I read about Google’s Search Console data scandal and felt that very specific sensation familiar to anyone who has spent serious time in this industry: not shock, not outrage, but a kind of weary, knowing recognition. The feeling you get when something you suspected for years is finally confirmed in writing.

A 47-word changelog entry, buried on a data anomalies page that nobody reads, quietly admitted that Google had been over-reporting search impressions for every website on the planet for fifty consecutive weeks. No email. No announcement. No apology. The SEO community spotted it before Google said a word.

I’ve seen this before. We all have. The platforms that built empires on marketers’ trust have a long and distinguished history of grading their own homework, then handing it back with a gold star. We kept accepting the grade. We called it being data-driven. We put it in board presentations. We made hiring decisions based on it. Agencies were fired over it.

Which is precisely why, in our PR for Robots series — where we track the shift from traditional search to AI-powered visibility — we keep returning to the same point: the digital metrics religion has been asking you to worship a god with a well-documented habit of making things up.

What Actually Happened

Between May 13, 2025 and April 27, 2026 — fifty weeks — Google Search Console was systematically over-reporting impressions in its Performance report. Every website. Every market. The entire tool, broken, for nearly a year.

Clicks were unaffected. Impressions were inflated. Which means every CTR calculation your team produced during that period used a wrong denominator, making your click-through performance look worse than it was. Every visibility trend line showing “growing impressions” may have been measuring the growing bug, not growing reach. Every board report, every agency presentation, every strategy document built on Search Console impressions was, to some degree, built on fiction. And given that DemandScience’s research across 750 senior marketing leaders found that 25% of the average marketing budget is already being spent on efforts that look productive in metrics but don’t drive outcomes, the Google bug didn’t create the waste. It just added to it.

Google has since fixed the logging error going forward. What it will not do is reconstruct the historical data. Fifty weeks of corrupted numbers are now permanent. They are the foundation of decisions already made, budgets already spent, strategies already locked in.

The disclosure, in full: “A logging error is preventing Search Console from accurately reporting impressions from May 13, 2025 onward.” Forty-seven words. For eleven months of bad data affecting every marketer on earth. You’re welcome.

This Is Not An Isolated Incident

If you are tempted to treat this as a one-off technical embarrassment, consider what else has happened recently and quietly.

On January 26, 2026, Meta deprecated all 10-second video metrics, a benchmark thousands of brands and agencies had used for years to evaluate video performance, with no direct replacement announced. Years of comparative data, rendered useless overnight. Whatever your video strategy was benchmarked against no longer exists as a measuring stick.

Instagram, across 2025 and into 2026, switched its primary metric from Impressions to Views, redefining what the number on your dashboard actually counts. Historical comparisons are now broken by design. Every trend line you drew crosses a definitional cliff edge you may not have noticed.

And underneath all of it, a structural rot that never left: $63 billion in global digital ad spend was wasted last year on invalid traffic — bots, click farms, automated scraping and non-human interactions that will never become a customer. That figure comes from Lunio’s 2026 Global Invalid Traffic Report, built on analysis of 2.7 billion paid clicks. If you think your website traffic at least tells you something real, NP Digital’s research across 602 tracked websites found that 51% of traffic came from bots and 21% were sessions too short to register anything meaningful. Sixteen percent, one in six visits, could be classified as genuinely engaged.

Your dashboard is full. Your funnel is largely empty.

The platforms are not being malicious. They are being something almost more dangerous: indifferent. Metrics change because products change. Errors get fixed when someone eventually notices. The marketing budgets built on top of those metrics are not their problem. They are yours.

“Data-Driven” Became A Religion. 

Here is the real indictment, and I say this as someone who has spent over four decades watching this industry evolve: we let the word “data-driven” do the work that thinking used to do.

It became a shield. It ended conversations. It silenced instinct and experience in rooms where instinct and experience were exactly what was needed. If the numbers said it, it was true. If the platform reported it, it was real. We outsourced our judgment to dashboards and called it rigor. Today, 75% of marketers say measurement is broken. That’s not a fringe view. The Interactive Advertising Bureau, the industry trade organization that sets standards and best practices for digital advertising, included it in their State of Data 2026 report. And yet, 66% of senior marketing leaders simultaneously admit their campaigns frequently look successful in metrics while failing to drive revenue. We built a system sophisticated enough to produce the illusion of performance, and then we congratulated ourselves for reading it.

Nobody asked whose data it was, how it was counted, or what commercial incentive the platform had to make the numbers look healthy. Nobody asked because asking felt unscientific. Unmodern. The kind of thing someone who didn’t understand digital would say.

The Google bug didn’t create this problem. It just made it impossible to ignore for one more news cycle, before everyone updates their anomaly annotations in GA4 and moves on.

The Pivot That Changes Everything

There is, buried in all of this, something that resembles good news, if you’re positioned to see it.

AI search doesn’t care about your impression count. ChatGPT, Claude, Perplexity, Google AI Mode and their expanding cohort don’t surface answers because your Search Console metrics looked healthy last quarter. They synthesize, they cite, they answer. The currency in AI search is authority, demonstrated expertise, genuine usefulness and earned trust — qualities that are considerably harder to inflate and considerably harder for a platform to accidentally break for fifty weeks without anyone noticing.

As we’ve been covering in PR for Robots, the brands that will win in AI search are the ones investing in being genuinely worth citing — not the ones optimizing for a metric that may or may not be accurately reported on any given Tuesday.

In a perverse way, the collapse of confidence in traditional digital measurement is the most compelling argument we have for taking AI search seriously now, before the same gaming, the same bot inflation and the same self-serving metric redefinitions arrive there too. Because they will. They always do.

What You Should Do This Week

Stop using impressions as a proxy for anything in a board conversation. They have been inflated, redefined and gamed too many times to carry the weight you’ve been placing on them.

Audit every performance target that was set using data from May 2025 to April 2026. You were measuring with a broken instrument. The targets need resetting.

Triangulate everything. Never rely on a single platform’s self-reported data. Cross-reference Search Console with Google Analytics 4, with third-party tools, with actual revenue. If the only source confirming your success is the platform you’re paying, that is not confirmation. That is a conflict of interest.

And start understanding AI search visibility now — before its measurement frameworks get captured, gamed and quietly disclosed in 47-word changelog entries too.

Finally

The platforms will keep changing their metrics, retiring benchmarks and disclosing year-long errors in language designed to be found by no one. That’s the deal. That’s always been the deal. We just chose not to read it.

The marketers who survive the next decade won’t be the ones with the cleanest dashboards. They’ll be the ones who never confused the map for the territory — and who understood early that in AI search, the territory just changed completely.

Sources: Google Search Console Data Anomalies Page, confirmed April 3, 2026; Search Engine Land, May 5, 2026l Lunio 2026 Global Invalid Traffic Report; DemandScience: The 2026 State of Performance Marketing, 750 senior leaders, October 2025; IAB State of Data 2026; NP Digital website traffic analysis, 602 properties, 2026; Meta Business Help Centre, January 26, 2026

Back in June I wrote about Brandformance—the not-so-ugly lovechild of brand and performance marketing. That piece argued for integration: emotional resonance plus measurable impact. But integration is only half the battle. The other half is recognizing that one of the most celebrated metrics in performance land—ROAS—might be the least interesting number you track.

I’ve sat in too many meetings where ROAS was toasted like gospel, champagne corks flying because a campaign “paid for itself” in six weeks while brand awareness, trust and repeat purchase scores quietly slid into oblivion. Despite the ubiquity of promotion and discounting that props up most performance marketing, I’ve never wavered on one principle: discounts erode brands. It’s a race to the bottom. Depending on your brand, sector and market, that race may be a marathon or a sprint.  But you are definitely headed to the same finish line.

In the old bricks-and-mortar world, “location, location, location” could save you. You could sell the same thing as the shop five blocks away, but if you were closer to a bus stop or sitting on the corner where the crowd passed, you won—sometimes even charging a bit more. In the online world there is no corner. If you’re a retailer without a unique proposition, you’re competing with everyone in your market. Price comparisons are instantaneous. You end up spending more to chase fewer sales. And soon, when AI shopping agents go mainstream, performance ads will matter even less. Those agents will leapfrog your offers and simply hunt down the cheapest, fastest, most trustworthy option. And trust in that scenario comes from brand equity. Legacy brands will have the edge—and they’re already exploiting it.

Why ROAS Is Seductive—and Misleading

On the surface, ROAS looks like the neatest metric ever invented. Spend a dollar, make five back—what could be clearer? It explains why marketers over-index on it and why finance teams love it. But ROAS measures only efficiency at a point in time; it ignores momentum.

Samsung Electronics, a company we worked with for over 20 years, proves the danger of myopia. In Interbrand’s 2024 Best Global Brands ranking Samsung’s brand value hit US $100.8 billion, up 10% year on year. That growth didn’t come from performance coupons—it came from positioning itself as a leader in AI and connected experiences which build trust and long-term preference. A short-term ROAS chart could never explain that kind of billion-dollar lift.

The Trace “72% Growth Advantage” study backs this up: brands that balance long-term building with short-term performance saw about 72% growth in brand value over five years while brands chasing only efficiency eked out 20%. It’s the difference between compounding returns and eking out margin.

The Real Risks of Chasing ROAS Alone

Marketers addicted to ROAS often find themselves discounting into oblivion. Coca-Cola is a case study in avoiding that trap. In Interbrand’s 2024 report Coca-Cola still sat in the Top 10 with a brand value of US $61.2 billion. Its secret isn’t efficiency metrics—it’s cultural saturation and trust which allow it to maintain pricing power while competitors rely on coupons.

There’s also the false comfort of attribution. Just because a conversion shows up after an ad impression doesn’t mean the ad caused it. Interbrand estimates that Best Global Brands have collectively left US $3.5 trillion in potential value unrealized by leaning too hard on short-term tactics. In just the past year alone that figure was US $200 billion. That’s what misplaced attribution bias costs.

And then there’s diminishing returns. Retarget audiences until their eyes glaze over and you’ll get declining ROAS and rising irritation. Ferrari illustrates the opposite path. In 2024 its brand value grew 21%—the steepest gain among global brands. That wasn’t the result of squeezing more retargeting out of lookalike audiences; it was Ferrari expanding its brand arena into fashion, lifestyle and experiences, deepening emotional equity so customers lean in willingly.

What Matters More Than ROAS

If ROAS is the sugar rush, brand equity is the long-term nutrition. Consider e.l.f. Beauty. From 2020 to 2024 its unaided awareness in the U.S. climbed from 13% to 3%, a twenty-point leap that translated into market share gains and revenue growth. They didn’t get there by hacking last-click ROAS—they invested roughly 22–24% of sales back into marketing and storytelling that people remembered.

Toyota is another example. In 2024 its brand value reached US $72.8 billion, up 13% year on year. Not bad for a 90-year-old automaker. Toyota’s consistency in quality and innovation means customers trust it which keeps margins healthy even when the competition discounts heavily. That resilience isn’t captured in a ROAS dashboard—it lives in brand perception.

And trust itself is quantifiable. Edelman’s Trust Barometer shows 81% of consumers say they must trust a brand to buy from it. That’s not a metric you’ll see in a performance report but it’s one of the most predictive numbers in marketing.

Winning the Budget Battle in the C-Suite

Of course, the pushback is always: “We need sales now.” But the right counter is: “We’ll need them later too—and at a lower cost per acquisition.” The Trace report is a powerful slide in that debate. A balanced 60/40 split between brand and performance doesn’t just protect long-term equity—it delivers higher total growth.

Legacy brands can point to compounding effects. Microsoft and Google each posted double-digit brand value growth in Interbrand’s 2024 table (+11% and +12% respectively) despite already being giants. That’s what happens when brand strength amplifies performance instead of leaving it to do all the work alone.

Challenger brands can point to cases like e.l.f., where spending aggressively on brand drove measurable equity gains and repeat purchase. The lesson is the same: ROAS can’t be your north star. At best it’s a mile marker.

So, What Should You Do?

Audit what you’re measuring. If your reports begin and end with ROAS you’re starving yourself of context. Add unaided awareness, repeat purchase, retention and trust.

Rebalance budgets. Not everyone will shift to 60/40 overnight but even moving 10 points away from performance-only will give you more compounding returns.

And tell better stories. Ferrari isn’t just selling cars; Samsung isn’t just selling screens. They’re embedding themselves in culture, technology and aspiration. That’s why they can grow brand value at double-digit rates while competitors chase click-through rates.

The Final Word

ROAS is too neat to ignore but too narrow to revere. AI shopping agents will soon strip the emotion out of buying decisions and prioritize trust, convenience and price the only brands that thrive will be the ones people already believe in.

The brands that survive aren’t the ones with the cheapest clicks—they’re the ones people would miss if the clicks disappeared.

Sources: Trace Brand Building, Interbrand Best Global Brands 2024б, Samsung Electronics press release, e.l.f. Beauty (WARC), Edelman Trust Barometer, Ferrari Interbrand 2024 report, Toyota Interbrand 2024 report