Why MarTech Stacks Are Built on Confidence, Not Data

Walmart just ran an experiment that every martech team in SEA should tattoo somewhere visible: its ChatGPT-powered checkout converted at one-third the rate of its own website. Not slightly worse. Three times worse. And yet the instinct across the industry right now is to pipe AI into every customer touchpoint and call it transformation.

The same week that result surfaced, MarTech published a separate analysis showing that marketing budgets are increasingly allocated not to what performs best, but to what teams feel most confident defending in a board meeting. Two data points, one uncomfortable pattern: the stacks we build and the channels we fund are shaped more by organisational psychology than by evidence.

The Confidence Tax on Media Spend

According to MarTech’s reporting, media spend is consolidating into a shrinking set of channels — not because those channels outperform alternatives, but because they produce metrics that stakeholders recognise and trust. Google Search, Meta, and a handful of regional platforms continue to absorb budget not purely on ROI grounds, but because their reporting infrastructure makes them legible to non-marketing executives.

In SEA, this effect is amplified. Shopee and Lazada ads are defensible because GMV attribution is direct and visible. LINE OA campaigns in Thailand or GrabAds in Indonesia are harder to defend even when they outperform on reach and cost, because their attribution outputs don’t slot neatly into a regional dashboard. The result: brands systematically under-invest in channels that work and over-invest in channels that report well. That’s not a media strategy. That’s organisational risk management dressed up as performance marketing.

The fix isn’t a new platform — it’s building internal fluency. Teams that can translate platform-specific attribution into business language (revenue impact, customer acquisition cost, lifetime value contribution) break the confidence monopoly that legacy channels hold.

Walmart’s ChatGPT Number Is a Warning, Not a Verdict

The 3x conversion gap Walmart reported for ChatGPT checkout isn’t an argument against AI commerce — it’s an argument against deploying infrastructure before the experience is ready. The failure mode here is familiar: a technology gets attached to a high-stakes conversion moment before the user journey around it has been properly engineered.

For SEA brands, the mobile-first context makes this even more acute. A checkout flow that adds cognitive load — asking users to interpret conversational UI when they’re used to tapping through a Shopee-style cart — doesn’t just underperform, it actively erodes trust at the most critical moment. The platforms that have cracked mobile commerce in this region (Lazada, Tokopedia, Sea Group properties) have done so by ruthlessly reducing friction, not by introducing novelty at checkout.

The strategic read: AI has a real role in discovery, personalisation, and pre-purchase support. It does not yet have a proven role in completing transactions. Brands experimenting with agentic commerce should sequence the deployment carefully — use AI to warm the funnel, then hand off to a conversion experience users already trust.

---

HubSpot’s February Updates Reveal Where CRM Is Actually Heading

MarTech’s roundup of HubSpot’s February 2026 updates is, on the surface, a feature changelog. Read it as a strategist and it’s a signal about where CRM infrastructure is consolidating. The updates expand AI capabilities across campaign workflows, tighten attribution modelling, and specifically target friction points that have frustrated operators for years.

What matters here isn’t any individual feature — it’s the direction. HubSpot is building toward a system where campaign performance is legible without analyst mediation. Better attribution plus AI-assisted reporting means that mid-market teams can now produce the kind of defensible narrative that previously required a dedicated data function. For SEA brands operating with lean marketing teams across multiple markets and languages, this is material. The overhead of interpreting performance data across Thai, Bahasa, and English-language campaigns simultaneously has historically been a real constraint on optimisation speed.

The caveat: better tooling in an over-tooled stack is still overhead. Before activating every new HubSpot capability, teams should audit what they’re actually using today versus what they’re paying to have available. Feature proliferation inside CRMs is its own form of the confidence trap — it feels like capability, but unused features just create noise.

Chaotic Content Workflows Are a MarTech Problem, Not a Creative Problem

MarTech’s analysis of content workflow dysfunction makes the point that rework, vague briefs, and reactive production cycles are the primary drain on content team output — not headcount or creative quality. What it doesn’t say loudly enough: this is almost always a tooling integration failure, not a people failure.

When a brand’s content brief lives in a Google Doc, approval lives in email, asset management lives in one DAM, and campaign scheduling lives in another platform entirely, chaos is the default state. The MarTech stack created the problem. The solution isn’t a workshop on better communication — it’s collapsing the number of systems a content workflow touches.

For SEA teams producing multilingual content across markets — a standard requirement for any brand operating across Singapore, Malaysia, Indonesia, and Thailand simultaneously — the integration gap is even more costly. A Vietnamese-language asset that misses a campaign window because of a broken handoff between a localisation vendor and a CMS isn’t a creative failure. It’s a workflow architecture failure. Fixing it means mapping the actual content supply chain, identifying where handoffs break, and eliminating the tools that create seams rather than closing them.

---

The through-line across all of this is the same: marketing teams are making expensive decisions — what to buy, what to fund, what to build — based on what feels defensible rather than what the evidence supports. That’s a rational response to organisational incentives, but it compounds over time into stacks that are over-bought, under-activated, and increasingly misaligned with actual customer behaviour. The question worth sitting with: if you audited every tool in your stack and every channel in your media mix purely on performance evidence, how many would survive?

---

At grzzly, we spend a lot of time inside exactly this problem — auditing MarTech environments across SEA where the gap between what’s been purchased and what’s being activated is wide enough to drive a budget cycle through. If your stack has grown faster than your team’s ability to use it deliberately, that’s worth a conversation. Let’s talk