AI Marketing in 2026: Why Proving ROI Is Getting Harder

A little while ago, a friend of mine — a Head of Marketing at a SaaS company — was venting to me.

He said the company had poured nearly two million into AI tooling this year. Content output had jumped fivefold. Email reach, personalized cold outreach, landing-page variants — all running on autopilot. The CEO was pleased at first; figured they had it locked down.

Then came the quarterly review. In front of the entire executive team, the CFO asked him one question:

"Did this AI investment actually bring in any revenue?"

He froze for three seconds.

It wasn't that there was no data. There was plenty. The dashboard was wall-to-wall charts — reach, click-through rate, time on page — every metric looked beautiful. But on the single most basic question — did we make money? — he had no answer.

It's not that he couldn't do his job. It's that the answer can't be measured.

An uncomfortable truth

One survey aimed at executives asked a simple question: can you confidently measure your company's return on AI investment?

Only 29% said yes.

Another report aimed at marketers was even harsher: between 60% and 75% admitted their measurement methods aren't up to standard.

In other words, out of ten marketing directors, seven are flying blind on ROI.

Why? Because the "ruler" we used to measure with no longer matches the shape of things today.

Your ruler was built for last era's buyer

What do I mean by "last era's buyer"?

I mean the buyer from ten years ago — searched a keyword, clicked your ad, browsed your site, filled in a form, bought. A straight line, every step visible to you.

Last-click attribution was designed for exactly that: whoever clicked the last link gets the credit.

Simple. Clear. Easy to account for.

But think about it — in 2026, how do you buy things?

You ask an AI assistant, "Which CRM is right for a mid-sized SaaS company?" and it hands you a synthesized answer. You read it, you've got a sense of it — and you never clicked into any official site.

A colleague drops a white paper into the company Slack channel. Everyone debates it for half a day. Nobody goes to the registration page to leave an email.

What you see in the search engine isn't ten blue links anymore — it's a synthesized AI answer at the top, mentioning three brand names. You remember them. But the search engine's click tracker doesn't remember you.

The influence really happened. The purchase decision really was shaped. But on your dashboard, none of it ever occurred.

Three invisible thieves

Let me break it down for you, and you'll see why it can't be measured.

The first thief: zero-click search.

Today, more than half of all searches end without a single click.

The reader gets their question answered inside the AI Overview and leaves with the answer. If your brand appears in that summary, that's tangible influence — but the attribution system can't see it, because there is no "click" to record.

This hurts B2B teams doing brand and thought leadership the most. Decision-makers often remember your name first, then accept your sales call at some later moment. And the credit for "being remembered" — quietly erased.

The second thief: dark social.

Research says more than 80% of B2B content sharing happens in private channels you can't see — Slack, forwarded emails, WhatsApp, DMs.

Example: a CTO at a company in Singapore drops your industry report into an executive group chat. Two weeks later, a VP in that group raises a demo request. Where do you think your CRM attributes that conversion?

"Direct traffic." Or "source: (none)".

The report that actually moved the deal circulated inside that group and never appears anywhere you can track.

In Asia-Pacific, this gets worse. Because business here runs on relationships, on introductions, on private trust. One recommendation inside a WhatsApp thread is worth ten cold emails. And that WhatsApp thread is exactly what you can't see.

The third thief: the content deluge.

AI tools are genuinely useful. The same team that used to write four blog posts a week can now write twenty.

But when your content output jumps fivefold, one question scales exponentially: which of those twenty posts actually moved which deal?

The signal stays the same. The noise multiplied fivefold. The signal-to-noise ratio collapses.

One study says companies using AI see ROI 10% to 20% higher on average. Sounds decent. But the same study also says 70% to 80% of AI projects last year failed to deliver the expected return.

The tools aren't broken. It's the way we measure them that can't keep up with what they produce.

So what are smart marketers doing now?

If the old ruler can't measure the new cloth, swap the ruler.

Not one new ruler that covers everything — combine three.

The first: Marketing Mix Modeling (MMM).

This is a holdover from the ad agencies of the last century, and it's hot again. Why?

Because it doesn't chase individual clicks. It looks at the macro trend — where you spent your money across channels over the past six months, and how business outcomes shifted. Offline events, brand ads, all those actions that don't generate trackable clicks — it can fold them in.

Suited for B2B companies fighting across multiple markets and channels in parallel. It captures your "invisible influence" — at least at the macro level.

The second: influence-based attribution.

Stop fixating on "the last click." Instead, look at what actually happens across a group of accounts — how long they spent with your content, how many times they came back, whether their deal cycle was shorter than accounts that never touched your content.

Put plainly, it doesn't measure "who clicked last." It measures "did your content actually change the behavior of this group of people."

That's how B2B buyers really buy: not a one-click impulse, but a loop of reading, comparing, discussing — and only then, a decision.

The third: self-reported attribution.

This one is the most underrated.

How? Just add one honest line to your demo request form or your first sales call: "How did you first hear about us?"

Sounds low-tech. But it's quietly saving you.

If, over a quarter, ten prospects tell you they were referred by a colleague in a Slack channel, or heard someone mention you at an industry meet-up — that's hard evidence of your "dark traffic." No tool can track it. But they told you, in their own words.

Each of the three rulers has blind spots on its own. Stack them together, and what you see starts to approach the truth.

The Asia-Pacific game is harder

Everything I've covered so far is a global problem. But drop it into Asia-Pacific, and you add another layer of complexity.

Let me walk you through three markets.

Singapore. The B2B tech and SaaS companies here are usually backed by a regional HQ or a global HQ holding the budget. The HQ wants one number — ROI. But Singapore's local data environment makes that number extremely hard to compute accurately. Explaining why it can't be computed accurately is harder than producing an accurate number in the first place.

Australia. Privacy legislation here saw further tightening this year, narrowing what first-party data can be collected, used, and stored. Layer on the continuing cookie deprecation, and the already-narrow tracking window gets squeezed even tighter.

Malaysia and Hong Kong. Business in these two markets runs on relationships, introductions, and dinner-table conversations. More than half of the influence on a deal happens in places your CRM will never see. The attribution models designed for digital-first markets simply don't fit here.

So marketers in Asia-Pacific shouldn't keep straining to close that "invisible gap." They need to actively build an honest framework — acknowledge that some things can't be seen, then plug the hole with frontline feedback, customers' own words, and signal-level insight.

One last thing

Back to my friend from the opening.

He told me later that the hardest part wasn't the absence of data. It was having too much data, and not one number willing to stand up and answer the CFO for him.

I told him: this isn't just your problem.

The entire industry is standing at the intersection of "AI is making marketing stronger" and "we can see this strength less and less."

The old attribution framework was designed for the click era. We've already walked into a new world where AI Overviews, dark social, and the content deluge stack on top of each other. Keep measuring the new cloth with the old ruler, and the only thing that grows is your anxiety.

Swap the ruler. Change the thinking.

Acknowledge what you can't see. Patch what you can.

And the rest — leave it to an honest relationship with the truth.