Meet Lift and Vantage: The Two Systems Inside Niti's Decision Engine

We've used a lot of words to describe what Niti does before landing on the two that actually stuck: Lift and Vantage. Not because the naming matters for its own sake, but because a brand's marketing team, CFO, and CEO are usually asking two different questions, and conflating them into one dashboard is part of why decisions get made on bad information in the first place.

Lift answers: what is this spend actually returning, once you account for what the platform doesn't tell you.

Vantage answers: why did this metric move, and what's the real cause.

Those sound similar. They're not the same job.

The problem Lift solves

Platform-reported ROAS is not the same as true ROAS, and the gap between them is usually bigger than teams assume. We worked through this with real spend data: ₹3.91 crore in Meta spend, evaluated two ways. Platform-attributed numbers told one story. A blended view - accounting for what actually happened across the full funnel, not just what one platform was willing to take credit for - told a different one: 2.76x blended ROAS against 0.71x on the same spend when measured the way the marketplace or platform wanted it measured. That's not a rounding error. That's the difference between a channel that's working and one that's quietly losing money while its dashboard looks healthy.

Lift exists to close that gap - to rank budget decisions against what spend actually returns, not what a single platform is incentivized to report.

The problem Vantage solves

A number moving is not the same as understanding why it moved. Revenue can be up while contribution margin is down. An adset can keep scaling into a SKU that's already out of stock, with nobody connecting the two systems that would have caught it. A best-selling SKU by units is not always the best SKU by margin.

Vantage is the investigation layer - it traces a metric change back to its actual cause across spend, SKU, supply, and margin data, instead of leaving a team to guess from a dashboard that shows the symptom but not the source.

We tested this against real decisions, not hypotheticals

When we looked back at 63 decisions Vantage and Lift together informed for Pokonut, the outcomes weren't uniformly good - and we think that's the honest and useful thing to report, not a flaw to hide. 46% were graded favorable, 38% neutral, 16% unfavorable. That's not a claim that the system is always right. It's a record of what actually happened, checked against outcomes instead of just checked against confidence.

That's the actual point of building this as two named systems instead of one undifferentiated "AI platform." A founder or CMO asking "is this spend working" needs Lift's answer. A founder or CFO asking "why did this happen and what do we do next" needs Vantage's answer. Bundling them into a single vague product blurs both questions and makes it harder to trust either answer.

We're not going to call either of these "agentic," because that word has become a placeholder for "AI did something" without saying what. Lift ranks budget actions against true return. Vantage investigates root cause before recommending a change. Those are specific, falsifiable jobs, and we'd rather be measured against them than against a buzzword.