KPI Media Leadership Interview Series: Miguel Louis Tolentino

THE ANALYST'S CONTRARIAN VIEW

What measurement trend is overhyped right now?

‍

What is overrated right now is deterministic multi-touch attribution as the single source of truth.

About 41% of enterprises have adopted MTA, but only 18% of teams call it highly accurate. That gap is the story. ATT, dark social, and zero-click behavior don't show up in the dashboard, yet teams treat it as gospel.

MTA is a good tactical compass for creative and keyword calls; it's just the wrong tool for macro budget decisions, especially on data where the platforms grade their own homework. Triangulate it with MMM for allocation, and geo-incrementality tests to stay honest.

‍

And what unglamorous data fundamental do brands keep under-appreciating?

Unified data taxonomy and governance is undervalued.

Everyone wants the AI models; nobody wants to fund the plumbing. Ask any data science team, and they’ll say 60 to 80% of an analytics project is untangling dirty data and mismatched campaign names, not modelling. Without clean naming across media and CRM, your models can only answer surface-level questions. That's a headline, not a decision. Your taxonomy is your analytics ceiling.

‍

What's the problem you most often see surface the moment a client's measurement is finally done right- the thing the old numbers were hiding?

Almost always the same thing: the budget had been preaching to the converted.

Fix the measurement, and you find optimisation was running on self-attributed conversions such as branded search, retargeting, some "performance" social that were mostly intercepting demand that already existed.

The system read that as efficiency. Meanwhile, the channels actually creating demand like display, CTV, creator, prospecting were converting elsewhere and getting no credit. It's structural: last-click rewards demand capture and punishes demand creation, so the budget quietly drifts toward the low-incrementality stuff the model can already see.

‍

MEASUREMENT AT ENTERPRISE SCALE

You've turned complex datasets into insight for high-growth brands at iProspect and beyond. What separates a dashboard that changes decisions from one that just looks impressive?

‍
Every executive dashboard looks impressive on day one. Imagine forty charts, three shades of green. Two weeks later, nobody opens it. The problem is teams build a data-first container instead of a decision-first tool: metrics go on screen because the data exists, not because anyone will act on them.

The ones that change decisions do three things. They show the vital few, which are five to seven numbers that touch the P&L; everything else is behind a drill-down. They swap vanity metrics for triggers: if this, then what do we do? If not, it's off the screen. Every number has a baseline, a target, and an owner, so when it goes red, it's a playbook and a phone call, not a debate.

‍

When is a brand's data or tracking foundation too weak to justify spending on ads at all, and how do you have that conversation?

When you can't connect an ad click to actual revenue. Spending on a broken foundation is just paying to generate noise at scale. In a way gambling with a layer bolted on, not an investment.

Ad-ready doesn't mean the website works; it means you're tracking the outcome you actually care about. Too many teams optimise toward page views or newsletter signups. If you're selling high-ticket B2B, a signup doesn't pay the bills; your conversion event does. Bad data has you killing winners and scaling losers for months.

‍

What's a metric big brands obsess over that you'd quietly ignore?

Hot-take ROAS
At the creative or keyword level, it's a fine diagnostic. It tells you which ad is pulling its weight. But promoted to a boardroom metric, it's dangerous: run a business on platform ROAS, and you're scaling a dashboard, not a company. It measures neither profit nor incrementality. I've watched brands celebrate a 4x while quietly bleeding cash, with platforms taking credit for sales that would've happened anyway.

‍

And what do you watch instead?

At the executive level, I watch three numbers instead: blended CAC, LTV, and payback, and MER. MER, in my opinion, is the kicker: if channel ROAS climbs while MER slides, your ads are cannibalising demand you already had.

‍

BRIDGING TECH, AI & STRATEGY

You lead the integration of AI-enabled tools into analytics. Where is AI genuinely elevating insight generation today, and where is it still overrated?

‍

It splits almost exactly along data quality. AI earns its keep in high-volume pattern recognition, not text generation — anomaly detection that flags a tracking drop while it's still a rounding error, predictive modelling that moves capital before the quarter turns, digests drafted off clean data that hand an analyst back two days a week.

Where it's oversold is "just chat with your data" laid over broken tracking or a disconnected CRM. It fixes none of that; it just accelerates garbage in, garbage out, and delivers it in confident prose. That's a real trap: a broken dashboard looks broken, but a hallucinated insight looks like an insight. AI is a multiplier on whatever foundation it's sitting on, in both directions.

‍

Your role bridges analytics and business strategy. What's an opportunity you've spotted in the data that the business side would never have seen on its own?

At iProspect, we once told a client to cut 20% of the spend on their best-performing platform. Sales didn't move.

Strategy runs on lagging indicators like last quarter's revenue and ROAS, so the logic collapses into arithmetic: double the spend, double the revenue. But channels don't scale linearly.

The response curve had flattened; the channel was still efficient, it had just run out of people to reach at that price. The next $10k had nowhere to go; when we moved it upper-funnel, it did. The business side couldn't see it because nothing in the reporting looked wrong. Nothing turns red when a channel simply stops growing; it just quietly stops, and the P&L doesn't tell you until a quarter later.

‍