Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallTo audit an ad campaign for demographic bias, check both who the advertiser made eligible and who the platform actually reached. Define a defensible eligible-audience baseline before looking at results, compare delivery across relevant groups, and document what the evidence can—and cannot—establish. A delivery gap is a reason to investigate, not by itself proof of its cause or of unlawful discrimination.
What an ad-delivery audit should answer
A useful audit distinguishes three questions that are easy to conflate:
As an Amazon Associate I earn from qualifying purchases.
- Who could receive the ad? This is shaped by targeting, exclusions, geography, eligibility criteria and the available audience.
- Who did receive it? This is the actual delivery outcome, measured through impressions or other suitable audience reporting when available.
- Why did the outcome occur, and was it unlawful? Those require causal evidence and a jurisdiction- and campaign-specific legal analysis; a disparity alone answers neither question.
Advertiser choices are only part of the picture. Platform optimization, campaign objective, creative, destination content, bid and budget decisions, and competing advertisers can all affect delivery. A broad audience setting therefore does not guarantee a demographically balanced result, while a measured gap does not identify which factor produced it.
Run the audit in a documented sequence
1. Define scope and preserve the original setup
Choose the campaign and observation period, then save the settings and reports before making changes. Record the platform, dates, geography, ad category, objective, budget, audience definition, exclusions, placements, creative and destination page. Identify which groups and forms of exclusion matter for this campaign and jurisdiction. Requirements can differ for housing, employment, credit and general consumer advertising.
#1 Best Overall
Keep a dated copy of each configuration and export, along with the report definitions and any later edits. Platform interfaces and reporting options change, so record the labels and settings actually shown for the campaign rather than assuming another account or period had the same options.
2. Define the eligible-audience baseline first
Write down the population that could legitimately receive the ad under the campaign’s actual criteria and was available on the platform during the observation window. The general population is not automatically the right denominator: a job may require specific qualifications, for example, and an offer may have genuine geographic or eligibility limits. At the same time, criteria should not be treated as legitimate merely because they were entered into an ad interface.
For an employment campaign, Imana, Korolova and Heidemann’s 2021 study framed delivery skew against qualified and available platform members in the campaign’s target audience. Use that as a methodological reference, not a universal legal rule. Document how each qualification or availability limit was defined and what evidence supports it.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errors3. Separate advertiser settings from delivery mechanisms
Inventory the possible influences before interpreting outcomes. Include explicit audience selections and exclusions, but also the campaign objective, optimization choices, creative and destination content, bidding and budget, audience availability, and auction competition. Note any inferred-interest or lookalike audience features in use, where those details are visible.
These factors can interact. For instance, an advertiser may choose a broad audience while the platform’s optimization delivers impressions unevenly. Conversely, a difference in reach may reflect unequal audience availability or qualifications rather than a delivery mechanism. The audit should describe observed settings and outcomes separately instead of attributing a gap to a single cause without evidence.
4. Measure outcomes against the baseline
Where the platform provides suitable reporting, compare actual impressions by relevant demographic groups with the corresponding groups in the eligible audience. State the source and definitions of both the impression data and the baseline. If demographic breakdowns are unavailable, incomplete or based on inferred attributes, say so; do not treat missing reporting as evidence that delivery was balanced.
One descriptive measure is each group’s share of impressions compared with its share of the eligible audience. For example, if a group accounts for 30% of the eligible audience and 15% of impressions, its impression-share-to-baseline-share ratio is 0.5. This signals a difference worth examining; it is not a legal threshold, a causal explanation or a measure that fits every campaign. Report the underlying shares and counts alongside any ratio, because a ratio can conceal small sample sizes or an uncertain denominator.
Free tools Windows power users keep installed
One-click scans. No signup required.
Use a measure suited to the question and data. A comparison of impression shares asks whether exposure reflects the eligible audience composition; a comparison of reach or frequency may be more appropriate when the concern is how many distinct people were exposed or how often. Do not compare metrics with different definitions or observation windows as if they measured the same thing.
5. Compare like with like when testing a hypothesis
If the audit seeks to test whether delivery differs beyond obvious campaign differences, use a matched comparison where feasible. Imana, Korolova and Heidemann’s 2021 black-box study ran paired job ads at the same time for jobs with similar qualification requirements but different existing workforce gender distributions. The design aimed to reduce the possibility that differences simply reflected qualifications, while controlling for other factors outside platform delivery.
For a practical comparison, align the campaigns as closely as possible on eligibility, timing, objective, creative, budget, placements, geography and other relevant settings; record every difference that remains. A matched design can make an explanation more testable, but it does not eliminate every confounder or establish a universal legal test. Do not change multiple campaign variables at once and then claim that one change caused the resulting difference.
6. Protect privacy and preserve reproducibility
Keep the analysis limited to data needed to answer the audit question. Document permissions, access controls, retention and any aggregation or suppression used to protect individuals. Do not infer a person’s protected status from a name, image or other sensitive attribute merely to fill gaps in platform reporting.
Preserve the observation window, campaign versions, report exports, definitions, calculations, exclusions, comparison design, data transformations, assumptions and limitations. The EU delegated audit regulation describes evidence as needing to be appropriate, sufficient and reliable and recognizes tests and substantive analysis as parts of an audit methodology. Those principles are useful for reproducibility; the regulation does not prescribe one demographic-parity metric for every advertiser’s campaign.
7. Write a conclusion at the strength the data supports
Separate the finding from its interpretation. State the measured difference, relevant denominator, observation period and data limitations. Then distinguish among a descriptive disparity, a supported hypothesis about a contributing factor, and a legal conclusion. Black-box audits may not expose platform code or user-level data, and platform-provided statistics may not offer enough demographic detail for direct measurement. Missing information should narrow the conclusion, not be silently filled with assumptions.
What a published platform experiment can—and cannot—show
In their 2021 paper, “Auditing for Discrimination in Algorithms Delivering Job Ads,” Imana, Korolova and Heidemann reported statistically significant gender skew in their Facebook job-ad experiment and did not find such skew in their LinkedIn experiment. These are results from the authors’ particular study design, platforms and period; they are not measurements of current platform delivery.
The authors described practical limits for outside auditors, including lack of access to platform algorithms and user data and the limits of available advertiser statistics. They argued for privacy-preserving delivery statistics and audit methods with rigorous privacy guarantees. They also reported that their own study took several months and cost close to $5,000. That is a historical estimate for that project, not a general price for conducting an audit.
Keep legal and transparency questions distinct
United States employment advertising
In testimony to the EEOC on January 31, 2023, witness ReNika Moore described how employment ads may be targeted using personal characteristics, online behavior, inferred interests, location and lookalike audiences. Her testimony explains why restricting direct demographic targeting alone may not address effects from proxies or delivery optimization. It is official testimony and historical context, not a current finding about a particular platform, advertiser or campaign.
Best Value
European Union platform transparency and targeting
The European Commission’s overview of the Digital Services Act says ads must be labeled and very large online platforms must maintain repositories with details about paid campaigns. It also says targeted advertising on online platforms is prohibited where profiling uses special categories of personal data, such as ethnicity, political views or sexual orientation. Application depends on the service and facts; platform transparency duties are not the same thing as a finding that an advertiser’s campaign discriminated.
Commission Delegated Regulation (EU) 2024/436 recognizes advertising systems among algorithmic systems that may be audited under the DSA. Its methodology combines assessment of internal controls, substantive analytical procedures and, where appropriate, system tests. It does not create a universal demographic-parity test for every campaign or replace other applicable anti-discrimination rules.
Before acting on a legal interpretation, confirm the jurisdiction, campaign category, protected classes, platform features and law in force. A voluntary fairness review, a platform’s transparency obligation and an anti-discrimination legal standard are related concerns, but they are not interchangeable.
Quick Recap
Audit record checklist
- Campaign scope, dates, geography, category and objective.
- Audience settings, exclusions, placements, creative, destination, budget and bidding choices as configured during the period.
- Written definition and basis for the eligible, available audience.
- Impression, reach or frequency data, with metric definitions, demographic categories and source.
- Comparison design and the campaign differences it controls for or leaves unresolved.
- Data handling, transformations, uncertainty, limitations and access safeguards.
- A conclusion that separates measured disparity, possible causes and any legal assessment.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
