The AI Reality Check Every Marketing Leader Needs in 2026

HubSpot’s CEO Yamini Rangan said the quiet part out loud last week: the conversation the market is having about AI and the conversation real business leaders are having internally are almost entirely different conversations.

The media cycle runs on displacement narratives — AI replacing humans, ripping out legacy stacks, token-maximisation as its own reward. Meanwhile, the leaders Rangan actually talks to are asking something far more grounded: which of this spend can I actually measure?

That gap is worth taking seriously. Because in markets like Southeast Asia — where digital marketing budgets are under pressure, growth targets are rarely softened by macro uncertainty, and CMOs are accountable to boards who read the same AI headlines — the hype-to-ROI translation problem is not abstract. It’s a quarterly conversation.

What Business Leaders Are Actually Asking About AI

Rangan’s framing, published on HubSpot’s blog in June 2026, identifies three questions that surface repeatedly from operators running real businesses. How do I make my people better with AI? Which systems can I trust? How do I measure the ROI of this spend?

Notice what’s absent: nobody credible is asking how to replace their marketing team with a model. The organisations that tried aggressive AI-for-headcount substitution in 2024–2025 have, with few exceptions, quietly rebuilt. What they’re doing instead is more interesting — and more difficult. They’re identifying the specific workflows where AI reduces cycle time or improves output quality, and they’re putting measurement infrastructure around those workflows before scaling.

For a regional brand managing campaigns across Shopee, Lazada, and LINE simultaneously, the first question — how do I make my people better — is genuinely the right starting point. The answer usually involves reducing the cognitive load of campaign QA and reporting, not replacing the strategist who understands why Thai consumers respond differently to urgency messaging than Filipino consumers do.

The Retention Baseline You’re Probably Missing

Here’s where the AI conversation gets structurally interesting: most organisations don’t have a clean measurement foundation to evaluate AI’s contribution to anything, because they don’t have clean measurement of the underlying business metrics to begin with.

Martech Zone’s Douglas Karr published a detailed breakdown of customer retention rate calculation in late May 2026, and the core point is deceptively simple — retention rate is ((customers at end of period minus new customers acquired) divided by customers at start of period) multiplied by 100. It’s not complicated maths. But the number of marketing teams who can produce that figure confidently, by segment, with a clean data trail, is lower than most would admit.

This matters for AI evaluation for an obvious reason: if you can’t measure your retention baseline, you cannot measure whether an AI-assisted CRM workflow, personalisation engine, or churn prediction model is actually moving it. You’re flying blind with expensive instruments.

For Southeast Asian brands specifically, this is compounded by fragmented data environments — customers acquired through TikTok Shop behave differently from those through direct brand sites, and the attribution models that work in consolidated Western markets often don’t map cleanly to regional platform ecosystems.

---

The Campaign That Got This Right (And What It Proves)

For contrast, consider how Sky Bet approached measurement discipline in their World Cup 2026 campaign — built by Anomaly, featuring Roy Keane and Micah Richards. The creative hook (a nation culturally consumed by football) was emotionally resonant, but the underlying strategic logic was retention-first: existing customers deepened through a cultural moment, not just acquisition targets addressed through celebrity reach.

The lesson isn’t about football or UK gambling markets. It’s about the sequencing: a clear audience (existing customers with demonstrated affinity), a measurable metric (re-engagement and bet frequency during the tournament window), and creative that served the strategy rather than substituted for it. AI tools played a role in content scaling and personalisation across customer segments — but the measurement framework existed before any model was deployed.

This is the model that transfers. Define the metric. Establish the baseline. Deploy the tool. Measure the delta. That sequence sounds obvious because it is. It’s also frequently skipped when the technology in question is exciting enough to distract from it.

Building the Measurement Infrastructure First

So what does this look like practically for a marketing team in 2026? Three moves that are unglamorous but decisive.

First: define your retention cohorts before touching any AI personalisation tool. Know your 30-day, 90-day, and 12-month retention rates by acquisition channel. If you’re running significant volume through Shopee or Lazada, get those platform-native metrics in a format you actually own — not just what the platform dashboard shows you.

Second: pilot AI tools against a single workflow with a binary success metric. Not “did this make things better” — that’s not a metric. “Did AI-assisted subject line generation increase open rates in our Q3 re-engagement sequence by more than 8% versus our Q2 control?” That’s a metric. One workflow, one number, 90 days.

Third: build internal AI literacy before building internal AI dependency. Rangan’s point about making people better with AI — not replacing them — has a practical implementation. Run structured prompt workshops with your content and performance teams. Document what works. Create shared libraries. The organisations that will use AI well in 2027 are the ones building institutional knowledge about it now, not just subscribing to tools.

---

Key Takeaways

• The productive AI question isn’t replacement versus retention — it’s which specific workflow, measured against which specific baseline, over which specific timeframe.
• Customer retention rate is the most under-measured metric in Southeast Asian digital marketing, and without it, evaluating AI’s contribution to growth is structurally impossible.
• Pilot AI tools against a single, binary success metric before scaling — one workflow, one number, 90 days.

The brands that will look smart in two years aren’t the ones that adopted AI earliest. They’re the ones that adopted it with the clearest measurement frameworks. The question worth sitting with: does your team currently have the baseline data infrastructure to know whether anything you’re deploying is actually working?

---

At grzzly, we work with marketing teams across Southeast Asia on exactly this challenge — building the measurement foundations that make growth investments (including AI tools) defensible and directional, not just optimistic. If your team is trying to get honest about what’s working and why, let’s talk.