The Measurement Gap That's Eating Your Strategy

2026-08-08 — marketing audit SME India

I watched a client meeting last week where someone asked what AI's ROI actually was.

Dead silence.

Kevin Indig's latest report on the first half of 2026 opens with a line that lands like a punch: AI's impact kept growing faster than anyone's ability to measure it. Not by a little. By a lot. And that gap is where most brands are getting slaughtered right now, because measurement isn't just a dashboard problem—it's the difference between strategy and guessing.

Here's what actually happened in the first six months of this year. Money moved. Traffic moved. People lost jobs. Market cap evaporated. And almost none of it traced back to proof.

The three-to-four consumer preference shift is real.

Research found that roughly three out of four consumers pick the top result in an AI answer, unless a trusted brand appears elsewhere on that list, then they pick the trusted name instead. That's not theoretical. That's how people actually make decisions now. But here's the problem: 91% of those AI citations appear on just one platform. Single-platform. Which means your rank tracking tool is blind to most of where your traffic should be coming from.

You're optimizing for visibility on something that's already shifting underneath you.

Software stocks fell about 30% over H1 2026, and when you dig into what actually caused the decline, it wasn't earnings misses or failed products. It was sentiment. Companies like Adobe and Salesforce traded at roughly 50% discounts to early 2025 levels because investors believed AI would destroy the SaaS model. No proof yet. Just belief. The stocks that crashed hardest were the ones investors thought were most vulnerable to disruption, not the ones that were actually disrupted.

The market punished narrative instead of performance.

Actually, that's not quite right—investors punished perceived vulnerability, which is a different animal. It's worse. Because if the market is going to tank your valuation based on a fear that hasn't materialized yet, you can't out-earn your way out of it. You have to prove the fear wrong. Good luck doing that when nobody knows what to measure.

Meta burned 73.7 trillion tokens in a single month on an internal leaderboard. An actual leaderboard ranking 85,000 employees by how many tokens they burned. No one could point to what improved because of it. The ROI was invisible. After four months, the CFO realized the annual token budget was gone in one quarter, and they shut the whole thing down.

That's not efficiency.

That's just spending money because the infrastructure exists to spend it. And it's happening everywhere right now because we have tools to do AI work but we don't have tools to measure whether the AI work mattered. We have the acceleration but not the brakes.

Companies blamed AI for 87,000+ job cuts through May. Roughly a fifth of all 2026 layoffs up to that point. But when Indig dug into the story, the pattern was different. The real causes were pandemic over-hiring and capital discipline. AI became the explanation that sounded smart, that sounded inevitable, that didn't require conversation with the board about bad hiring forecasts two years ago.

Blame the algorithm instead.

Here's where this matters for your brand: Everything that happened in H1 2026—the consumer preference shifts, the valuations, the hiring decisions, the token budgets—came down to the same problem. Impact was outrunning measurement. We could see things moving. We couldn't see why.

And when you can't measure why, you stop trusting anything. Not the data. Not the investment. Not the strategy.

The gap between impact and measurement isn't just a dashboard gap. It's an attribution gap. You can see that AI citations matter. You can see that trust is a ranking factor now. You can see that consumers prefer answers from sources they already trust. What you can't see is which of your content appears where, how often, across which platforms, for which queries. Most brands don't even know where 91% of their potential AI citations are happening.

That blindness is the real story.