Microsoft Advertising's August product package opens four control areas at once. AI Visibility in Clarity now makes topics and query clusters easier to inspect. Performance Max gains two experiment types. Ad Preview Hub makes it easier to review and share creative and Bing result previews before launch. Audience Ads also adds excluded content terms that can block pages whose titles contain unwanted words or phrases.
Treating these as isolated menu changes would miss the larger operating shift. The package brings AI-search visibility, incrementality testing, creative approval, and brand-safety boundaries into one control loop. Microsoft also shares aggregate performance claims for advertising inside Copilot. Those figures are platform claims, not guaranteed outcomes for every account.
For advertisers in Türkiye, availability may vary by product, account, and market. The newsletter is a broad product update, but individual capabilities can remain subject to phased rollout. The sound response is to verify access, then connect each feature to an existing measurement and approval process.
What changed across the four areas?
The first area is Clarity AI Visibility. Topic Insights and Query Topics help teams see which topic clusters are associated with a brand and which grounding queries use its pages as sources. Microsoft brings metrics such as citations, citation share, grounding queries, and share of authority into one reporting layer.
This is not a replacement for web analytics. When an AI system cites a page, the user may never click through. Citation growth is therefore not the same as traffic growth. It can still identify content that is repeatedly used as evidence and reveal gaps in the site's expertise portfolio.
The second area is Performance Max experiments. An Upgrade experiment is designed to compare moving an existing Search or Shopping campaign into Performance Max. An Uplift experiment asks whether adding Performance Max to the current campaign mix creates incremental conversions. Microsoft recommends conditions including at least 30 conversions in the previous 30 days, aligned goals and products, and a test lasting four to twelve weeks.
The third area is Ad Preview Hub. It adds Performance Max creative previews, shareable links, and a Bing search result view. Media, design, and client teams can review one preview instead of exchanging screenshots. A preview cannot guarantee every live-auction placement and context, but it standardizes pre-launch creative and brand review.
The fourth area is excluded content terms for Audience Ads. An advertiser can define words and phrases it does not want to appear in page titles, at account or campaign level. Search Engine Land reports that the list supports up to 1,000 terms and is rolling out gradually. The control complements site exclusions and IAS content categories.
The four modules represent AI visibility, experimentation, creative preview, and content suitability connected to one campaign centre. It is a Fark Studio illustration, not a reproduction of the Microsoft interface.
How should AI visibility be measured?
Separate at least three layers. The first is citation: is a URL used as a source? The second is topic share: how often is the brand visible in a relevant cluster? The third is business impact: does that visibility correspond with branded search, direct visits, forms, or sales conversations?
A frequently cited source may not produce a strong brand mention in the answer. A brand may also be named while the link points elsewhere. Citation, mention, and traffic therefore need separate reporting. AI Visibility can begin a content diagnosis, but it should not be the only business case for investment.
Evaluate topic changes over a meaningful period rather than reacting to weekly noise. A new content series, PR campaign, or technical change may take time to be discovered. Conventional organic results and referring sources should be recorded in the same period.
The critical distinction in Performance Max testing
Upgrade and Uplift do not answer the same question. Upgrade tests a move from one campaign type to another. Uplift tests the incremental contribution of adding Performance Max to existing activity. Teams should write the decision question before selecting the experiment.
Microsoft says eligible tests have shown an average 8% incremental conversion lift from adding Performance Max. This is Microsoft's aggregate product claim. Category, conversion definition, budget, existing brand demand, and measurement quality vary, so the figure should not become a forecast for a local account.
A sound experiment keeps conversion goals, geography, product scope, and evaluation windows aligned across test and control. Large budget or creative changes during the run make the result difficult to interpret. The 30-conversion threshold is not a promise of statistical certainty. It is a minimum signal recommendation for the learning system.
What do excluded terms solve?
Excluded content terms help catch recurring words associated with sensitive news or content. A brand can build lists around crises, death, violence, political conflict, or category-specific risks. The feature cannot understand every synonym, image, body-copy context, or emerging event.
An overly broad list may unnecessarily reduce reach. A word can appear in a safe context as well as a risky one. Terms belong inside a layered brand-safety strategy with site exclusions, content categories, placement reporting, incident review, and human judgment.
Central account-level management improves consistency, but campaign tolerance can differ. Children's products, finance, health, and corporate reputation campaigns may need tighter campaign-level controls. Keep a change log and measure whether a list also removes valuable inventory.
Eight steps for teams in Türkiye
1. Verify feature access in the account. Do not interpret a product-newsletter date as universal same-day availability.
2. Build an AI visibility topic dictionary. Separate brand, category, problem, solution, and competitor clusters. Do not merge citations and mentions into one field.
3. Attach a business metric. Read Clarity visibility beside branded search, direct traffic, qualified forms, and sales conversations.
4. Write the experiment question. Replace “Is PMax better?” with a testable question such as whether it creates incremental qualified leads under the same budget and goals.
5. Clean the conversion definition. Duplicate, low-value, or unqualified events can mislead automation tests. Audit UET, revenue, and offline lead-quality imports first.
6. Assign an owner for Preview Hub approval. Record who approved the shareable preview and when.
7. Manage brand-safety terms by risk level. Keep global exclusions, campaign terms, and temporary crisis lists separate. Give each change a review or expiry date.
8. Report all four controls in one loop. Visibility should inform content, experiment results should inform budget, preview should control launch, and suitability should define the boundary.
The illustration shows diagnosis, experiment, pre-launch control, and feedback as a continuing operating loop. A decision in one module should not invalidate the others without evidence.
How should Copilot ad claims be treated?
Microsoft reports a 53% higher conversion rate and 25% higher relevance for ads in AI experiences. These are aggregate internal platform claims. The account-level method, sample, and Türkiye breakdown are not public, so the numbers should not be inserted directly into a media forecast.
Performance Max, standard Search, and Shopping campaigns may be eligible for Copilot placements. The newsletter also mentions a Search AI Max pilot. Advertisers may not need a separate campaign for every surface, but the ability to interpret queries, creative, and conversion quality by surface remains an important reporting uncertainty.
Fark Studio interpretation: the package matters less because it adds more automation buttons and more because it adds places to test and constrain automation. A mature team treats AI visibility as a content signal, experiments as incrementality evidence, previews as a launch gate, and excluded terms as a live risk inventory.
If you want to connect Microsoft Advertising across performance marketing, measurement and analytics, and content production, contact Fark Studio.
Sources
Microsoft Advertising, Product Newsletter August 4, 2026, August 4, 2026. Official product newsletter covering AI Visibility, Performance Max experiments, Ad Preview Hub, and Copilot advertising.
Microsoft Clarity, Analyze AI citation queries by topic, July 22, 2026. Official product source describing Query Topics and topic-level citation analysis.
Microsoft Advertising Help, Excluded content terms, live help page as of August 5, 2026. Official documentation for Audience Ads content-suitability controls.
Search Engine Journal, Microsoft Advertising adds AI visibility insights, PMax testing, and creative preview updates, August 5, 2026. Secondary source consolidating the official update for marketers.
Search Engine Land, Microsoft Advertising adds excluded content terms for Audience Ads, August 5, 2026. Secondary report on the brand-safety rollout.
