You Approved the Budget. But Do You Know Where It Actually Went?

There is a conversation that doesn't happen enough in Indian advertising.

A brand CMO approves a ₹10 crore programmatic budget. It moves through an agency, into a DSP, through one or more SSPs, across sub-exchanges, and eventually reaches a publisher. By the time it gets there, a significant portion has been consumed by fees, margins and intermediary commissions at every hop — none of which are fully disclosed on any standard campaign report.

The CMO gets a deck showing reach, frequency, CTR and view-through conversions. The KPI boxes are ticked. The campaign is declared a success.

But here's the question nobody in that chain is incentivised to ask: how much of that audience was real?

The Illusion of Precision in Programmatic Audience Data

Programmatic advertising was sold to the industry on a promise. The promise was simple and intoxicating: stop buying context and start buying people. Know exactly who you're reaching. Target with surgical precision. Make every rupee accountable.

The reality of how most programmatic campaigns are executed in India is considerably more complicated.

When a brand buys an "In-Market Auto Intender" segment from a data marketplace, what are they actually buying? In most cases: a probabilistic model built on behavioural signals of unknown age, collected via SDKs of undisclosed origin, passed through data brokers who may have re-sold, modelled and expanded it multiple times before it arrives in the DSP as a ready-to-activate audience.

No origin transparency. No consent rating. No freshness timestamp. No signal integrity disclosure. No supply chain map.

The segment label says "In-Market Auto Intender." What the label doesn't say is that a significant proportion of it is modelled from a lookalike expansion, the original signal is months old, and the data passed through multiple intermediaries before it reached the buying platform — none of them disclosed.

A consumer reading the back of a cereal box has more transparency into what they're consuming than a CMO approving a ₹10 crore programmatic buy.

What Genuine First-Party Data Looks Like: A Bobble AI Perspective

I spent over a year at Bobble AI building commercial frameworks around first-party data — and the contrast with what passes for "data" in most programmatic supply chains is stark.

Bobble AI operates at the input layer. As a keyboard platform competing directly with Google Gboard and Microsoft SwiftKey, Bobble sits at the most intimate moment in a user's digital behaviour: the moment of intent. Not inferred intent. Not modelled intent. Actual, real-time typing behaviour — what a user is searching for, responding to, expressing — captured at the exact moment of origination.

There was no signal integrity problem for the core data. No modelling required to determine whether a signal was real. No supply chain transparency question to answer because the answer was unambiguous — the data came from the user's own input, on their own device, in real time.

Where AI modelling did come in was in categorisation — using Bobble's AI Labs capability to organise behavioural signals into structured audience cohorts for advertiser activation. That's a legitimate and transparent use of modelling: not to manufacture signals that don't exist, but to make real signals commercially actionable.

The difference between this and what most data products in the programmatic market offer is not incremental. It is categorical.

Most programmatic audience segments start with a thin slice of observed behaviour and model outward — sometimes at 11x lookalike expansion — until the original signal is so diluted it barely exists. The segment is sold as if it were deterministic. It isn't. It never was. And nobody in the chain is required to tell the buyer that.

The Commission Problem: Why No One in the Chain Is Incentivised to Tell You the Truth

Here's the structural issue that makes this worse — and that the industry actively avoids discussing.

Every intermediary in the programmatic supply chain earns more when more budget flows through it. Agency trading desks take a margin. DSPs take a tech fee. SSPs take a supply fee. Sub-SSPs take another cut. Data vendors charge CPM uplift. The more complex the chain, the more entities there are to take a piece.

This creates a profound misalignment of incentives. The people advising brands on where to spend their programmatic budgets are, in many cases, financially rewarded for routing that budget through more intermediaries, not fewer. The people selling audience segments are rewarded for making those segments look as large and targetable as possible — not for disclosing that most of the scale is modelled.

It is the brand's money. It is the brand's consumer. But the accountability for how that money travels through the supply chain — and how much of it actually reaches a real human being in a relevant context — sits with nobody.

Agency KPI sheets measure delivered impressions, click rates and view-throughs. They do not measure how much of the audience segment was inferred. They do not measure signal freshness. They do not measure how many intermediaries touched the supply path or what each one cost. The boxes get ticked. The budget gets renewed. The problem compounds.

Why Publisher-First Planning Is the Supply Chain Antidote

The antidote to a broken supply chain isn't a better algorithm. It's a shorter, more transparent path — starting with a publisher who actually knows their user.

A publisher with genuine first-party data has something no data broker can manufacture. They have signal origin clarity — they know where their data came from because it came from their own platform, their own registered users, their own content interactions. They have consent — real, specific, purpose-bound consent — not a checkbox buried in a terms and conditions scroll. And they have freshness — signals that reflect what their audience is doing now, not what a modelled lookalike was doing eight months ago.

This is why platform selection should follow data quality — not the other way around. The question a media planner should be asking before a single rupee of budget is committed isn't "what's the CPM on this SSP?" It's "what is the signal origin of the audience I'm buying at this publisher, and can I verify it?"

In India, publishers who have invested in building genuine first-party data infrastructure — through logged-in user bases, consent frameworks, behavioural intelligence and identity solutions — are sitting on an asset that the programmatic industry has consistently undervalued. Not because the asset isn't valuable, but because the opacity of the supply chain has allowed inferior data to compete at the same price point.

This publisher-first logic connects directly to broader revenue architecture thinking. As we explore in The First Sale Is Never About the Product, the brands and platforms that invest in building genuine value infrastructure — not just reach — are the ones that scale sustainably.

Supply Path Optimisation Done Properly: Beyond Cost to Signal Quality

Supply path optimisation became an industry talking point as a cost efficiency play. Fewer hops, lower fees, better working media ratios. All of that is true and worth pursuing.

But SPO done properly is not just a fee reduction exercise. It is a signal quality and accountability exercise.

The full chain looks like this: Client budget → Agency → DSP → SSP → Sub-SSP → Publisher. At every handshake in that chain, two questions should be answerable: what does this intermediary cost, and what does this intermediary add? If an SSP cannot demonstrate that it is delivering incremental publisher access, verified audience quality or measurable supply transparency — it is a cost without a corresponding value.

The brands and agencies that approach SPO through this lens — asking not just "how many hops?" but "which hops can I audit and which publisher's data can I trust?" — will build programmatic supply chains that are not just more efficient but more defensible.

Because the regulatory environment is moving in one direction. India's Digital Personal Data Protection Act is now operational, with enforcement beginning May 2027 and penalties up to ₹250 crore. The questions that DPDP raises about consent, data origin and purpose limitation are exactly the questions that a rigorous SPO framework forces the supply chain to answer. The brands that have already built clean supply paths grounded in publisher first-party data will be prepared. The ones that haven't will be scrambling.

The Three Questions Every Brand Should Demand Answers To Right Now

On audience data: What percentage of the segments I'm activating are deterministic versus modelled? What is the average age of the signals in those segments? Who collected this data, under what consent framework, and how many times has it been re-sold?

On supply path: How many intermediaries sit between my DSP and the publisher? What does each one cost and what does each one contribute? Which publishers in my plan have verifiable first-party data that I can audit?

On accountability: If my campaign delivers its KPIs but a majority of the audience reached was inferred and stale — did it actually work? Who in my agency is accountable for data quality, not just delivery metrics?

The programmatic supply chain will not fix itself. The incentive structures are too entrenched and the opacity is too profitable for too many participants.

But brands have more leverage than they think. The budget is theirs. The consumer is theirs. The right to ask these questions — and to route spend only through supply chains that can answer them — belongs entirely to the advertiser.

The brands that start exercising that leverage now will not only spend more efficiently. They'll build audience intelligence that compounds, supply relationships that are defensible, and a programmatic practice that doesn't fall apart the moment a regulator asks where the data came from.

For a broader picture of how AI search is changing where consumers discover brands before they even enter your programmatic funnel, read The Funnel Didn't Break — The Top of It Just Moved Somewhere You're Not.