Google Discover in 7 Indian Languages: What Publishers Should Track Now

On 29 September 2026, Marfeel extended its Discover Monitoring to Tamil, Telugu, Marathi, Kannada, Punjabi, Bengali and Malayalam. Together these languages reach more than 700 million speakers. For the first time, regional publishers can see card-level Discover data in the language they actually publish in.

That matters because most Indian-language newsrooms have been flying blind. They get Discover traffic, often a lot of it, but their reporting stops at a single Search Console line. This brief explains what changed, what it does not change, and the six numbers every regional desk should start tracking this week.

What actually changed

Google Discover itself is not new to Indian languages. What is new is measurement. Marfeel's update lets publishers track card performance, pull out content topics and read engagement for stories in these seven languages. It also lets national outlets compare their language editions side by side.

The coverage is weighted toward very large audiences:

Hindi is not on this list. If you run a Hindi edition, keep using your existing stack and treat these seven as a separate benchmark set.

Why this lands at the right time

Google's February 2026 Discover core update was the first core update built only for Discover. It aimed to show more content from sites based in the user's country, cut clickbait, and reward in-depth, original, timely work from sites with topic expertise. Google launched it for English-language users in the US and said it would expand to all countries and languages.

For an Indian-language publisher, every one of those goals points the same way. Local, original, expert reporting in the reader's own language is exactly what the update describes. Check the Google Search Status Dashboard for the India rollout date before you attribute any traffic change to it.

The 6 things regional publishers should track now

1. Discover share of Google traffic, per language edition. Split Search Console by property or URL folder (/ta/, /te/, /mr/). A language edition where Discover is over 60% of Google clicks needs a different editorial rhythm from one led by Search.
2. Card CTR by language. Compare each edition's Discover CTR against your English site. A lower CTR usually points to weak images or translated headlines that read unnaturally.
3. Topic mix that earns cards. List the 20 stories per language with the most Discover impressions each month. Look for local topics (state politics, regional cinema, district weather) that your English desk never covers.
4. Original vs. translated performance. Tag every story as original reporting or translated from English. If translated stories earn far fewer impressions, that is your case for local reporters.
5. Image source. Tag each lead image as original photo, agency or stock. Discover rewards large, relevant, compelling images; stock photos tend to lose.
6. Author coverage. Track what share of stories carry a named, in-language author with a bio page. Anonymous "Web Desk" bylines are a weak expertise signal.

Language editions vs. English: the gap to close

Most national outlets built their Discover playbook on English. They then copied it into regional editions: same headline formulas, same stock images, machine-translated body copy. That approach leaves value on the table in two ways.

First, local relevance is language-specific. A Chennai reader's Discover feed rewards Chennai news written for Chennai readers, not a translated Delhi story. Second, headline habits do not transfer. Wordplay that works in English often reads awkwardly in Marathi or Malayalam, which can hurt CTR.

A 30-day starter plan

1. Week 1: Set a baseline. Export 90 days of Discover data per edition and record CTR, clicks and top topics.
2. Week 2: Tag the last 100 stories per edition as original or translated, and by image source.
3. Week 3: Commission five original, locally reported stories per edition with original photos.
4. Week 4: Compare those stories to the baseline and decide where to move reporter time.

How to set up Search Console for language editions

You do not need a paid tool to start. Search Console already gives you Discover clicks, impressions and CTR. The problem is that most Indian sites see one blended number for all languages. Fix that in three steps.

1. Add each language folder as its own property. Verify example.com/ta/, /te/ and /mr/ as separate URL-prefix properties. Each one then gets its own Discover report.
2. Export monthly, not daily. Discover traffic swings sharply from day to day. Export the last full month for each edition and compare month on month, so you react to trends, not spikes.
3. Join it with your CMS tags. Match each Discover URL to your CMS data: author, section, original vs. translated, image source. That join is where the useful answers appear.

A paid tool such as Marfeel adds card-level views and topic extraction on top. Search Console alone is enough to prove the case internally.

Five mistakes regional desks keep making

• One headline, seven languages. Translating the English headline word for word is the most common error. Native readers notice immediately, and CTR drops.
• The same stock image across editions. If the Tamil, Telugu and Bengali versions of a story all use the same agency photo, none of them stands out in the feed.
• "Web Desk" bylines. A regional edition with no named authors gives Google nothing to judge expertise on.
• Canonicals pointing to English. This tells Google the regional page is a copy. Each language page should canonical to itself.
• Chasing viral spikes. Sensational headlines may win a day of traffic, but Discover's 2026 update is designed to show less of them. Steady, specific, local stories build longer-lasting reach.

What good looks like: one example

Imagine a Telugu edition covering a new Hyderabad metro line. The weak version is a translated national wire story with a stock train photo and a "Web Desk" byline. The strong version is filed by a named Hyderabad reporter who rode the first train. It shows an original photo from the platform, explains fares and timings for local commuters, and quotes riders in Telugu. Only the second version gives Discover a reason to show it to Telugu readers in Hyderabad, and now you can measure the difference by language.

For the step-by-step editorial and technical fixes, read our playbook: how to optimize Discover in Tamil, Telugu and Marathi.

The bottom line

Regional publishers finally have language-level Discover data. Use it to prove what your desks already suspect: original, local, in-language reporting outperforms translated English copy. Start measuring now, while most competitors are still reading one blended number.

Related reading: Google Discover optimization in 2026 and the AI "Dive deeper" test and the max-image-preview:large tag for Discover.

Source: Marfeel (speaker figures are Marfeel's), Google Search Central.