Google Ads is now showing AI-powered conversion recommendations

AI-powered conversion recommendations are automated suggestions made by Google's artificial intelligence systems, designed to enhance the accuracy and effectiveness of conversion tracking in Google Ads. These recommendations help advertisers optimize their conversion setup, ultimately boosting the precision of Smart Bidding and other AI-driven bidding strategies.

Google Ads has introduced AI-generated recommendations directly inside the conversion summary dashboard. The suggestions are labelled "Recommended by Google AI" and are designed to help you improve conversion measurement in your account. This is separate from the existing Recommendations tab. The feature was spotted by Arpan Banerjee and shared publicly by Adriaan Dekker on LinkedIn, before being picked up by Search Engine Roundtable in July 2026. The key point for any advertiser: these suggestions affect your conversion tracking setup, which directly feeds Smart Bidding. Review them carefully before applying anything.

Why Google Ads conversion tracking is the base of every campaign

Smart Bidding depends on signal quality. Google's automated bidding strategies  Maximize Conversions, Target CPA, Target ROAS  are machine learning models built on clean measurement data. When that data is broken, incomplete, or misconfigured, every bidding decision the algorithm makes is built on distorted input. It does not know something is wrong. It just optimises toward whatever signal it receives.

This is not a niche technical concern. Without conversion tracking, you are bidding blind  Smart Bidding has no signal to optimise toward, ROAS calculations are guesswork, and there is no way to understand which campaigns, ad groups, or keywords are actually generating leads and revenue.

The new AI recommendations in the conversion summary are significant precisely because of this. Google is not adding another layer of campaign advice. It is intervening at the measurement layer  where the data quality problems that limit performance actually start. For e-commerce brands especially, that distinction matters. A shopping feed may be fully optimised, bidding strategies may be correctly configured, but if the conversion setup has gaps, the whole system underperforms.

How Google Ads conversion tracking works

To evaluate what the AI is recommending, it helps to understand the mechanism it is examining.

When someone clicks on your ad, the Google Tag  formerly the global site tag  fires and stores a click identifier (gclid) in a first-party cookie in the user's browser. This captures critical information: the gclid, timestamp, campaign details, and device data. When that user then completes a conversion action  submits a form, makes a purchase, calls your business  a second piece of code fires and sends a conversion signal back to Google Ads, connecting the original ad click to the completed action.

Google's servers match that signal back to the specific campaign, ad group, keyword, and audience segment that drove it. The result is attribution data that tells Smart Bidding which inputs actually lead to outcomes.

You set up and manage conversion actions inside your Google Ads account under Goals > Conversions. Each action requires a name, a category, a conversion value, and a count type. Those choices are not cosmetic. An account tracking "every" instance of a lead form submission will inflate numbers and send the algorithm the wrong signal about what is generating new customers. Choose "one" for lead actions. Choose "every" for purchase events where repeat buying is the goal.

How to set up Google Ads conversion tracking: the core steps

Setting up Google Ads conversion tracking correctly requires several decisions in sequence. Getting any one wrong corrupts the data downstream.

Step 1: Define what counts as a conversion

Start with the goal, not the tag. What action does your business actually need users to complete? A purchase, a form submission, a phone call, an app install, a key page visit? Each type requires a different setup. Select the type that matches your business model before you configure anything else.

Give each conversion action a clear, descriptive name. "Purchase completed" or "Demo request submitted" are useful. "Conversion 1" is not. You will need to find and manage these actions later, and vague labels make that harder.

Set the conversion value. For e-commerce, use dynamic values so Google receives actual transaction amounts, not a flat number. This is what enables Target ROAS to work. If you set a static value or leave it at the default, bidding strategies that depend on value data will generate unreliable results.

Step 2: Choose your implementation method and add the tag

Google Tag Manager is the right method for almost all setups. It lets you add and configure your tracking tag without editing site code each time, provides a central place to manage all tags across your site, and makes debugging significantly easier.

Inside Google Tag Manager, add a Google Ads Conversion tag. Provide the Conversion ID and Conversion Label from your Google Ads account  both are required. Add the correct trigger so the tag fires only on the confirmation or thank-you page, not on every page of the site. A tag that fires on every page will inflate conversion numbers and send bad signals.

If you do not want to use Google Tag Manager, you can install the Google Tag directly in the code of your site. Place the global tag in the <head> section of every page, and the event snippet on the specific page where the conversion action is completed. This method works, but it is harder to maintain and audit.

Step 3: Enable Enhanced Conversions

Enhanced conversions capture first-party data that customers voluntarily provide  such as an email address or phone number during checkout  hash it using the SHA-256 algorithm, and send the hashed data to Google. Google then matches this hashed data against its own logged-in user data to connect the conversion to the ad click that drove it.

This matters because browser restrictions, ad blockers, and cross-device journeys all create gaps in standard cookie-based measurement. Enhanced conversions help recover those gaps using data your customers have already given you. Advertisers typically see a 5–30% increase in reported conversions after implementing Enhanced Conversions, with improvements in Smart Bidding performance following as the algorithm sees a more complete picture of actual outcomes.

Enable Enhanced Conversions in your conversion action settings. You can set this up automatically via the Google Tag if your confirmation page already displays user data, or manually via Google Tag Manager if you need to specify which fields contain the relevant information.

Step 4: Check the setup with Google Tag Assistant

Do not assume the tag is working because you installed it. Use Google Tag Assistant to verify that your tracking tag fires correctly on the right page and does not fire on pages where it should not. Check this in your live site environment, not just in preview mode. The conversion column in your Google Ads account should populate within 24–48 hours of a verified test conversion. If it does not, there is an issue to find and fix before running live campaigns.

What "Recommended by Google AI" in the conversion summary means

Google Ads has introduced AI-powered recommendations in the conversion summary dashboard, labelled "Recommended by Google AI," to help improve conversion measurements. The suggestions aim to enhance performance, but users should carefully review them before implementation.

This is a new location for AI-generated guidance. The existing Recommendations tab has covered bidding strategies, keywords, ads, and assets for years. The suggestions now appearing in the conversion summary are scoped specifically to measurement quality — not campaign structure.

In practice, these recommendations might flag a conversion action that appears inactive, identify a tracking tag that is not shooting off consistently, suggest you enable Enhanced Conversions on an action that does not yet have them, recommend switching to a data-driven attribution model, or note a secondary conversion action that is being counted as a primary goal and distorting bidding signals.

The exact suggestions depend on what Google detects in your specific Google Ads account. The AI is pattern-matching across account data to identify configurations that differ from what typically produces better measurement quality. That is genuinely useful. It is not a directive.

How to review and act on these recommendations

Just be cautious when using these recommendations. A human should always review them.

The AI does not know your business model, your sales cycle, or the deliberate choices you made when you first configured your setup. It knows patterns across accounts. When your configuration diverges from those patterns, it generates a suggestion. Sometimes that divergence is a mistake worth fixing. Sometimes it is intentional.

A practical review process:

Understand what is being proposed. Read the full description before touching anything. Some recommendations are straightforward  adding a tag to a conversion action that is missing one. Others are structural and carry downstream consequences for your bidding data.

Verify the underlying issue. Use Google Tag Assistant to confirm whether a tag is actually misfiring before assuming the diagnosis is correct. Cross-reference the conversion data in your account with what you know about real business results. If the numbers in your report roughly match what your business processes show, the flagged issue may not be as serious as it appears.

Consider timing. Changing conversion actions or attribution models mid-campaign resets Smart Bidding's learning period. If your campaigns are performing well, implement measurement changes at the start of a new period, not in the middle of a live flight. Good housekeeping at the wrong moment can cost you weeks of accumulated learning.

Save your changes before exiting. Google's interface does not always auto-save configuration changes in this section. Confirm each change is saved before navigating away.

Note what you dismiss and why. If you dismiss a recommendation, record the reason. This creates a record you can follow up on later and helps distinguish deliberate setup decisions from things you simply forgot to action.

Attribution: the area most likely to be flagged

One area where AI recommendations in the conversion summary are likely to surface issues is attribution model selection.

Google's data-driven attribution model assigns credit based on real user paths rather than arbitrary rules like last click. User journeys are longer and more fragmented. Multiple touchpoints influence conversion outcomes. Automated bidding depends on accurate attribution signals.

Data-driven attribution is now the default model in Google Ads. Accounts still running last-click attribution are providing Smart Bidding with an incomplete picture of which touchpoints actually influence purchases. The algorithm optimises toward what it can measure. If early-funnel clicks never receive attribution credit, the bidding strategies built to improve ROAS will systematically undervalue the channels and keywords that showcase awareness and consideration.

If the AI flags your attribution model as an area to improve, get a grasp of what the change will do to your reported numbers before applying it. Switching models does not change your actual results  it changes how credit is distributed across the path to conversion. Your headline conversion volume may appear to shift. Make sure you have a grasp of the before-and-after comparison before drawing conclusions.

Offline conversions and what the AI may detect

For some accounts, the AI recommendations will point to offline conversions as a gap. These imports can now be automated via API, and AI can even estimate conversion value when no base data is returned  for example, for store visits, phone calls, or form fills without thank-you bubbles.

When someone clicks on your ad but completes their purchase or conversion offline  by phone, in a store, or through a sales process that happens outside your website  that signal never reaches Google Ads through standard tag-based measurement. The result is that Smart Bidding only sees a fraction of the outcomes your campaigns are actually making.

Offline conversion imports close that gap. You capture the gclid at the point of the click, store it alongside the lead or customer record in your CRM, and upload the completed conversion data back to Google Ads once the sale or outcome is confirmed. This requires additional setup — the gclid needs to be captured and stored at the time of the initial click but the improvement in bidding signal quality is significant for any business where conversion journeys extend beyond the website.

If your Google Ads account shows a recommendation related to offline conversions in the conversion summary, it means Google has detected that some of your conversion types may not be fully tracked. That is worth investigating before dismissing.

The broader direction: AI moving into measurement

This feature is part of a clear pattern. Google has been extending AI from campaign optimisation into the measurement and data quality layer. AI reporting tools in Google Ads now work together to create a cohesive data-driven strategy: Report Generator translates complex questions into targeted data; custom dashboards consolidate that data into a single view; and an agentic capability turns those insights into action, suggesting and executing optimisations directly.

The conversion summary recommendations sit at the measurement end of that stack. The intent is not to change what campaigns do. It is to improve the accuracy of what the account knows.

That distinction matters because measurement improvements compound. Better conversion data improves bidding signals. Better bidding signals improve campaign performance. Better campaign performance generates more and richer data for the algorithm to learn from. A single tracking configuration fix, applied at the right time, can have a larger impact on long-term results than weeks of bid adjustments.

Whether any specific recommendation in your conversion summary is worth applying depends entirely on what it is proposing and whether it reflects a real gap in your setup. The label "Recommended by Google AI" is a prompt for review, not a reason to click apply.

The right approach is the same one that applies to every automated suggestion in Google Ads: read it, verify it against your own data, understand what implementing it will change, and then decide. The AI can identify patterns at scale. You understand your business, your customers, and your goals. Neither of those things should operate without the other.

‍