How do keyword blocklists overblock news, and how do you scope one?
Part of Brand suitability
The short answer
A keyword blocklist stops ads from serving on pages whose URL or text contains listed words. It can overblock news because reporting uses the same words as harmful content, so responsible stories lose ads. Scope a list by giving each term a purpose, preferring exact phrases, setting expiry dates, exempting trusted news pages and reviewing it each quarter.
Key takeaways
- A blocklist matches words, not meaning, so it can remove responsible reporting and still miss a harmful page that avoids the listed words.
- The IAB's 2020 guide put the false positive rate of keyword URL tools at generally around 20 percent.
- Generic risk words and identity terms cause the overblocking; brand specific crisis phrases are where a list still earns its place.
- Each term should carry a purpose, a match level, an owner and an expiry or review date.
- Category settings and inclusion lists can carry the broad protection, leaving the keyword list short and specific.
How does a keyword blocklist decide what to block?
Definition
- Keyword blocklist
- A list of words or phrases that stops an ad from serving when they appear in a page's URL, its text or a video's metadata. Some tools call it a keyword exclusion list.
Where the list looks changes what it catches. The IAB's December 2020 brand safety guide (opens in a new tab) separates matching on the URL from matching on the body of the page and asks buyers to learn which one each tool uses. Its example is a crime report about a fatal crash with a suspected drunk driver: a URL only check with “death” or “drunk driving” on the list does not block the page, because neither word is in the address.
| Match level | Strengths | Weaknesses |
|---|---|---|
| URL only | Fast, cheap and usable before the bid, because the address arrives with the ad request | Misses pages whose address lacks the word; flags harmless addresses that contain it |
| Page text | Catches words in headlines and body copy, not only in the address | Blocks a page for a passing mention; cannot tell reporting from promotion |
| Platform list, such as Google Ads excluded content keywords | One list applied across YouTube, Display Network and Google video partner inventory | Exact match only, up to 1,000 keywords at account level, and not applied to YouTube Sponsorship campaigns |
Google's own content suitability help page (opens in a new tab) warns that “some words have multiple meanings” and that excluding a keyword stops ads on related content “even in a different context than you intended.” The IAB guide gives the same warning with homonyms: Corona is a beer brand and the Spanish word for crown, “die” is the German word for “the”, and “killing” appears in a song title.
Why do keyword blocklists overblock news?
News is written in the vocabulary of risk. A report on a flood, a trial or an election uses the same words as content that glorifies harm, and a word list cannot see the difference. The IAB guide put the industry false positive rate for keyword URL tools at generally around 20 percent, meaning about one in five flagged impressions was on content later judged safe for advertising.
The problem grows during big stories. In the IAB's example of a large advertiser, about 1 percent of impressions were normally flagged by its keyword list, and the flag rate rose to 8 to 10 percent when terms such as “Coronavirus”, “COVID 19” and “George Floyd” were on the list. The guide notes that some brands excluded words such as “protest” and “black lives matter” in 2020, and says this potentially limited scale and caused short term revenue losses across the publishing industry.
Demonetisation of news content through brand safety tools is already widespread in display advertising.
Identity terms are a second source of harm. Griffin's paper cites an investigation in which keywords such as “gay” triggered demonetization on YouTube, and argues that LGBTQ+ content was demonetized indiscriminately. A 2024 review by the Reynolds Journalism Institute (opens in a new tab) at the University of Missouri reports that Black and LGBTQ+ publishers have been hit especially hard. Words that describe who people are say nothing about whether a page is hateful.
Which words belong on a keyword blocklist, and which do not?
The IAB guide says keyword lists work best for content a brand is “hypersensitive to avoiding”, used alongside other tools. That points to a short list of specific phrases rather than a long list of general words. Sorting each term into one of five types makes the cleanup decision quick.
| Term type | Example | Recommended action |
|---|---|---|
| Brand specific crisis phrase | The brand's name next to “recall” or “lawsuit” | Keep as an exact phrase, with an owner and a review date |
| Named event | The name of a specific storm or incident | Keep only for a set period, then remove |
| Generic risk word | Death, attack, fire, crash | Remove and cover the risk with category settings |
| Homonym | Corona, die, killing | Remove, or replace with a phrase that fixes the meaning |
| Identity term | Words for race, religion, sexuality or disability | Remove; handle hate speech through its own category |
How do you scope a keyword blocklist step by step?
Seven steps to a keyword blocklist that protects without overblocking
- Write a purpose for each term. One line saying what content the term keeps the brand away from, and why a category setting would not do it.
- Choose the match level on purpose. Decide whether the term should match the URL, the page text or both, and confirm which one the tool actually uses.
- Prefer phrases to single words. A phrase that pairs the brand name with the risk is far narrower than the risk word alone.
- List language variants deliberately. Platform lists such as Google's are exact match only, so each spelling, plural and translation the brand needs has to be added on purpose.
- Set an expiry date for event terms. Terms added for a breaking story come off on a fixed date unless someone renews them.
- Exempt trusted news pages. The IAB describes advertisers using exception lists for news site homepages and section pages to keep scale and support trusted publishers.
- Review the list each quarter. TAG's certification guidelines require certified buyers to document how keywords are added and removed and to review lists quarterly, a sound rhythm for any team.
The IAB also suggests monitoring a sensitive term before blocking it, so the team can see where it would fire before it costs reach. The RJI review asks advertisers to drop overly broad terms and to favor inclusion lists of vetted sites, and it notes that some tools read only the URL, which is why publishers try to keep listed words out of their addresses.
What does a keyword blocklist cleanup look like in practice?
Fictional example
A fictional car dealer group in Florida
A fictional group of car dealerships in Tampa, Orlando and Jacksonville inherits a keyword blocklist of several hundred words from a previous agency. The new planner runs a sample of recent pages from the plan's news and automotive sites against it before the hurricane season campaign.
The list removes crash test reviews and stock market stories because of “crash”, hurricane preparation guides because of “storm” and “flood”, and Spanish language local news because the Spanish versions of several words were added without a purpose. These are pages the dealer group wants to appear beside.
The planner rebuilds the list as a dozen exact phrases pairing the brands the group sells with “recall”, each with a review date. Death, injury and disaster coverage move to category settings with a stated tolerance, and the homepages and weather sections of trusted Florida news sites go on an exception list. The client signs the new list, and the next review is booked for the following quarter.
What should replace a long keyword blocklist?
Keyword lists, category settings and inclusion lists each have a real job, and a plan can use all three. The trade-off is between precision, effort and how easy each one is to explain to a client.
| Tool | Strengths | Weaknesses | Best use |
|---|---|---|---|
| Keyword blocklist | Simple, fast, easy to read and audit line by line | Matches words, not meaning; overblocks news and identity content; misses pages that avoid the words | A few brand specific phrases and short lived event terms |
| Category settings | Grade a topic by how it is treated, so reporting and promotion can be told apart | Depend on a vendor's classification, which is harder to inspect, and vendors can disagree | Standing tolerances for sensitive topics such as violence or alcohol |
| Inclusion list | Only vetted sites run, so unlisted sites do not carry the ad | Takes time to build and maintain, and limits reach to the sites on it | Core plans where the brand wants control over each site |
Since an MRC policy of October 2025, a service accredited only for keyword or text checks at the domain level may not be marketed as brand safety, which is a useful reminder of what a word list can and cannot prove. For how category settings grade sensitive content, see the brand suitability guide; for the content no tolerance allows, see the brand safety floor.
Keyword blocklist audit checklist
- Each term has a written purpose, an owner and a review or expiry date.
- No generic risk word is on the list where a category setting could cover the risk.
- No identity term is on the list.
- Homonyms were removed or replaced with phrases that fix the meaning.
- Each language variant on the list was added on purpose.
- Trusted news homepages and section pages are on an exception list.
- A sample of blocked pages was reviewed this quarter, and the changes were logged.
How does PlacementGraph judge pages without a keyword list?
Frequently asked questions
Do keyword blocklists need French or Spanish versions?
Only if the campaign runs on content in those languages, and then each version should be added on purpose. The IAB describes keyword lists as language specific, and Google's excluded content keywords are exact match only. A French term added without a purpose can remove Canadian French news the brand wanted, just as an English one can.
How long should a keyword blocklist be?
There is no standard length, and length is a poor measure of protection. Google Ads caps excluded content keywords at 1,000 at the account level. A list that only holds brand specific phrases and current event terms stays short, while broad risk coverage belongs in category settings.
Can a publisher find out that a keyword blocked its pages?
Not easily, unless the buyer tells it. The IAB guide says every publisher on a plan should have access to the same data and details about how its content was categorized. Buyers who share why a page was blocked help publishers fix real issues and spot terms that should come off the list.
Should LGBTQ+ or other identity terms ever be on a blocklist?
No. Identity terms describe people, not risk, and research shows they remove responsible coverage and creators. Hateful content can be handled through the hate speech category, which judges how a page treats a group rather than whether it names one.
Sources
- Understanding Brand Safety and Brand Suitability in a Contemporary Media Landscape (opens in a new tab)IAB, December 2020
- About content suitability (opens in a new tab)Google Ads Help
- From brand safety to suitability: advertisers in platform governance (opens in a new tab)Rachel Griffin, Internet Policy Review 12(3), July 11, 2023
- How advertising blocklists undermine online journalism (opens in a new tab)Andy Lee Roth, Reynolds Journalism Institute, November 25, 2024
- Brand Safety Certified Guidelines, version 5.0 (opens in a new tab)TAG (Trustworthy Accountability Group), July 2026
- Policy for Property-Level Ad Verification Representations (opens in a new tab)Media Rating Council, October 18, 2025
Researched and written by PlacementGraph Research. Read how we research, source and update these pages. Corrections: diana@placementgraph.com.