YARA FarmWeather
1M Installs. 74% Lower CPI.
01 · Company Overview
YARA is a Norwegian agritech company operating globally. In India, YARA launched FarmWeather; a hyper-local weather app designed specifically for farmers, giving them field-level weather forecasts for sowing, spraying, and harvesting decisions. India has 150M+ farmers. Most are in rural areas, speak regional languages, and many were encountering a smartphone app for the first time.
| Dimension | Detail |
|---|---|
| Product | YARA FarmWeather; hyper-local weather forecast app for Indian farmers |
| Target audience | Active farmers across all regions of India, primarily vernacular-speaking, rural and semi-urban |
| Business goal | Drive 1M+ qualified app installs. Reduce CPI. Improve retention and month-on-month active users. |
| Challenge | Reaching a highly distributed, non-English-speaking audience not easily targeted through standard digital demographics |
02 · Market Context
Why this was a hard app marketing problem: Most app campaigns target urban, English-speaking users via standard interest and demographic targeting. Farming in India is radically different; the audience is vernacular-first, spread across 29 states with different crops, climates, and languages. A generic install campaign would produce installs but not retention. I needed installs from farmers who would actually open and use the app.
03 · Audience Insight; Farmer Segments
| Farmer Segment | Language / Region | Primary Weather Need | Channel |
|---|---|---|---|
| Wheat / rice farmers (North India) | Hindi, Punjabi | Pre-sowing forecast, frost alerts | Facebook + Google (location + farming interests) |
| Cotton / sugarcane farmers (Maharashtra, Gujarat) | Marathi, Gujarati | Rainfall prediction for harvest timing | Facebook vernacular ads + YouTube |
| Rice / spice farmers (South India) | Tamil, Telugu, Kannada | Cyclone and flood alerts, monsoon tracking | Google UAC + YouTube regional content |
| Vegetable farmers (across states) | Multiple | Daily micro-forecast for irrigation decisions | Facebook automatic placements + Google |
The Vernacular Insight: Farming is hyper-regional. A wheat farmer in Punjab has completely different needs and language from a spice farmer in Kerala. When I switched from single-language Hindi creatives to region-specific vernacular video ads, CTR doubled and post-install retention improved significantly. The ad felt like it was made for them because it was.
04 · Strategy
| My Thinking | What I Did and Why |
|---|---|
| A single national campaign cannot reach farmers effectively across 29 states with different languages and crops. | Built a full-funnel, multi-platform strategy with regional segmentation: Google (location + keyword targeting), Meta (interest-based + vernacular creative), Instagram, YouTube (vernacular explainers). Multi-channel presence at all stages of the install funnel. |
| Farmers need to understand what the app does before they will install it. | Created vernacular video ads in 6 regional languages (Hindi, Marathi, Gujarati, Tamil, Telugu, Kannada). Each showed a farmer from that region describing a weather problem specific to their crop. CTR improved 2.1x vs generic Hindi creative. Post-install Day 7 retention improved 35%. |
| Location-based targeting combined with farming keywords creates a very precise audience signal. | Used Google Ads with district-level location targeting layered with farming-related keywords in local languages. On Meta, combined geographic targeting with interest signals: farming, agri equipment, cooperative societies, Kisan TV. Reduced wasted install spend by 45% vs broad national targeting. |
| Lookalike audiences built from best users are more valuable than cold interest targeting at scale. | Built lookalike audiences from two seeds: (1) Users with 3+ app opens in first week, (2) Users who set up location-based weather alerts. Expanded across India in 2% and 5% bands. Lookalike campaigns produced 60% lower CPI than cold interest targeting. |
| Automatic placements let the algorithm find the cheapest qualified install. | Ran automatic placements across all Meta surfaces. Tested against manual placement. Automatic placements reduced CPI by additional 30% vs manually locked placements. |
05 · Results
| Metric | Result |
|---|---|
| Total app installs | 1 million+ qualified app installs achieved |
| Cost per install | 74% reduction over campaign duration |
| Cost per click | 54% reduction |
| Play Store ranking | #1 in Top Free and #1 in Top Trending, Weather Apps category |
| Play Store rating | 4.5 star average; users who installed also used and rated it |
| Month-on-month active users | Grew consistently through the campaign period |
Key Learning: For apps targeting non-urban, vernacular audiences, creative language match is not a nice-to-have; it is a performance variable. Switching to region-specific vernacular video was a bigger lever than any audience or bidding optimisation. Lookalike audiences built from behavioural signals outperform those built from install events alone.