Audience signals can help Performance Max learn who is likely to convert. They do not fence the campaign into the audiences you choose.
That distinction changes the job. You are not building an audience-targeted campaign. You are giving Google AI a useful starting point – then judging the campaign by business outcomes, search behavior, creative performance, and conversion quality.
A practical workflow is: define what each asset group is meant to sell, inspect the reusable audiences already available in the account, propose a small signal set, review it, attach it, and monitor the wider asset-group result.
The quick answer
Performance Max audience signals are optional suggestions applied at the asset-group level. Google says they can help guide its machine-learning models, but ads may still show to people outside those signals when Google predicts they are likely to convert.
Use signals to describe the likely early converters for a specific offer – not every person who could conceivably buy.
Start with strong first-party data when it is relevant and eligible.
Use custom segments or Google audience segments to add useful intent and context.
Keep each signal set aligned with the asset group’s product, landing page, and creative.
Treat search themes as a separate input for relevant searches, not as audience signals or keywords.
Review additions and removals before applying them; a plausible audience can still be strategically wrong.
Table of Contents
Audience signal, search theme, or targeting?
These three ideas are easy to blur together:
Audience signal: a suggestion about the people most likely to convert, attached to an asset group.
Search theme: a phrase that helps Performance Max understand searches relevant to the asset group. It is not a keyword and does not restrict reach.
Targeting control: a rule that limits where, when, or to whom ads may serve. Audience signals are not this.
Google explicitly describes audience signals as optional and says Performance Max can serve beyond them. If your requirement is a strict boundary – for example, a regulated location or a business constraint – use an actual campaign control where one exists. Do not rely on a signal.
Choose signals from the asset group outward
A good signal plan begins with the asset group, not the audience menu.
Write one sentence that defines the asset group’s job:
“This asset group sells [offer] to [buyer] through [landing page], using creative about [specific problem or outcome].”
If that sentence is vague, the signal set will usually become a kitchen sink. If one asset group mixes several offers or audiences, fix the structure before adding more signals.
A useful order of operations
1. Start with first-party data when it matches the offer
Website visitors, app users, customer lists, past converters, and video viewers can provide Google with direct evidence of who has already interacted with the business. The strongest list is not always the largest one; it is the one most closely related to the conversion you want.
Do not automatically attach “all visitors” to every asset group. A high-intent product-viewer list may be more useful for one product line, while an existing-customer list could be appropriate for upsell – or inappropriate if the goal is new-customer acquisition.
2. Add intent and context deliberately
Google supports custom segments built from relevant search terms, URLs, or apps, plus additional segments such as in-market, affinity, life-event, and detailed-demographic groups.
Use these to sharpen the hypothesis, not inflate the list. For a B2B workflow tool, a search-based custom segment around the problem being solved may provide a better starting cue than a broad “business professionals” audience.
3. Match the signal set to the creative and landing page
Signals, assets, and the landing page should tell the same story. If the signal implies enterprise buyers but the creative speaks to freelancers, the asset group is giving the system conflicting clues.
This is also why copying one audience set across every asset group is usually weak practice. Reuse an audience only when the underlying buying context is genuinely shared.
A review-first AI workflow
AI is useful here because it can inventory account resources, compare them with the asset-group brief, and turn a messy audience list into a reviewable plan. The assistant should not jump directly from “find audiences” to “apply them.”
Step 1: Inspect the asset group
Ask the assistant to summarize the offer, final URL, existing assets, current signals, and any search themes. Resolve structural mismatches before editing signals.
Prompt:
Review PMax asset group [ASSET_GROUP_ID]. Summarize its offer, landing page, assets, current audience signals, and search themes.
Identify inconsistencies, but do not make changes.Step 2: Inventory available audiences
List the reusable Google Ads Audience resources already available in the account. Names and descriptions help, but they are not proof that a list is populated, eligible, fresh, or strategically appropriate.
Prompt:
List the reusable audiences available in account [CUSTOMER_ID].
Group them into first-party, custom-intent, and broader Google audience themes where the names and descriptions allow. Flag anything ambiguous.
Do not attach audiences yet.Step 3: Propose a small signal plan
Give the assistant the asset-group job, acquisition goal, exclusions or sensitivities, and any first-party data constraints. Require a rationale for every proposed audience.
Prompt:
For asset group [ASSET_GROUP_ID], propose a concise audience-signal plan for [OFFER].
Prioritize signals closest to conversion intent.
For each audience, explain why it belongs, what could make it misleading, and whether it is already available in the account.
Do not apply changes.Step 4: Approve exact audience IDs
Review the proposed audience names and IDs. Check that first-party lists reflect the right users and that using them aligns with consent, policy, and the campaign’s acquisition objective. Remove generic filler.
Step 5: Apply and verify
Prompt:
Add the following approved audience IDs to PMax asset group [ASSET_GROUP_ID]: [AUDIENCE_IDS].
Then list the asset group’s signals again and confirm exactly what was added.
Do not remove or alter any other signal.When removing a signal, follow the same pattern: list current signals, select the exact audience-signal resource, obtain human approval, remove only that resource, and list the signals again.
Step 6: Evaluate the asset group – not an imaginary audience report
Audience signals help the system learn, but they are not a clean reporting dimension that tells you the incremental value of each signal. Judge the asset group through outcomes such as conversions, conversion value, CPA or ROAS, search-term quality, network mix, asset performance, and lead quality.
Google’s current help documentation says new signals can take up to two weeks to be fully integrated by its models. That is not a universal waiting rule: account volume, conversion lag, budget, and the size or freshness of audience lists still matter. Avoid making repeated signal changes faster than the account can produce useful evidence.
Common mistakes
Treating a signal as hard targeting. Performance Max can serve beyond it.
Using one catch-all audience for every asset group, regardless of the offer.
Adding broad segments simply to make the signal set look comprehensive.
Assuming a descriptive audience name guarantees list quality or eligibility.
Confusing audience signals with search themes, keywords, exclusions, or demographics reporting.
Changing signals while simultaneously changing budget, bidding, assets, and conversion goals – making the result impossible to interpret.
Removing a signal because there is no per-signal conversion report. That report does not exist in the way the premise assumes.
How HireOtto helps with this workflow
HireOtto turns the review-first sequence into a conversation inside an MCP-compatible AI assistant. For Performance Max audience signals, it can:
List reusable Google Ads Audience resources already available in the account.
List the audience signals and search themes currently attached to a specific asset group.
Add approved audience IDs to an existing asset group or include them when creating a standard PMax campaign or new asset group.
Remove specific existing audience signals after you identify their exact signal resources.
Manage search themes separately and pull related PMax reports for the wider performance review.
The human review gate still matters. HireOtto does not decide whether an audience matches your acquisition strategy, prove that a first-party list is compliant or high quality, or turn a signal into hard targeting. Its current workflow reuses audiences already defined in Google Ads; it does not create a new reusable audience definition for you. Google also does not provide clean performance attribution for each individual audience signal.
Use the HireOtto Performance Max guide for the end-to-end workflow, the Google Ads MCP tools reference for current actions, and the HireOtto quickstart to connect your AI assistant.
FAQs
Are audience signals required for Performance Max?
No. Google describes them as optional. They can provide a useful starting point, especially when you have relevant first-party data or a clear audience hypothesis.
Do audience signals restrict who sees my ads?
No. Google may show ads to relevant users outside your signals when it predicts they are likely to convert.
Should every asset group use different signals?
Not automatically. Different offers and buyer contexts usually deserve different signal plans, but an audience can be reused when it genuinely applies to more than one asset group.
Are search themes the same as audience signals?
No. Audience signals describe likely people; search themes describe relevant search intent. Both are asset-group signals, but they are managed as different inputs.
Can I see conversions for each audience signal?
Not as a clean, causal per-signal report. Evaluate asset-group and campaign outcomes, then use search-term, network, asset, and lead-quality evidence to refine the broader setup.
Should I remove old signals?
Remove a signal when the underlying hypothesis is wrong, the offer has changed, or the audience is no longer appropriate – not merely because an individual signal lacks its own performance column. Inventory and approve the exact removal first.
About Me
I’m Suyash – badminton junkie, ex‑GroupM ad‑ops grunt, first marketer at a B2B SaaS startup, and creator of Hiretto: Google Ads MCP Server.
My mission: less clicking, more thinking.
Let’s build leverage together.