Microsoft Ads Creative: Responsive Search Ads and Asset Review

#01What should an asset set communicate?

Begin with the buyer's decision and the evidence the advertiser can support. A useful set might communicate product type, compatibility, installation method, delivery conditions and a clear next step. Fifteen slight rewrites of the same slogan do not provide fifteen distinct reasons to choose the offer.

Create a claim ledger before writing. For each proposed statement, record the evidence, relevant products or markets, owner and expiry condition. A shipping promise may depend on stock, order time and location. A compatibility claim may apply only to a particular variant. The ad system cannot infer these qualifications reliably from a brand tone guide.

Separate enduring product facts from temporary promotions. An evergreen claim can remain valid while a discount expires. If both live in the same reusable asset set, establish who removes or updates the time-sensitive material. Avoid relying on an operator remembering a promotion's end date after it has been copied across campaigns.

The Google responsive-search lesson covers transferable copy principles. Microsoft's limits, generated-asset behaviour and current AI Max controls must be reviewed independently rather than assumed identical.

#02How do you write compatible combinations?

Microsoft documents support for up to 15 headlines and four descriptions. These are capacity limits, not a requirement to fill every slot with weak or repetitive text. Prioritise distinct, accurate assets that can coexist without contradiction.

Read possible pairs and groups aloud. Two headlines that each begin with the same phrase can sound awkward together. A broad delivery promise can conflict with a market-limited offer. A description referring to “this discount” may make no sense when its corresponding headline is absent.

Keep essential qualifications attached to the claim they qualify. Do not spread a strong claim into one asset and its only limitation into another optional asset. A combination that omits the limitation may become misleading even if the complete spreadsheet looked balanced.

#03When is pinning appropriate?

Pinning restricts which positions can use an asset and reduces available combinations. It can serve a genuine business or review requirement, but it is not a universal performance tactic. Record why the restriction is needed rather than pinning everything to recreate a fixed ad by habit.

Microsoft cautions that not every headline or description position is always shown. Therefore, a required disclosure cannot depend on an optional slot appearing. If a message is only truthful when a particular description is visible, redesign the claim and consult the relevant policy or legal reviewer.

Review current automated-text behaviour as well. AI Max text customisation and URL expansion can change the practical creative system around an ad. Do not assume a pinning procedure copied from another platform establishes the behaviour of every Microsoft generated asset or eligible format.

When evaluating a pinned variant, define the business question. A change in ad strength is not enough to decide whether the restriction is commercially justified. Consider qualified outcomes, valid claim presentation and customer understanding, while recognising that observational asset reports do not isolate a clean causal effect.

#04What should you know about generated text?

Microsoft's responsive-search documentation describes autogenerated assets sourced from landing pages. Its August 2026 announcement places the existing text capability under AI Max text customisation without automatically opting older advertisers into every broader AI Max component.

This distinction matters in inherited accounts. An operator may see generated text and assume broad search-term matching or URL expansion is also enabled, or may see AI Max disabled at one level and assume no generated copy exists anywhere. Inspect the actual components and asset sources.

Generated text inherits the quality of its source material. Old promotions, unsupported superlatives, outdated delivery terms and unrelated page content can become creative risk. Fix the website source and review the automation controls rather than repeatedly removing symptoms while leaving the underlying material unchanged.

Use Microsoft's asset-source and combination reporting where available to distinguish advertiser-supplied from generated material. Keep a review record of problematic claims, the source page and the proposed response. Deleting or changing an asset can affect multiple associations, so inspect scope before an authorised operator acts.

#05How should extensions and other assets be governed?

Additional ad assets can provide useful routes or information, but support and serving vary by asset type, market and campaign. Review the currently available asset types rather than promising that every source-platform extension imports or appears identically.

Choose assets by customer need. A sitelink should lead to a useful destination with a distinct purpose, not repeat the same page under several labels. A promotion should match actual terms. A call or location-related asset should reflect staffed operations and genuine business presence where applicable.

Inspect association scope. Account-level assets can be inherited by campaigns with different markets, products or offers. A delivery statement valid for one collection can become misleading when attached to another. A complete creative review includes account, campaign and ad-group associations, not just the visible ad editor.

Track asset ownership and expiry alongside the claim ledger. Product, finance and legal reviewers may own different facts. The campaign operator should know which evidence can be reused and when renewed approval is required. This reduces both stale claims and unnecessary reapproval of unchanged material.

#06What does editorial approval establish?

Microsoft reviews ad text, assets, keywords and related components for policy compliance, and review can continue after ads begin serving. A previously approved item can later receive a different status. Monitor delivery and policy notices rather than treating first approval as permanent clearance.

Approval does not establish that a claim is commercially wise, fully supported in every business context or legally sufficient for every jurisdiction. The advertiser remains responsible for the offer and customer experience. Maintain internal review even when the platform accepts the asset.

When an item is disapproved, read the specific reason and affected scope. Check language, destination, claim evidence and associated assets before resubmitting. Do not evade a policy decision by disguising words, changing to an unrelated domain or creating duplicate accounts.

Prepare a concise support or appeal package when appropriate: affected identifiers, exact status, relevant policy, corrected evidence and date of change. Avoid including customer records or unnecessary personal data. A controlled explanation is more useful than repeatedly making speculative edits without a record.

#07How should you assess creative performance?

Separate diagnostic quality, observed association and causal evidence. Ad strength can suggest completeness or diversity, but it is not a profit score. An asset associated with stronger results may have served in different contexts or combinations, so its report is not a simple independent A/B test.

Choose a primary business metric and supporting diagnostics before reviewing. For a storefront, purchase value on a consistent basis may matter more than click-through rate. For office enquiries, qualification can matter more than raw form volume. A high-click message can be costly if it attracts the wrong interpretation.

Use controlled experiments where eligible and feasible. Otherwise, label before-and-after comparisons as observational and record concurrent changes in bids, targeting, seasonality and landing pages. Avoid claiming a headline caused an outcome when the entire campaign changed at the same time.

#08How should a creative review package be assembled?

Give reviewers the claim ledger, source pages, proposed assets and examples of possible combinations together. A legal or product reviewer cannot reliably approve a fragment without knowing the product, market and conditions it may accompany. Mark supplied and generated material separately and identify any inherited association that broadens use beyond the original campaign.

Record approval at the level of the actual claim and context rather than calling an entire folder approved forever. When the offer, destination or market changes, the operator should know which evidence remains valid and which decision must return to the responsible reviewer. This makes reuse faster without treating old approval as unlimited permission.

#09Worked example: Ledgerline

Ledgerline drafts assets for a cable tray. One illustrative headline says fits every desk, another promotes free delivery, and a description contains the only condition limiting delivery to qualifying orders. The initial set sounds attractive when read in spreadsheet order but fails when assets are combined differently.

The product reviewer replaces the universal compatibility claim with an accurate product description and a prompt to check dimensions. The offer reviewer ensures the delivery message remains truthful without relying on an optional description. The marketer also finds an old promotion on the source page that could inform generated text.

Walkthrough

1. Build a claim ledger with product, market, evidence and expiry conditions.
2. Draft distinct assets around identity, useful differences, constraints and next action.
3. Read combinations for repetition, contradiction and missing qualifications.
4. Review pins only where a documented requirement justifies them.
5. Inspect generated-text settings, asset sources and associated landing pages.
6. Check inherited assets and current editorial status at every relevant scope.
7. Plan evaluation around valid purchases or qualified enquiries, not ad strength alone.

The example illustrates review quality. It does not invent an improvement in click-through rate, conversion rate or platform approval time.

#10Common mistakes

Filling every available slot with near-duplicate copy can reduce useful diversity. Placing a critical condition in a description that might not appear can make an otherwise accurate offer misleading. Both problems arise when the asset spreadsheet is treated as one fixed advert.

Another mistake is blaming the generator while leaving inaccurate source pages unchanged. Finally, do not use approval or a favourable diagnostic label as proof of commercial effectiveness. Policy, truthfulness, customer understanding and performance are related but separate review questions.

#11Try it yourself

Create an offline asset set for one offer and attach a claim ledger. Include at least one product fact, one verified distinction, one buyer constraint and one clear action. Do not publish it as part of the exercise.

Ask a colleague to assemble awkward combinations and remove any optional description containing a qualification. Identify which versions remain truthful. Then review the intended source page for stale claims and write a generated-text acceptance checklist. Your output should explain what evidence makes the set safe, not merely list creative alternatives.

#12Quick quiz

1. Do all responsive-search text positions always appear?

Answer: No. Important qualifications should not depend on an optional headline or description slot being shown.

2. Does ad strength prove that a creative set is profitable?

Answer: No. It is a diagnostic, while business performance requires valid outcomes and appropriate comparison.

3. Does existing generated text mean every AI Max feature was intentionally enabled?

Answer: No. Microsoft distinguishes the existing text capability from broader AI Max matching and URL-expansion choices.

#13What should the creative handover contain?

Provide the asset set, claim evidence, association scope, automation settings, expiry owners and evaluation plan. A reviewer should be able to distinguish supplied text, generated text and the business facts supporting both.

Previous: Microsoft Ads Structure: Locations, Language and Distribution · Next: Microsoft Ads Landing Pages: Relevance, Quality and Policy QA

Or return to the Microsoft Ads for Marketers track.

Related: Microsoft Ads Match Types: Search Terms and Negative Controls · Microsoft Ads Landing Pages: Relevance, Quality and Policy QA · Microsoft Ads AI Max and Copilot: Governed Paid Discovery

#14Common questions

How many responsive-search assets does Microsoft support? The documentation checked supports up to 15 headlines and four descriptions. Capacity is not a reason to fill slots with repetitive or unsupported copy.

Can a required condition live only in an optional description? That is unsafe. Not every text position always appears, so an important qualification should not depend on an optional asset being shown.

Does generated text prove all AI Max features are enabled? No. Microsoft distinguishes existing autogenerated text under text customisation from broader AI Max matching and URL-expansion choices. Inspect actual component settings.