An agile social media model built for speed, scale, and success

Gemini Social, Google

The Challenge

The AI race is on, and it’s playing out in real time on social media.

To gain an advantage in APAC, the world’s fastest growing digital region, Google wanted to use its country-level social media handles to start building a more locally relevant presence for Gemini. Sounds simple, but in fact, winning in APAC comes with multiple challenges:

Challenger Mindset

For the first time in a long time, Google was the challenger brand. ChatGPT and even Meta AI had already claimed a larger share of voice in APAC.
APAC Complexity

APAC is a complex region, with work in each market requiring local insights and nuance. Simply translating and adapting global assets wasn’t an option.
Org Complexity

In organizations as big and complex as Google, creative assets often have to go through multiple approvers and endless feedback loops which cause delays.
Volume of Assets

Winning in social takes both quality & quantity. Creating localised assets for 7 different markets, with the aim of at least 1 post per week was daunting.
Pace of AI Evolution

Gemini is evolving rapidly, with new product launches becoming an almost weekly occurrence. Keeping pace required an agility the org didn’t yet have.
Local vs Regional

Local markets ran their own country level Google social media handles, with several competing priorities (Search, Gemini, Pixel etc.), and each with their own agenda.

The Solution

A radical new agile social media model built for speed, scale, and success. The guiding principle of this process was prioritizing speed & efficiency, without compromising on creative quality & local relevance

Working hand-in-hand with the Google Gemini product marketing team, every week would follow a process of rapid strategy and ideation, same-day approvals (or troubleshooting and revisions), followed by storyboarding, and market opt-ins. The goal was to hand over creative to the production team by the end of every week. The production team would work on execution the following week, as we repeated the creative process all over again.

This was a groundbreaking new approach for the team at Google, who were used to working with agencies that took weeks to deliver a single video asset. Here, we were delivering 5-6 localised assets a week, and continuing to learn and iterate with every one.

Degrees of Localization

70% original concepts. 30% adaptations. 100% locally relevant.

With a project of this size, each week we had to prioritize where we needed simple local adaptations v/s localized concepts v/s wholly original concepting and execution. In the end, we created a model with roughly 70% original local concepts, and 30% localized adaptations. A ratio that’s still evolving, and continuing to skew towards more and more original concepting.

Localized adaptation

Localized adaptations are created for global concepts which can work almost entirely as originally concepted by the global Gemini social team, with small tweaks to localise the talent, the scenarios, and the language. Here are a few simple examples.

Localized concepts

Localized concepts take the intent of a global concept, but created something completely new for markets, based on what we think will resonate. Here is an example for the launch of vertical video in Gemini, where the original concept didn’t feel “APAC enough”. This is a simple example of how speed and efficiency can work hand in hand with creativity and local insight. Instead of simply translating the global asset, we captured the intent, and created new original assets which better resonated with local audiences.

View this post on Instagram

A post shared by Google Gemini app (@googlegemini)

Global Asset

Glass Elevator (JP, KR)

Vertical Stories (IN, ID, VN)

Original local concepts

These are concepts created specifically around APAC moments, trends, or needs, completely independent of any global input. Examples were festivals like Diwali in India, the fall season in Japan, or the totally unique “rambutan festival” of Indonesia. These helped us increase local relevance, while continuing to educate users about Gemini’s features.

The Results

In the first 6 months, we built 300+ assets across 7 markets, and tackled 19 different product launches. Google Gemini awareness is now at its highest in the APAC region, and this project is cited internally as a benchmark for other teams to try and emulate.