Social analytics is bought as a measurement programme and justified as one, but it only ever pays for itself as an early-warning system — the value is not the annual brand-health score, it is the signal that something is changing, fast enough to act on it.
The Indian version of this argument comes with sharper constraints than the global one. India’s consumer conversation is multilingual — the discussions that move fast-moving consumer goods (FMCG) categories happen in Hindi, Tamil, Telugu, Bengali and eight other languages, not in the English that most global listening platforms parse well. It is fragmented across Instagram, YouTube, regional-language platforms and short-video apps. And an ever-larger share of it is dark: WhatsApp groups and private messages that no listening tool can see, which means the visible conversation systematically understates the real one. A programme built on an English-language dashboard of public mentions is measuring a shrinking, unrepresentative slice of Indian consumer opinion — slowly. If the output is also an annual tracking report, you have paid enterprise money for a rear-view mirror with a blind spot.
A decade ago the diagnostic question we asked marketing leaders was blunter: who owns the social data your agency captures on your behalf? The answer, almost universally, was the agency. The brand received a monthly report; the raw conversations, sentiment classifications and historical baselines lived in the agency’s tools, and changing agencies meant losing years of accumulated baseline. That question has largely been settled by the platforms themselves — every brand now has native access to its own account analytics. But settling it exposed the harder question underneath, and this is the one we now put to every leadership team: who owns your social intelligence? Not the data — the standing capability to turn the data into a decision. In most Indian organisations the honest answer is that social intelligence lives in an agency slide deck, presented monthly, filed quarterly, read never. Ownership of data was a contractual problem. Ownership of intelligence is an operating one, and it is still mostly unsolved.
Across a decade of deployments, the engagements that justified their cost share one property: the insight arrived faster than the conventional alternative could have delivered it, and someone acted on it. The public case we have cited to boards for years is Kraft’s Vegemite work in Australia. A product bought on habit and toast, listened to at scale across languages and platforms, turned out to be used and discussed in ways the brand’s own positioning did not reflect — and the company repositioned and re-packaged on the strength of what consumers were already saying, publicly crediting the work with changing how it looked at its own brand. The vendor-reported figures from the period, offered here as reported claims rather than audited fact:
| Kraft Vegemite listening programme (public case; vendor-reported at the time) | Reported figure |
|---|---|
| Languages monitored | 38 |
| Postings analysed | ~1.5 billion |
| Claimed improvement in time-to-insight | ~70% faster |
Strip out the vendor gloss and the durable lesson is the last row. The repositioning did not come from a superior annual study; it came from hearing the market months before a research cycle would have reported it. We saw the same mechanic in Indian food-category work in the mid-2010s: a listening exercise for a large Indian foods conglomerate surfaced, within weeks, a specific and persistent consumer complaint about the category leader’s product — the kind of competitive opening a syndicated panel would have delivered a year later as a footnote, if at all. The point of the story is not the finding; it is the clock. Launch and positioning decisions in Indian FMCG were, and largely still are, made on gut feel plus syndicated panel data that arrives on a monthly or quarterly cadence. Listening’s entire economic case is that it beats that clock. The brand-run social command centre that a global sports-drink brand made famous — real-time monitoring staffed by the brand’s own people, in public view — was the early institutional form of the same idea: listening as an operations room, not a reporting function.
Here is the dual lens. The brand voice wants listening to guard equity — track sentiment, measure campaign reception, produce the annual health narrative for the board. The evidence voice observes that everything in that list is a lagging indicator, and that lagging indicators are precisely what panels, trackers and sales data already provide, more rigorously. What listening uniquely offers is lead time: the quality complaint before the returns data, the competitor campaign before the share shift, the emerging use-case before the category re-segments. Fund it as a leading-indicator system or do not fund it. And in India, fund the hard parts: regional-language coverage with models actually trained for it, proxy measurement for the dark channels — trackable forward links, platform-specific signals — and explicit acknowledgement in every readout of what the tool cannot see. A listening programme that is silent about WhatsApp is not conservative; it is misleading.
On Monday, do three things. First, ask who in your organisation — a named person, not a vendor — owns social intelligence, meaning the authority and the standing process to turn a signal into an escalation. If the answer is your agency, you own a report subscription, not a capability. Second, instate a decision log for the programme: every week, one line — what did listening tell us, and what did we do about it? Run it for a quarter. Third, act on the log. If a quarter passes and the honest answer is “nothing changed because of this,” do not renew the dashboard and do not commission a bigger annual study; redirect the budget to fixing speed — language coverage, alerting thresholds, an owner with authority — because that is the only axis on which this discipline has ever earned its money. The annual brand-health score will survive without social analytics. Your response time will not.