie set on one site cannot be read on another, which breaks cross-site profiles. |
| Other privacy-focused browsers | Usually blocked | No | Treat as cookieless inventory. |
| Mobile apps | Not cookie-based | Depends on device ID consent | Where users decline tracking, app context and app categories take over. |
| Connected TV and audio | Not cookie-based | Limited | Targeting relies on content, channel and household signals. |
A note on change: browser policies have shifted several times since 2020. In 2025 Google also wound down most of the replacement APIs it had proposed. Plan for a web where cross-site identity is unreliable, whatever the next announcement says.
Why cookieless planning still matters while Chrome keeps cookies
It is tempting to read the Chrome decision as "nothing changed". In practice, the share of impressions with a usable cross-site identifier keeps shrinking for other reasons.
Browser mix
Safari has a large share of mobile browsing in many rich markets. Those users are already cookieless.
Consent choices
Under consent rules in Europe and elsewhere, many users decline tracking, which removes them from profile-based audiences.
Regulation
Privacy laws keep tightening around profiling and sensitive data, which raises the cost of behavioral targeting.
Fragmented IDs
Alternative IDs cover only parts of the web, so audiences built on them have gaps that context can fill.
The contrast between the two approaches is covered in depth in our guide to contextual targeting vs behavioral targeting. This page focuses on running context well when identity is missing.
Signals that survive without third-party cookies
Cookieless does not mean signal-less. These are the inputs still available on most impressions, ranked from most to least universal.
| Signal | Personal data needed? | Available on | Best use |
|---|---|---|---|
| Domain and its categories | No | Almost every web impression | Topic targeting, suitability, exclusions and audience proxies through personas |
| Page content | No | Pages that have been analysed | Finer topic and sentiment on broad sites |
| Coarse location and language | Usually limited | Most impressions | Market targeting, combined with country-level popularity of the site |
| Time and device type | No | All impressions | Dayparting and format choices |
| Publisher first-party data | Yes, with consent | Logged-in or consenting users on that site | Audience segments sold by the publisher itself |
| Advertiser first-party data | Yes, with consent | Your own customers | Retention and lookalike modeling inside walled environments |
The domain row is the backbone because it is present on nearly every request and needs no consent decision. Our database of websites by category explains each field that comes with a domain.
How cookieless contextual targeting works with domain data
The mechanics are simple enough to run inside any bidder, ad server or planning tool. The domain is looked up in a category file, and the labels decide what happens next.
Request
A bid request or placement report arrives with a domain.
Lookup
The domain is matched against the category file in memory.
Labels
IAB categories, web filtering label, personas and popularity come back.
Rules
Suitability rules and contextual segments decide bid or no bid.
Learn
Results by category feed back into the next round of segments.
Why a local file helps
- No network call inside the auction, so no added delay
- No placement data leaves your systems
- Same answer every time for the same domain
- Works for planning and reporting as well as bidding
What to check in the data
- Several ranked categories per domain, not just one
- Both IAB versions 2 and 3 if partners differ
- Daily coverage of newly registered domains
- Popularity bands to separate real reach from long tail
If you are deciding between a hosted service and a file, our guide to URL categorization API vs database compares the two for latency, cost and freshness.
Personas: audience thinking without tracking anyone
Personas are labels on a domain that describe the kind of visitor the site is built for. They let a planner say "reach bargain hunters" or "reach investors" and translate that into placements, with no profile of any person.
Among the 300,000 most popular domains in our data, more than 1,100 distinct persona labels appear, and most domains carry nine or ten. Read website personas explained for how to use them, and audience segmentation for campaign examples.
Contextual segment recipes for common audiences
Behavioral audiences can usually be rebuilt as a mix of subject categories, web filtering labels and personas. These recipes are starting points; test and refine them against your own results.
| Audience you used to buy | Subject categories | Personas to include | Exclude |
|---|---|---|---|
| In-market car buyers | Automotive, car reviews, auto finance | Auto Enthusiast, Affordable Car Enthusiast, Luxury Car Enthusiast | Parked domains, car piracy forums |
| Small business owners | Business, accounting, software for small firms | Business Owner, E-commerce Entrepreneur, Freelance Professional | Job boards if you want owners, not job seekers |
| Home improvers | Home & Garden, interior design, DIY | DIY Enthusiast, Home Improvement Enthusiast, Interior Design Enthusiast | Real estate listings if you sell tools, not homes |
| Personal finance seekers | Financial Services, investing, banking | Beginner Investor, Savings Enthusiast, Long-Term Investor | Gambling and Cryptocurrency if your compliance team requires it |
| Travel planners | Travel, hotels, airlines, destination guides | Budget Traveler, Luxury Traveler, Business Traveler | News/Media during sensitive events, by page rule |
| Students and learners | Education/Reference, Webinars/E-learning | College Student, Adult Learner, Online Learner | Sites labeled for adult-only categories |
| Gamers | Gaming, esports, game reviews | Casual Gamer, Hardcore Gamer, Role-Playing Gamer | Gambling, which shares some gaming personas |
| Home cooks | Food & Drink, recipes, kitchen equipment | Cooking Enthusiast, Baking Enthusiast, Foodie | Alcohol for youth-focused brands |
Note the gaming row: some gaming and betting personas overlap, so always pair personas with a web filtering exclusion. Industry-level lists, such as travel and hospitality domains, are a quick way to see what a recipe returns.
Mistakes that make contextual campaigns underperform
Lists that are too narrow
A few hundred domains cannot deliver at scale. Widen by related categories and personas before you raise bids.
Only the biggest sites
Top portals are broad and expensive. Mid-popularity niche sites often match the context better and cost less.
Ignoring the long tail risk
The far long tail holds parked, abandoned and made-for-advertising sites. Use popularity bands and exclusions.
One taxonomy version
Partners may still use IAB version 2 while you plan on version 3. Keep both IDs so segments survive the hand-off.
Stale data
A list built once a year misses new sites and sites that changed subject. Refresh monthly at least.
Judging too early
Contextual line items need a few weeks of learning. Compare like with like over the same period.
A four-week plan to launch cookieless contextual campaigns
Most teams can move a share of spend to context within a month. The plan below assumes you already buy programmatically and have a list of past placements.
Find your cookieless share
- Split last quarter's impressions by browser and by presence of an identifier
- Categorize every placement domain and see which categories drove results
- Note which audiences had no reach on Safari and Firefox
Translate audiences into context
- Map each audience to IAB categories and personas
- Set suitability rules and exclusions, including parked and made-for-advertising sites
- Choose popularity bands per market so lists are not all long tail
Run context next to your current setup
- Launch contextual line items with the same creative and budget share
- Hold back a control group or region where possible
- Keep inclusion lists to a manageable size per line item
Compare and expand
- Compare cost per outcome by category, not only by line item
- Drop categories that underperform and widen the ones that work
- Refresh domain lists with newly categorized sites
Measuring results without cross-site cookies
Attribution that follows a person across sites is the first thing to break. These methods measure contextual campaigns without it.
Lift tests
Show ads to a random part of eligible inventory and compare outcomes with a held-out part. Simple and robust.
Geo experiments
Run a campaign in some regions and not in comparable others. Works well for retail and app installs.
Mix modeling
Statistical models that link spend by channel and category to sales over time. Needs a few months of data.
First-party conversions
Sales and sign-ups recorded on your own site and shared with partners under consent, without third-party cookies.
Category reporting
Report results per IAB category and persona group, so you learn which contexts work rather than which users.
Attention and quality
Viewability and time in view by category. Useful when outcome data is thin, but not a replacement for it.
What context does well
- Reaching people while they read about a related subject
- Scaling across browsers, apps and markets with one method
- Keeping ads away from unsuitable content at the same time
- Prospecting for new customers who are not in any of your lists
Where context needs help
- Retargeting a specific past visitor, which needs consented first-party data
- Strict frequency capping across sites without any identifier
- Very broad portals, where page signals add precision
- Products with tiny audiences spread across general sites
Tip: keep at least one always-on contextual line item even while cookie audiences still work in Chrome. It gives you a benchmark, and it means you are ready if browser or consent rules shift again.
Privacy and compliance notes
What contextual targeting avoids
- Building or buying profiles of individual people
- Following users across sites
- Inferring sensitive traits about a person
- Most consent dependencies for the targeting decision
What still needs care
- Storing identifiers on a device for frequency capping
- Precise location or IP-based targeting
- Combining context with personal data in one segment
- Sensitive categories such as health, which some laws restrict for ads
This is general information, not legal advice. For more on data protection and categorization, see privacy compliance and GDPR, and for a full adtech setup, contextual targeting for adtech.
How each team uses cookieless context
The same category file serves different jobs depending on where you sit in the supply chain.
Brands
Write audience briefs as context: subjects, personas, markets and suitability. Ask every partner to deliver against the same domain lists so results can be compared.
Agencies
Keep a library of tested contextual recipes per vertical. Reuse them across clients and refresh the domain lists monthly from one source.
DSPs and bidders
Load the file in memory, enrich every bid request with categories and personas, and offer contextual segments as a standard targeting option.
Publishers and SSPs
Package inventory by category and persona so buyers without cookies can still find it. Check how your own domains are labeled.
Key terms in plain words
| Term | Meaning |
|---|---|
| Third-party cookie | A small file set by a domain other than the site you are on, often used to recognise the same browser across many sites. |
| Partitioned cookie | A cookie kept separate for each top-level site, so it cannot be used to follow a browser from one site to another. |
| Contextual segment | A group of placements chosen by subject, persona or other site attributes rather than by user profiles. |
| Inclusion list | The set of domains a line item is allowed to buy. Often called an allow list. |
| Exclusion list | Domains a campaign must never buy, such as sites below the brand safety floor. |
| Popularity band | A group of sites with similar visitor volume, used to balance reach against niche relevance. |
Cookieless contextual targeting: questions and answers
Are third-party cookies gone in 2026?
Not everywhere. As of October 2026 Chrome still allows them by default, while Safari blocks them and Firefox isolates them per site. Check current browser documentation, as policies change.
What is cookieless contextual targeting?
Choosing where ads run based on what the site or page is about, not on a tracked profile. It works in every browser because it only needs the URL or domain.
Does contextual targeting need user consent?
Classifying a site does not involve personal data. Other parts of ad delivery, such as storing identifiers, may still need consent, so check with your legal team.
Can contextual targeting reach specific audiences?
Yes, indirectly. Domain personas describe the kind of people a site serves, so you can choose placements where your audience is likely to be.
How do you measure cookieless campaigns?
Lift tests, geo experiments, mix modeling and first-party conversion data. Report by category so you learn which contexts work.
Should we stop using cookie audiences in Chrome?
Not necessarily. Where they work and users have consented, they can still perform. The point is to stop depending on them: run contextual segments alongside, measure both the same way and shift budget based on results rather than on browser announcements.
How fresh does contextual data need to be?
Monthly refreshes cover most established sites. New domains appear every day, so daily additions matter if you buy open exchange inventory where fresh sites show up first.
Is domain level enough for contextual targeting?
For most niche and mid-sized sites, yes. For very broad portals, add page-level signals. You can check a domain's categories to see how specific it is.
Contextual data for every browser
IAB categories, web filtering labels, personas and country popularity for over 120 million domains, as files you run in your own stack with a lookup API included.
Chrome is a trademark of Google LLC, Safari and iOS are trademarks of Apple Inc., Firefox is a trademark of the Mozilla Foundation, and IAB is a trademark of the Interactive Advertising Bureau. They are named only to describe browser and industry behaviour. URL Categorization Database is not affiliated with, sponsored or endorsed by any of them.