VP of Sales shares how revenue teams can join the 6% that have successfully implemented AI

Byron Munson is the Vice President of Sales for JAPAC at Supermetrics.

We sat down with Byron to learn why only 6% of teams have successfully implemented AI — and how he's making it happen for his team. Here's what he told us.

Building commercial engines from scratch

I'm Byron Munson, Vice President of Sales for JAPAC at Supermetrics, a global growth-stage MarTech business. Our customer base ranges from SMBs to the largest enterprise brands. Our GTM structure centers on a lean Key Account Manager, Commercial Account Executive model, and a partner organization.

I've spent twenty-plus years across APAC and EMEA, building commercial engines from scratch to regional scale. I started in advertising at DDB and McCann in Sydney, then moved into ad tech at Adconion London and Videology SEA, and later into MarTech at Datorama. I helped run APAC for Datorama and landed the Salesforce acquisition in the region.

Going through an acquisition from the inside, twice now, taught me a lot I wouldn't have learned from a strategy deck: what survives in a larger organization, what gets rebuilt, and what quietly falls away.

I also founded my own company, a climate tech startup called Climagap, which gave me a healthy dose of humility regarding ideating, securing funding, and bringing an idea to market from scratch. Ultimately, the business didn't scale for various reasons, so I decided to close operations.

At Supermetrics, I lead a regional team of ~20 cross-functional employees. We built JAPAC from no dedicated in-market presence into a significant regional operation. This includes a partnership ecosystem that provides access to strong in-market implementation and technology operators, which is essential to sustainable scaling and our customers’ long-term success.

A revenue organization affected by AI on two fronts

More recently, this extended into AI on two fronts. Internally, we built our own AI-driven revenue infrastructure — covering pipeline health, customer risk, and deal coaching — grounded in our own data. This taught me firsthand that AI adoption is primarily a trust and change-management problem, not a technology one.

And externally, I've watched the same shift play out with our customers: brands and agencies drowning in data overload. The real unlock hasn't been more data; it's been helping them decide faster.

Our LLM partnerships, including our more recent role as Google’s Gemini Enterprise launch partner in the region, sit at that intersection — putting AI-powered marketing intelligence directly into marketers' hands. Honestly, I've learned it's not about the technology; people only act on insight that they trust, whether it's a rep trusting a coaching nudge or a customer trusting a recommendation enough to change their marketing investment plans.

Where AI excels and where its utility ends

AI empowers decisions by providing insights. Upgrade propensity is the clearest case; in a self-serve motion, usage, billing, and engagement signals can sit in disconnected systems with no natural connection. AI pulls these data sources together, revealing a clear signal of who is ready to expand.

But AI is never the sole driver. What remains entirely human includes pricing, negotiation, and anything built on trust and interpersonal interaction. AI can tell me an account's ready. It can't have the conversation.

And it's worth noting that not every AI bet converts to revenue as quickly as AI enthusiasm suggests. Some consume significant capacity before producing much.

That's why the risk I watch hardest is false confidence. AI output can look polished and definitive even when built on thin data. I treat it like a junior analyst's first draft: useful, but I do not act on it unchecked.

Two places where AI misses the mark in revenue organizations

I'm seeing two dangerous misses when it comes to AI in revenue.

The clearest is AI-generated outbound personalization. It's efficient on every activity metric, but it didn't move win rate because buyers can spot synthetic personalization instantly.

Here's a subtler one: letting a model score deal risk automatically. It was right often enough to be trusted, and wrong exactly where it mattered — relationships going cold, which never show up in structured data.

For both, a human is needed in the loop.

How using AI in performance management and market strategy improves decisions

I've made two big changes recently:

1. Performance management: We now surface individual and team risk before it becomes a performance challenge, not after. Proactive solutioning instead of a post-mortem approach. We do this by pulling call transcripts, pipeline history, CRM notes, product-line revenue, and customer feedback — and AI surfaces underlying behavior.
2. Market strategy: Pulling 4 years of operating data across the region surfaced a trend in Thailand, specifically, which led us to expand with a partner-first strategy. We built out a robust solution partner network in market, rather than landing boots on the ground ourselves.

We achieved this by connecting all our internal data and third-party platforms with our Claude environment. Ultimately, this provides the foundation to make repeatable, commercially impactful decisions at scale. Not a one-off deep dive, but a long-term, consistent approach.

Byron Shares

We built out a robust solution partner network in market, rather than landing boots on the ground ourselves.

How customer-centric collaboration becomes possible with AI

Sales and Customer Success always ran alongside each other: same customer, two functions, two systems of record, two versions of the truth. That's not a partnership. It's parallel play.

The industry caught up to that diagnosis. Serious revenue leaders now agree on "One number, one motion, one source of truth." GTM as a connected system, not a relay of handoffs.

That idea is no longer controversial. But most stop at structure: shared quotas, unified dashboards, CRM consolidation. That's plumbing. It doesn't answer the question, "How do two functions, sitting on two different systems, ever describe the same customer the same way?"

Here's the mechanism: When language from every customer touchpoint (sales, onboarding, and renewal) sits in an LLM that's pointed at your data, there is no longer a sales version of the customer and a customer success version.

This is how collaboration and transparency show up in practice. Not a value on a slide, but the structure that makes it impossible for teams to lose sight of the customer. And this is what customer-centric collaboration looks like.

How to successfully adopt AI in revenue organizations

89% of teams feel pressure to adopt AI. Only 6% have actually implemented it. That gap is the opportunity (Supermetrics, Marketing Data Report 2026).

Here's how to bridge that gap:

1. Audit your gaps and needs. Establish your goals, map A/B tests to surface the true long-term direction. Then, map your data. If you don't connect it, the LLM has no context. Most stalled "AI initiatives" stem from data hygiene problems. If you don't trust the inputs, you cannot trust the outputs.
2. Don't roll out AI org-wide on day one. Find your team's single highest-drag task: deal summaries, QBR prep, pipeline hygiene. Rebuild that one workflow with its full context. Prove the efficiency gains before scaling.
3. Write the rules down. Your reps use undocumented judgment — what a healthy deal looks like, what "stalled" means at month nine, and how they distinguish renewals from upsells. If you don't write it down, your LLM can't use it, and neither can new hires.
4. Measure adoption by decisions changed, not seats activated. License counts and login stats tell you nothing. Track whether AI-surfaced insights change forecast calls, deal strategies, or coaching conversations. That's the only signal that separates real implementation from those who never get past a pilot.

Byron Shares

Here’s how to bridge that gap: Audit your gaps and needs. Don’t roll out AI org-wide on day one. Write the rules down. Measure adoption by decisions changed, not seats activated.

Why LLMs must be onboarded like new hires

And one more thing: Onboard your LLM as you would a new hire. Skip this onboarding, and you get confidence without competence.

A new employee doesn't receive access to the CRM or your playbook without knowing roles and responsibilities and understanding what "good" looks like. An LLM needs the same four key elements before it touches live data:

• Your goals: What "healthy" means for this pipeline, this quarter, this region
• Your definitions: What "at risk" means, what a healthy pipeline looks like
• Your guardrails: What it must never decide alone — for example, no reclassifying a deal stage or calling a churn reason without a human in the loop.
• Your history: The context that changes what a number means — for example, "pipeline coverage is low" reads very differently once you know that we just pulled a major renewal forward a quarter.

Skip that onboarding and the model doesn't fail quietly — it fails convincingly.

Follow along

You can follow Byron Munson's work on LinkedIn.

More expert interviews to come on The CRO Club!