If you are still treating ROAS as the single metric that defines PPC success, you are working with an incomplete picture. Not because ROAS is irrelevant, but because the AI algorithms inside Google Ads now decide who sees your ad, on which channel, and at what moment. The data you get back only partially reflects what actually happened.
A recent article on Search Engine Journal puts this shift into context and proposes a measurement framework built for 2026. We have been testing several of these approaches with our clients and can confirm: the transition from mechanical metrics to business metrics is no longer optional.
What changed in Google Ads auctions
With AI Max, Google moved beyond advertiser-defined keywords. The system now finds contextually relevant queries you never explicitly targeted. A keyword like “trail running shoes” might trigger impressions for “best shoes for rocky terrain running” or “ultra marathon gear.” This means the search terms report is no longer a controllable list but a continuous flow of queries the algorithm deemed relevant.
Performance Max took this logic even further by distributing budgets simultaneously across Search, YouTube, Display, Discover, Gmail, and Maps. Channel reporting, launched in April 2025, finally provides visibility into budget allocation per channel. But it also means keyword-level attribution has become insufficient. When a user sees a YouTube ad, searches for the brand on Google, and buys on mobile two days later, no last-click attribution model captures reality.
Add the fact that nearly 60% of Google searches end without a click, according to a SparkToro/Datos study. AI Overviews, shopping modules, and answers generated directly in the SERP reduce organic visibility and change user behavior. In this context, you need to look at PPC performance from a completely different angle.
A 4-layer framework for PPC measurement
In our work with clients, we reached a similar conclusion: measurement needs to be layered. No single KPI tells the full story. Here are the four levels we recommend:
1. Profitability over ROAS. ROAS tells you how much revenue each dollar generates, but not whether you made a profit. A 5x ROAS on a product with 10% margin is worth less than a 3x ROAS on a product with 50% margin. Include cost of goods sold (COGS) in your calculations and measure contribution margin at the product or category level. For lead generation, ROAS does not apply directly. Track lead qualification rate, sales acceptance rate, and close rate. A low CPL means nothing if the leads never become customers.
2. Incrementality testing. The key question: are your campaigns creating new demand or just capturing existing demand? This distinction is critical, especially for Performance Max, which tends to gravitate toward users already at the end of their journey. Practical methods include geo holdout tests (pause campaigns in one geographic area and compare results), Google incrementality tests (now with a minimum investment of 5,000 USD, significantly more accessible than before), and branded search suppression tests. We ran geo holdout tests for two e-commerce clients this year and the results were revealing: one discovered that 40% of Performance Max conversions were cannibalization from organic.
3. Blended CAC. The formula is simple: total acquisition spend divided by total new customers. It seems elementary, but few advertisers calculate this correctly. Why does it matter? Because when you only look at paid CAC, you ignore the halo effect that campaigns have on organic searches and vice versa. YouTube brand awareness campaigns can reduce Search CAC by 15-20% over the medium term, but if you measure each channel in isolation, you never see this connection.
4. First-party data quality. Segment conversions by type: new vs. existing customers, high vs. low margin, single purchase vs. repeat buyer. Integrate offline conversions, CRM revenue, and lifetime value indicators. Without this data, Google's algorithm optimizes for volume, not real value. We have seen accounts generating hundreds of conversions per month where half were existing customers who would have purchased anyway. When they segmented and fed quality data to the algorithm, CPA on new customers dropped by 25%.
What this looks like in practice
The four layers above are not theoretical. They represent a practical shift in how PPC teams should report and optimize. The good news is that none of these require new tools. Everything can be done with Google Ads, GA4, and a well-structured CRM integration. The hard part is changing the habits and expectations around reporting.
At the operational level, we recommend a few immediate changes. First: extend your attribution window to 60-90 days. The purchase journey has lengthened significantly, and a 7 or 30-day window misses conversions that matter. In B2B or for high-value products, the decision cycle easily exceeds 60 days.
Second: analyze performance by intent clusters, not individual keywords. The “Search terms insights” report in Google Ads groups queries into search categories. This gives you a much clearer picture than looking at each query individually. Instead of optimizing keyword by keyword, you optimize on intent segments.
Third: change the conversation with management. Instead of “we had 5x ROAS this month,” present: “Blended acquisition cost dropped 12%, contribution margin from paid grew 8%, and our incrementality test shows 70% of conversions are incremental.” It is a massive difference in perspective and in data confidence.
Why this matters for the Romanian market
The shift from keyword-centric to outcome-centric measurement is not a Silicon Valley trend that will arrive here in two years. It is happening right now, because the same Google Ads algorithms run everywhere. A Performance Max campaign in Bucharest uses the same AI bidding as one in Berlin or Chicago. The difference is that many Romanian advertisers have not yet adapted their measurement stack to match.
Many advertisers in Romania still operate with dashboards built around CPC and ROAS. This is not a criticism but an observation: these metrics were sufficient when you had full control over keywords and placements. But in 2026, with ads embedded in AI conversations and fully automated auctions, measurement needs to evolve.
The good news: the tools already exist. Channel reporting in Performance Max, incrementality tests with lower thresholds, conversion lag reports in GA4. What is usually missing is not the technology but the mental framework. The shift from “how much did I spend and earn” to “how profitable and incremental is every dollar invested” is the change that matters most right now.
