Documentation
Audiences Audit gives you two things in one view: a performance analysis that shows which segments are already working and which ones are draining budget, and an AI suggestion engine that reads your top-performing ad copy and keywords to propose audiences worth adding or excluding. Use it before any significant targeting or budget change to make sure your audience strategy is grounded in data, not guesswork.
- Best for
- Audience performance review, coverage gaps, and AI-powered targeting ideas
- Core analysis
- Spend vs. CPA or spend vs. ROAS quadrant chart with detailed sortable table
- Winner signals
- Top-left for CPA efficiency (high spend, low CPA); top-right for ROAS strength
- Suggestion source
- Your ad copy and keywords interpreted by AI
- Action posture
- Review and prioritise before changing live targeting
#What Audiences Audit does
Audiences Audit is Adnine's Audience Insights workflow. It identifies which audiences are driving efficient results, which ones are costing money without delivering, and which new segments are worth testing or excluding. The right audience strategy can dramatically reduce your CPA or boost your ROAS, but managing audience coverage across multiple campaigns is hard without a clear view of performance.
Questions it answers
Which audiences are working?
Compare audience spend with CPA or ROAS to find segments that outperform the account average and deserve more coverage.
Which audiences need review?
Surface segments with meaningful spend but weak efficiency, low return, or results that do not match the campaign goal.
Where is coverage missing?
Check whether useful audiences are applied to the right campaigns or ad groups, or are absent from campaigns where they should appear.
What should you test next?
Review AI-generated include and exclude ideas based on your ad copy and keywords, and decide which fit the business.
#Who it is for
Audiences Audit is most useful when you are running campaigns across multiple audience types, managing a mix of prospecting and remarketing goals, or preparing for a budget change and want to know which segments to protect, scale, or cut.
When to run the audit
| Situation | What to look for |
|---|---|
| Routine optimisation review | Confirm CPA and ROAS winners are still performing, catch any high-spend segments that have softened. |
| Before a budget increase | Identify which audiences have the efficiency and conversion volume to absorb more spend safely. |
| After an ad copy or keyword change | Use Refresh Suggestions to get updated AI ideas based on the new ad copy and keywords, then check whether existing audience performance has shifted. |
| Coverage gap review | Check whether strong-performing audiences from one campaign type or region are missing from related campaigns. |
| New product or seasonal campaign | Run the suggestion workflow to find audience ideas matched to your updated ad copy and keywords. |
#How the audit works
- Select the account scope and date range Choose a window with enough spend and conversions to make CPA or ROAS comparisons meaningful. Extend the range when audience volume is low or the account is newer.
- Read the performance quadrant Toggle between CPA and ROAS depending on the campaign goal. Each audience segment is plotted by its spend and its performance relative to the account average. Winners appear in the top-left (CPA) or top-right (ROAS) of the chart.
- Open the detailed table Sort and filter the underlying rows to confirm spend, impressions, clicks, conversions, conversion value, CPA, ROAS, and the applied campaign or ad group scope for each segment.
- Check setup context Review audience type, inclusion or exclusion role, targeting mode, list freshness, and whether the segment fits the funnel stage before treating chart position as a final verdict.
- Classify each segment Assign an action category to each meaningful audience: scale, test, fix setup, monitor, reduce, or exclude. Avoid changing live targeting from chart position alone.
- Review AI suggestions Read the explanation for each include or exclude idea. Compare suggestions with business context, the actual offer, and customer profile before promoting any to a test.
- Verify after changes After audience, bid, or budget changes have collected enough data, return to the same CPA or ROAS view and confirm performance moved in the intended direction.
#Audience types and signals reference
The audit reviews all eligible audience types present in the account. Because remarketing lists, customer lists, Google in-market segments, and affinity segments behave differently, it is important to understand what each type represents before comparing performance across rows.
Audience types and what to expect from each
Availability varies by campaign type and account configuration. Performance expectations should match each audience's role in the funnel.
| Audience type | What it represents | Performance context |
|---|---|---|
| In-market segment | Google-defined users actively researching a product or service category. | Useful for prospecting; compare against other prospecting segments, not remarketing lists. |
| Affinity segment | Google-defined users with a long-term interest or lifestyle pattern. | Typically broader reach; may show higher volume but lower conversion intent than in-market. |
| Website remarketing list | Users who visited specific pages of your site within a defined window. | Higher intent; judge against past-visitor and page-specific benchmarks, not cold traffic. |
| Customer match list | Users matched from your CRM or email data. | Often used for retention, upsell, or exclusion; freshness and match rate directly affect performance. |
| Combined segment | A composite of two or more audience criteria joined by AND, OR, or NOT logic. | Inspect constituent parts if performance is confusing; overlap or conflict between layers is a common cause. |
| Custom segment | Built from search query patterns, app usage, or URL-based intent signals. | Closer to intent-based targeting; useful for bridging prospecting and in-market gaps. |
| Similar audiences / Optimised targeting | Google-expanded reach based on existing list characteristics; behaviour varies by campaign type. | Treat as an extension of the source list; check whether reach expansion is helping or diluting results. |
Key data fields the audit surfaces
These fields appear in the detailed table and inform every quadrant position.
| Field | What it tells you | How to use it |
|---|---|---|
| Audience name and type | The exact segment under review and its source category. | Confirm it matches the intended targeting before treating performance as actionable. |
| Applied scope | The campaign or ad group where the audience is active. | Decide whether an action belongs at campaign level, ad group level, or across multiple campaigns. |
| Inclusion or exclusion role | Whether the audience is used to target, observe, or exclude traffic. | Avoid scaling an audience used only as an exclusion; confirm exclusions are intentional. |
| Spend and volume | Cost, impressions, clicks, conversions, and conversion value. | Use minimum-volume thresholds before trusting CPA or ROAS readings. |
| CPA and ROAS vs. average | How the segment compares with the account average for the selected metric. | Use relative performance to identify outliers worth scaling, testing, or monitoring. |
| List freshness and size | Estimated audience size and recency of membership updates. | A stale or very small list can underperform regardless of historical strength. Fix it before judging performance. |
#How to read your results
The quadrant chart places each audience segment based on its spend and its performance relative to the account average for the selected metric. Toggle between CPA and ROAS to match the chart to the campaign's actual goal.
Quadrant reading guide
CPA winners are top-left (high spend, low CPA). ROAS winners are top-right (high spend, high ROAS). Use the table to turn chart position into a concrete next step.
| Quadrant position | What it signals | Typical next step |
|---|---|---|
| High spend, lower CPA than average | This is the strongest CPA pattern. The segment is spending meaningfully and achieving a much lower CPA than the account average. | Review coverage and consider broader testing, higher priority bid, or budget protection where conversion quality supports it. |
| High spend, higher ROAS than average | This is the strongest ROAS pattern. The segment combines high spend with strong return. | Consider scaling or protecting the audience in value-focused campaigns if margin and conversion value quality hold up. |
| High spend, weak CPA or ROAS | The audience is using enough budget to matter but is underperforming the selected average. | Inspect campaign goal, targeting mode, list quality, audience overlap, and landing-page fit before reducing or excluding. |
| Low spend, strong CPA or ROAS | The audience looks promising but may not have enough volume to trust the signal. | Monitor, extend the date range, or test in a controlled scope before scaling. |
| Low spend, weak CPA or ROAS | The signal is too small to prioritise, or the audience may not fit the current campaign structure. | Keep on a watchlist unless the segment is clearly irrelevant or conflicts with the campaign goal. |
#Fixing audience issues
Audience performance problems are not always caused by the audience itself. Often the issue is a setup problem: the segment is applied at the wrong level, excluded by accident, too small to serve, stale, or mixed with incompatible campaign goals. Before changing bids or budget, check setup first.
Common issues and how to fix them
| Issue | Likely cause | Fix |
|---|---|---|
| High spend, poor CPA, no obvious explanation | Audience may be too broad, the campaign goal may not match the segment type, or landing-page fit is weak. | Narrow the audience, confirm the campaign type supports the intended use, and review landing-page relevance. |
| Strong efficiency but very low volume | Audience list is too small, stale, or restricted by overlap with exclusions. | Refresh the list source, extend the membership window, or check for accidental exclusions at a higher level. |
| Audience shows no impressions | List is below the minimum eligible size, excluded at campaign or ad group level, or not supported in the campaign type. | Check eligibility for the campaign type, verify exclusion layers, and rebuild the list if membership has dropped. |
| Conflicting include and exclude rules | A user could simultaneously match an included and an excluded audience, often from combined or layered segments. | Audit include and exclude logic together; simplify or reorder layers to resolve the conflict. |
| Customer list underperforming | Match rate may be low, the list may be outdated, or the targeting goal may have changed. | Re-upload a fresh list, confirm the audience goal (retention vs. exclusion vs. lookalike), and validate match rate in Google Ads. |
| Remarketing list results inconsistent | Tracking may have gaps, the membership window may be too short or too long, or the page visited does not match purchase intent. | Verify tag coverage, adjust the membership duration, and segment lists by page type or funnel stage. |
#AI audience suggestions
The suggestion workflow uses AI to recommend audiences based on your account's own signals. Adnine sends your top-performing ad copy and keywords to the model, which proposes a list of audiences to include and exclude from a built-in Google Ads audience taxonomy. Each suggestion comes with an explanation so you can assess relevance before taking action.
How to evaluate suggestions
| Suggestion type | What it can mean | How to evaluate it |
|---|---|---|
| Include audience | Your ad copy and keywords suggest a customer interest, intent, industry, or use case that may fit the offer. | Check whether the audience describes a real buyer profile, has enough reachable volume, and fits the campaign's funnel stage. |
| Exclude audience | Your ad copy and keywords point to traffic that is likely irrelevant, too broad, or unlikely to convert. | Confirm the exclusion will not block useful prospects, existing customers, or high-value research behavior before applying it. |
| Adjacent or niche suggestion | The AI may surface a related but not identical segment that could reveal a useful test idea. | Use observation mode or a controlled experiment before treating it as a core target. |
| Weak or off-topic suggestion | Some suggestions may not fit the business if page copy is broad, outdated, or ambiguous. | Skip it, refine your ad copy or keywords, or use Refresh Suggestions after those change. |
#Pitfalls and limits
Audiences Audit is a decision-support tool. It helps you prioritise audience work, but it does not replace account strategy, privacy requirements, data quality checks, or a final review before changing live targeting.
Limits to keep in mind
| Limit | Why it matters | How to handle it |
|---|---|---|
| Low-volume segments can be noisy | A small number of clicks or conversions can make an audience look unusually strong or weak. | Set minimum-volume thresholds, use longer date ranges, or assign a monitor status before acting. |
| Averages depend on selected scope | The chart compares each audience with the average in the selected view. Mixing campaign goals inflates or deflates the benchmark. | Segment reviews by goal (prospecting, remarketing, retention) to keep comparisons meaningful. |
| Audience overlap can blur attribution | Users often belong to multiple segments simultaneously, so one audience may receive credit for conversions driven by another. | Treat audience performance as directional and investigate overlap before assigning sole credit to a single segment. |
| AI suggestions require human review | Suggestions are generated from your ad copy and keywords and may include ideas that are too broad, irrelevant, or risky for the account. | Read each explanation and apply human judgment before adding or excluding an audience based on a suggestion. |
| Campaign type restrictions apply | Audience availability, reporting, and targeting behaviour can vary by campaign type. Not every audience action is supported everywhere. | Confirm the action is supported in the specific campaign type before applying it. |
| Privacy and policy still apply | Customer lists, remarketing, sensitive categories, and personalised advertising rules may limit how audiences can be used. | Follow your legal, privacy, and platform policy process before changing audience collection or targeting. |
#Best practices
- Toggle between CPA and ROAS based on the campaign's actual goal before deciding whether an audience is strong or weak.
- Use the detailed table to confirm spend, conversions, and applied scope behind every chart position before acting.
- Separate prospecting, remarketing, customer, retention, and exclusion audiences during review. They have different performance baselines.
- Check audience setup (type, role, level, list freshness) before assuming poor performance is caused by the audience itself.
- Treat high-spend, strong-performing segments as scale candidates, not as automatic budget increases. Verify conversion quality first.
- Treat low-volume, high-efficiency segments as test ideas until they collect enough traffic to trust.
- Use Refresh Suggestions after major ad copy, keyword, offer, product, or geography changes.
- Reject AI suggestions that do not match the real buyer profile, service area, policy constraints, or campaign goal.
- Document each audience action with the metric pattern, date range, affected campaign or ad group, and expected outcome.
- Return to the same CPA or ROAS view after changes to confirm performance moved in the intended direction.
#Frequently asked questions
What is the difference between CPA mode and ROAS mode in the quadrant chart?
CPA mode plots segments by spend and cost per conversion. Winners appear top-left (high spend, low CPA). ROAS mode plots by spend and return on ad spend. Winners appear top-right (high spend, high ROAS). Switch to the mode that matches the campaign's billing goal. For conversion-volume campaigns, CPA is usually the right lens; for value-based or revenue-focused campaigns, use ROAS.
Can I trust an audience that looks great on the chart but has very few conversions?
Not without more data. A small number of conversions can produce a very strong CPA or ROAS reading by chance. Before acting, extend the date range, check absolute spend, and look for at least enough conversions to represent a pattern rather than a one-off. Assign a monitor status until volume grows.
How do AI suggestions work and where does the content come from?
Adnine sends your top-performing ad copy and keywords to an AI model, which generates audience ideas to include and to exclude from a built-in Google Ads audience list based on those signals. Each suggestion includes an explanation. Use Refresh Suggestions after significant content updates to get recommendations based on the latest version of the page.
What does Refresh Suggestions do?
It re-runs the AI on your latest top ad copy and keywords and generates a new set of include and exclude audience ideas. Run it after you update your ad copy or keywords, launch a new product, change your target region, or run a seasonal campaign with different messaging.
Will changing an audience in Adnine automatically apply the change in Google Ads?
The Audiences Audit is a review and prioritisation tool. It surfaces performance data and AI ideas but does not automatically apply changes to live campaigns. Review findings here, then make the targeting adjustment directly in Google Ads or through the relevant Adnine workflow.
Why does an audience show in the quadrant but not in the detailed table, or vice versa?
The chart may require minimum spend or conversion data to plot a point, while the table can show audiences with partial data. If you see a discrepancy, check the date range, any applied filters, and whether the audience has enough impressions to appear in both views.
How should I handle audiences that appear in multiple campaigns with different results?
The audit aggregates data at the scope you select. If the same audience appears in both a Search and a Display campaign, results may be combined. Narrow the scope to a single campaign type or use the detailed table filters to isolate the view before making a campaign-specific decision.
Is there a minimum list size required for an audience to be included in the audit?
Google Ads enforces minimum audience eligibility thresholds for delivery and reporting, which vary by campaign type and list type. Very small or newly created lists may appear in the table but show no impressions or spend until they reach the required threshold. Treat these as setup items rather than performance signals.