Which KPIs Replace CTR and CPC When AI Runs the Campaign

The KPIs to replace CTR and CPC when AI controls ad placements and targeting are the ones a platform finds hard to improve just by changing who sees the ad or where it runs: profit per ad dollar, new customer share, incremental return, blended efficiency and creative-level results. Click-through rate and cost per click earned their place when a buyer chose the keyword, the placement and the bid, because each number pointed at a setting the buyer could change. Once the system makes those choices, both become by-products of the inventory it bought that week.

Google's guidance on evaluating Performance Max explains the mechanics. It calls marginal ROI, the additional return on investment you get from additional spend, foundational to the campaign's cross-channel bid optimization, and adds that a channel with a lower average ROI can still carry the highest marginal ROI within a single auction. A campaign bidding this way can buy cheap, rarely clicked video impressions one week and costly search placements the next, so average CTR and CPC can swing with the mix while ROAS and profit hold steady.

Five numbers that hold up under automation

The KPIs to replace CTR and CPC in AI advertising share one property: each is built from an outcome the business books, so a change of placement or audience has far less room to flatter it.

Profit per ad dollar

(revenue - cost of goods - variable costs) / ad spend. Above 1.0, the campaign covered its own cost. Revenue-based ROAS pays the system for selling whatever converts, low-margin lines included. Your order system holds the inputs, and Google documents that conversions with cart data, combined with the cost of goods sold from a Merchant Center feed, lets Google Ads generate additional metrics based on gross profit. Our ROAS calculator is a further resource for setting the break-even bar with your own margin.

New customer share

new customers / all customers who bought in the period, read beside cost per new customer. An optimizer paid on conversions has every reason to find past buyers, who tend to convert more easily. When the new customer acquisition goal is active and the campaign optimizes toward Purchase conversions, Google reports new customers (under its auto-detection, people with no purchase in the last 540 days) and defines its CAC column as "The ad spend allocated to new customers divided by the total of unique new customers acquired through a campaign." The same page says CAC columns are unavailable for Demand Gen at this time. A first-order flag in your own order data covers other platforms, and our explainer on CAC covers the blended and paid versions.

Incremental return

incremental conversion value / total spend, where incremental value is what the exposed group produced beyond a comparable holdout. It is the one number here that addresses whether the sales would have happened anyway, the question behind incrementality. Google describes Conversion Lift as a comparison between a treatment group that sees your ads and a control group that doesn't, with incremental conversion value and incremental ROAS among its metrics. The page opens by saying Conversion Lift isn't available for all Google Ads accounts and directs advertisers to their Google account representative. A geo holdout read against your own sales produces the same ratio.

Blended efficiency

total revenue / total marketing spend, often called MER, paired with blended CAC. The inputs sit with finance: the store or billing system for revenue, and every channel's invoice for spend. It catches the month in which a campaign's reported ROAS climbs while company revenue stays flat.

Creative-level results

conversion value / conversions for each asset, ranked within one asset group or ad set. Creative is one of the few inputs still in the buyer's hands, and Google lets you add conversions, conversion value and value per conversion to the Performance Max asset group, asset and asset associations reports. The page adds a caveat that belongs on the scorecard: "Conversions aren't split across multiple components of an advertisement." If a headline, a description and an image together produce one conversion, each is credited with one, so asset totals exceed the asset group's count. Rank assets against one another and keep the figures out of any sum. The same page says conversion data for some asset types is not available for now, so read a blank as a reporting gap.

Which decisions CTR and CPC still support

Both keep a job as diagnostics, and three decisions still lean on them. The first is comparing creative on equal footing: CTR ranks two assets fairly when audience and inventory are held constant, as inside one asset group or one split test. The second is locating a fault. When conversion rate drops and CTR holds, look at the landing page, the feed or the checkout; when CTR drops and conversion rate holds, suspect creative fatigue or a mix shift. The third is reading auction pressure. Since CPC = CPM / (1,000 x CTR), splitting a CPC change into its two parts shows whether inventory got more expensive or the ads got clicked less. For Performance Max, Google's evaluation page points to a channel distribution table with clicks, conversions and cost, which is where a mix shift shows first.

Targets are a different matter, because which clicks a platform credits with a conversion can change under you. On March 3, 2026, Meta announced: "Going forward, we are changing the definition of click-through attribution for website and in-store conversions to exclusively include link clicks." Conversions that follow a share, a save or another non-link click move into engaged-view attribution, which Meta is renaming engage-through attribution, with the changes beginning later that month for campaigns optimizing for website or in-store conversions. A click-through conversion trend that spans the rollout compares two definitions. The same post states Meta's belief that incrementality experiments such as its Conversion Lift are the gold standard in measurement.

What the IAB says about incrementality, outcomes and attention

IAB measurement guidance on incrementality, media outcomes and attention as advertising metrics includes two documents released in November 2025.

The IAB and IAB Europe released their Guidelines for Incremental Measurement in Commerce Media on November 3, 2025. The report outlines four families of method (experiments, model-based counterfactuals, econometric models and hybrid proxies) and when to use each, and it rests on three principles: credible counterfactuals, control of bias, and separation of signal from noise. It is written for commerce media; we read those principles as a fair test of any incremental return figure on a scorecard, whoever produced it.

For attention, the IAB and the Media Rating Council published their Attention Measurement Guidelines in November 2025, covering four approaches: data signal-based measurement, visual and audio tracking, physiological and neurological observation, and panel- or survey-based methods. The IAB's page is plain about where attention ranks: "Attention is not a standalone measure of ad effectiveness." It calls attention a complementary signal to combine with delivery and outcome metrics. On a scorecard, we place it in the diagnostics column beside CTR, with outcomes above both.

A one-page scorecard by campaign type

In AI era advertising, the KPIs that replace CTR and CPC depend on what the campaign is for, so the scorecard gets one row per campaign type and four columns: the target KPI, the guardrail that stops the system gaming it, the proof you check each quarter, and the diagnostics you watch without setting targets.

Two rows need a note. Google describes AI Max for Search campaigns as a suite of targeting and creative enhancements, including search term matching, text customization of headlines and descriptions, and final URL expansion. Those change the query, the copy and the landing page around a click, so CTR and CPC from before and after describe different ads. Google states its own claim in outcomes: advertisers that activate AI Max will typically see 14% more conversions or conversion value at a similar CPA or ROAS (Google internal data, 2025, non-Retail advertisers). On the Meta row, Meta says its most advanced AI optimizations are enabled when a sales campaign uses Advantage+ audience, Advantage+ campaign budget, Advantage+ placements and Advantage+ destination. With those on, creative and the conversion signal are the inputs left to judge.

A hypothetical month on the scorecard

Every figure here is an assumption, invented to show how the readings can disagree. A store spends $40,000 on Performance Max in each of two months. Month A brings 2,000,000 impressions, 40,000 clicks and $160,000 in reported revenue at a 40% margin after cost of goods and variable costs. Month B brings 5,000,000 impressions, 50,000 clicks and $168,000 at 35%. New customers are 480 of 1,600 buyers, then 420 of 1,680, and company revenue is $520,000, then $515,000, on $100,000 of total marketing spend.

A click-based review stalls on the first two rows: CTR halved, CPC improved by a fifth, and neither says whether the month was good. Reported ROAS rose. The outcome rows agree with one another, though: profit per ad dollar fell from 1.60 to 1.47 as sales tilted toward lower-margin products, new customer share dropped five points, and MER slipped. The campaign found more of the buyers the store already had. A tighter new-customer guardrail and a lift test come before any budget increase, and when spend does go up, our guide to scaling paid ads without killing ROAS applies the same marginal logic in steps.

Set the target on one KPI per row and write each guardrail as a number. CTR and CPC stay on the page, in the last column, where they explain a change without deciding a budget. If you would like a second pair of eyes on your scorecard, our paid media team is a further resource.

Sources

• Google: Evaluate Performance Max Results
• Google: About conversions with cart data
• Google: Measure your lifecycle goals campaigns
• Google: About Conversion Lift
• Google: Conversion reporting by asset in Performance Max campaigns
• Google: About AI Max for Search campaigns
• Meta: Simplifying Ad Measurement for a Social-First World (March 3, 2026)
• Meta: Advantage+ sales campaigns
• IAB: Guidelines for Incremental Measurement in Commerce Media (November 3, 2025)
• IAB: IAB and MRC Attention Measurement Guidelines (November 2025)