Your tools aren't the problem. Your operating model is. Here's why connected customer experiences keep failing — and how to fix what's actually broken.
Brands across SEA are sitting on martech stacks worth hundreds of thousands of dollars — and still delivering customer experiences that feel like they were designed by three separate companies that have never met. The tools aren’t broken. The operating model is.
The Stack Isn’t the Problem. The Org Chart Is.
MarTech’s Gene De Libero, reporting from a panel of practitioners, landed on a diagnosis that should make every CMO uncomfortable: connected customer experiences keep failing not because of technology gaps, but because of how teams are structured, incentivised, and empowered to act. Siloed ownership — where CRM sits in one team, paid media in another, and lifecycle email in a third — means that even when data flows correctly between systems, the decisions downstream are made by people optimising for different goals.
In SEA, this problem compounds quickly. A brand running campaigns across Shopee, LINE OA, and their own app is almost certainly managing those channels through separate teams, separate agency briefs, and separate KPI frameworks. The customer experiencing all three simultaneously has no idea — and no patience for — the internal org chart that explains the inconsistency. The fix isn’t a new CDP. It’s cross-functional ownership with shared accountability for journey-level metrics, not channel-level ones.
Personalisation Has Passed Its Useful Limit
There’s a version of email personalisation that works — and a version that’s become a liability. MarTech’s Alexander Melone makes the case that hyper-personalisation has an overuse problem: when every email references browsing behaviour, purchase history, location, and predicted intent simultaneously, subscribers don’t feel understood. They feel watched.
The distinction that matters here is relevance versus surveillance. Personalising a cart abandonment email with the specific product left behind is relevant — it closes an open loop. Personalising a re-engagement email with a reference to a product someone viewed once, six weeks ago, signals that you’ve been tracking them more closely than the relationship warrants. In markets like Thailand and Indonesia, where WeChat-style super-apps have normalised a certain level of data integration, the threshold for what feels intrusive may differ — but the underlying principle holds: personalisation earns trust when it solves a problem the customer actually has, not when it demonstrates how much data you’ve collected.
Practically, this means auditing your email programme not just for open rates but for unsubscribe triggers. If your most personalised sends are also driving your highest unsubscribe rates, that’s the signal.
AI in the CRM: Compound Advantage or Expensive Autocomplete?
Salesforce’s move to embed AI directly into its SMB CRM workflows — surfacing next-best actions, drafting follow-up copy, and flagging at-risk accounts without requiring a data science team — is a meaningful shift in where AI creates value. For smaller teams operating in cost-sensitive SEA markets, the promise is clear: reduce the analyst overhead required to turn customer data into action.
But the honest read on embedded CRM AI is that it compounds existing data quality. If your CRM contact records are patchy — and in most mid-market SEA businesses, they are — AI suggestions built on that foundation will be confidently wrong rather than obviously broken. The prerequisite work is unglamorous: deduplicating contact records, enforcing field completion at the point of entry, and establishing consistent lifecycle stage definitions across sales and marketing. That operational groundwork is what determines whether AI becomes a genuine force multiplier or an expensive layer of automation built on sand.
Salesforce’s Agentforce platform, which now extends into contact centre workflows, points toward where this is heading — autonomous agents handling routine customer interactions while human teams focus on complex or high-value cases. For SEA brands managing high volumes of inbound queries across WhatsApp, LINE, and web chat simultaneously, that architecture is genuinely interesting. The implementation risk is the same: agents trained on poor data or unclear escalation rules will erode trust faster than they build efficiency.
The Operating Model Fix: What It Actually Requires
Pulling these threads together, the pattern is consistent. Personalisation fails when it’s deployed without a theory of what the customer actually needs at that moment. AI fails when the data infrastructure beneath it hasn’t been treated as a strategic asset. Connected experiences fail when the teams responsible for them are measured on disconnected outputs.
The organisations getting this right in SEA are doing three specific things: first, they’re appointing someone — a head of customer experience, a growth lead, whatever the title — who owns journey-level metrics and has the authority to prioritise across channels. Second, they’re treating first-party data infrastructure as a product, not a project — with ongoing ownership, quality standards, and investment cycles. Third, they’re applying AI to well-defined, high-frequency decisions where the feedback loop is short enough to catch errors quickly, rather than using it to automate complex judgement calls where failure is expensive.
The martech category will keep shipping new capabilities. The brands that extract disproportionate value from them will be the ones who’ve already done the organisational work to absorb them.
The real question for 2026 isn’t which tools to buy — it’s whether your operating model is actually capable of using the ones you already have. Most aren’t. And no vendor roadmap is going to fix that for you.
At Grzzly, we work with SEA brands at exactly this intersection — helping growth teams diagnose where their martech investment is leaking value and building the operational frameworks to capture it. Whether it’s first-party data strategy, cross-channel journey design, or making sense of AI vendor claims, we’d rather have a straight conversation about what’s actually broken. Let’s talk at grzz.ly.
Sources
Written by
Rogue GrizzlyOperating at the contested frontier of cookieless targeting, clean rooms, and identity resolution. Comfortable where the infrastructure is shifting and the playbooks have not yet been written.