The foundation was doing serious work in India and could not say whether any of it was landing. There was no benchmarking data for the kind of advocacy it wanted to build, no shared definition of who it was talking to, and nothing written down about when it should stay quiet. Programs in this situation usually respond by producing more communications. This one had to start by admitting the instrument was missing.
I won the mandate in May 2015 and led the work with a team of roughly 25 people. The brief was not campaign work. It was the communications architecture for the foundation's largest country program, and it covered seven things at once: India communications strategy, audience strategy, narrative strategy, a communications framework, voice strategy, a risk and reputation framework, and a measurement framework. They were written as one system rather than as seven documents that happened to share a client.
Audience, sized rather than described
I mapped eleven audiences, from national, national-language and regional media through partners, the central government, state governments, beneficiaries and employees, to two distinct groups of Indian millennials who both want the country to do better and disagree about what better means.
Then I sized them. About 77 million Indians were in the 16 to 24 age group, of whom roughly 23 million showed active interest in social good, narrowing through a pyramid to about 2,300 people at the top who could genuinely move an issue. Every named influencer was scored on three things: the total reach of their social properties, the daily visitors and page rank of the portals and blogs they wrote for, and the strength of their followers' followers. Scoring produces findings that a description never does. In one issue area, 53 people qualified as genuinely influential, 38 of them already followed the foundation's account, and the foundation followed none of them back.
Narrative, and who is allowed to carry it
The narrative framework ran a single chain from belief to purpose to vision to mission to value proposition to the work to messages to stories, so that any sentence anyone published could be traced back to the belief it came from, and anything that could not be traced was cut.
The voice architecture assigned a distinct communications function to each of the foundation's principals, so that senior voices were not competing for the same ground or arriving in the same register. Who speaks, on what, and when.
When the client should not speak at all
The political engagement doctrine set different rules for the central government, for the state governments and for the press, because an identical statement carries different risk in each of those rooms. The page I am still proudest of is the filter for when the foundation should not speak, and should place a credible third party instead: when it lacks standing on the issue, and when it will be heard as foreign whatever it says. Knowing when a client should stay silent is the rarest judgment in this work, and almost nobody writes it down where a junior person can use it at nine on a Friday night.
Alongside it sat a reputation grid that classified any live issue on reach against impact, with four named quadrants, so that the classification argument happened once rather than every time.
A measurement framework built to generate its own benchmarks
The measurement framework crossed a five-stage path, from encounter to act, against owned, earned, shared and paid, with a named metric in every cell. It also said in writing that there was no benchmarking data available, and that the framework existed to produce it.
That sentence is the whole method. When the number you need does not exist, you build the instrument that makes it exist, and you say out loud that the first year is calibration rather than performance. Eleven years later I still write that paragraph into everything I build.
Outcomes and outputs
The outputs were the frameworks, the audience and influencer maps, the voice doctrine, the silence filter, the risk grid, the measurement matrix, an editorial calendar and a response protocol for handling everything from a genuine query to a rant.
The outcome was narrower and more useful. The foundation's India team could answer five questions in one meeting instead of five: who are they talking to, what do they stand for, who says it, when should they stay quiet, and what counts as working.
What I would do differently
I would build the measurement instrument first and let it argue with the narrative while the narrative was still soft, instead of designing it last against a story that had already hardened. I would also build the audience pyramid as a live graph rather than a document. It was accurate on the day it shipped and it started decaying the day after, and nothing in the system told anyone that.
This was live client strategy for an active political and philanthropic program, so the casting of individual principals, the model of the political system, and the commercial terms stay where they belong. The method is describable. The specifics are not mine to publish.