If you run a store in Kochi or Thrissur and you're about to pay someone to convert the whole site into Malayalam, stop at the product page. A Malayalam ecommerce website earns its money in exactly three places — ingredient and sourcing detail on ayurveda and food products, the story behind a handloom or kasavu piece, and support conversations on WhatsApp. Everything after "Add to cart" converts better in English for the same buyer, including the one who read the whole description in Malayalam two minutes earlier. We've set this up both ways and the full-translation version loses, usually at the payment step.
That sounds backwards. It isn't. The Malayali shopper on a 4G phone in Kozhikode consumes in Malayalam and transacts in English, because every system downstream of your checkout — the UPI app, the bank OTP screen, the Delhivery label, the Razorpay receipt — is in English anyway.
Translate what the buyer reads to decide, not what they click through
Draw the line at the decision. Anything the customer reads while deciding whether to buy is worth translating. Anything they interact with to complete the purchase should stay in English.
Worth translating: product descriptions for ayurveda, Ayush and food SKUs; ingredient lists and sourcing notes; usage and dosage instructions; care instructions for handloom; the "why this is different" paragraph on a spice or cashew listing; FAQ answers about shelf life and shipping; the making-charge explanation on jewellery.
Leave in English: size charts, cart and checkout labels, payment method names, the address form, order status and tracking pages, shipping and returns policy headers, discount code fields. A customer who sees "വലുപ്പം" above a dropdown that contains S, M, L has gained nothing and lost half a second.
The address form is the one people argue about until they ship. If a buyer types a Thiruvananthapuram address in Malayalam script, that string goes onto a courier manifest and into a pincode-matching API that has never seen Malayalam. We've had to clean up those orders by hand. Keep the form in English and the error messages in English, because the error messages tell the buyer what the system will accept.
How to add Malayalam to a Shopify store without breaking checkout
Shopify's own Translate & Adapt app is free and it's the right tool. Install it, add Malayalam as a language, and it gives you a side-by-side editor for every translatable string in the theme, your products, collections, policies and metafields. You publish the language under Settings → Languages, and it appears on your storefront at a subfolder URL — yourstore.com/ml/products/….
Three things to check before you publish, in this order.
- Test the checkout in Malayalam before customers see it. Shopify's own checkout strings are translated for a defined list of languages, and the coverage for Indian languages is thinner than for European ones. A checkout that renders half its labels in Malayalam and half in English looks broken, and a buyer who thinks the page is broken abandons rather than asks.
- Follow the buyer out to payment. Shopify Payments, Razorpay and every UPI intent screen will hand your customer an English page regardless of what your storefront is set to. If the handoff is jarring, that's an argument for keeping the final steps English end to end.
- Machine translation into Malayalam needs editing, not a glance. The auto-translate output is fine for navigation labels and genuinely unreliable on compound product terms. "Cold pressed" and "single origin" come back as things no Malayali writes. Budget editing time for anyone who sells food or ayurveda.
Under Shopify Markets, language is assigned per market. The practical setup for most Kerala merchants: one India market with English as default and Malayalam as a secondary published language, and a separate GCC or international market in English with its own currency and duty settings. Resist the urge to publish Malayalam to the UAE market because your customers there are Malayali. They are shopping on an English phone in Sharjah, and what they actually need from you is AED pricing, a delivery estimate they believe, and a shipping cost that doesn't triple at checkout.
What earns the translation: ingredients, food, and the rate card
Three categories where we've seen Malayalam copy do real work.
Ayurveda and Ayush products. The buyer wants to know what's in it in the words they'd use talking to a vaidyar. Sanskrit-derived ingredient names in Malayalam script carry meaning that the English transliteration loses — someone who recognises the herb by its Malayalam name does not recognise the romanised version on sight.
Food. Banana chips fried in coconut oil, pickles, spice blends, cashew grades. The sensory language is simply better in Malayalam and the English version reads like a packaging brief.
Gold jewellery. Not the product copy — the rate breakup. Older buyers in Thrissur and Kottayam verify the arithmetic before they trust the price, and a table that shows metal rate, making charge, wastage and GST separately in Malayalam gets read line by line. If you're displaying live rates and repricing as the spot moves, our Gold & Silver Live Rate app handles the breakup table, and the labels are yours to set in either script.
Translating ayurveda product claims is a regulatory job
A claim in Malayalam is still a claim. If your English page says "supports digestion" and the Malayalam version slides into "cures acidity" because the translator chose a stronger verb, you have created a drug claim on a licensed Ayush product, and you've done it in a language your compliance reviewer probably didn't read.
Our rule: the Malayalam claim must be a translation of the approved English claim, not an independent piece of copy. Write English first, get it signed off against your Ayush licence and label, then translate with the constraint that no new benefit appears. For packaged food, FSSAI's labelling requirements are met in English or Hindi; Malayalam on your website is additional context for the buyer, never a substitute for what's on the pack. Check the specifics with whoever handles your licensing, because the categories differ and the penalties are not symmetrical with the upside.
Malayalam product descriptions and SEO: the uncomfortable part
People in Kerala watch, read and chat in Malayalam. They search in English. Open Search Console on any Kerala store with a mixed catalogue and look at the query report — the queries arriving are "ayurvedic hair oil kerala", "kasavu saree online", "nendran chips online order", with a long tail of romanised Malayalam typed in Latin script. Queries in actual Malayalam script are a small slice, and they're mostly informational rather than commercial.
So treat Malayalam product descriptions as a conversion asset, not a traffic asset. They lift the buyer who is already on the page. They will not bring you a second stream of organic sessions, and anyone promising that should show you the query data first.
Two technical notes. Shopify generates hreflang alternate tags automatically for every published language and market, so your /ml pages and your default pages get declared as alternates without you touching anything. Do not hand-add hreflang in theme.liquid as well — we find duplicated or conflicting tags on about a third of the stores we audit after someone has followed a generic tutorial. And if the Malayalam page is raw machine output that's near-identical in structure to the English one, Google treats it as what it is. Publish fewer translated pages, edited properly.
The arithmetic on a bilingual Kerala store
Numbers, so you can argue with them.
Say you have 420 SKUs, descriptions averaging 180 words, and a decent human translator at ₹2.50 per word. That's ₹450 a product, ₹189,000 for the catalogue, before theme strings and policies.
Now the return. Onam fortnight brings 60,000 sessions. If 4% of them switch to Malayalam, that's 2,400 sessions. Suppose the Malayalam copy lifts conversion on those sessions from 1.6% to 2.0% — a real lift, not a rounding error. That's 0.4% of 2,400 = 9.6 extra orders. At ₹1,900 AOV, ₹18,240 in incremental revenue for the peak fortnight.
Full translation takes roughly ten Onams to pay back. Translating only the 40 SKUs that carry most of your festive revenue costs about ₹18,000 and clears in one. Do that version. Add more SKUs after you've seen the language switcher data for a quarter.
The switcher itself is worth instrumenting from day one. Fire an event on every language toggle. If under 2% of sessions ever touch it, you have your answer and you've spent almost nothing finding out.
Where Malayalam actually pays: WhatsApp
The highest-return Malayalam you will write isn't on the site. It's the WhatsApp reply to "ith original aano?" at 9pm. Pre-sale questions, COD confirmation calls, return pickups, order-not-received panic during Onam — all of that in Malayalam, from a human or a well-written template, moves more revenue per rupee than translating your returns policy.
Same for the COD confirmation message. A buyer who gets "please confirm your order" in English from an unknown number ignores it. In Malayalam, with the product name and the amount, they reply. That's directly fewer RTOs.
Questions we get from Kerala merchants
Does regional language increase conversion in India? On the right pages, yes — and the effect is category-dependent, not universal. Ayurveda, food, handloom and jewellery see it. Electronics, accessories and anything where the buyer is comparing specs don't, because the spec vocabulary lives in English in the buyer's head too.
Should a bilingual store use a subfolder or a separate domain? Subfolder. Shopify's Markets implementation handles the /ml path, the hreflang and the canonical for you. A separate .in domain for Malayalam splits your authority and doubles your maintenance for no gain.
Which Shopify plan do I need? Basic or Shopify handles multi-language and Markets fine. Nobody needs Plus to publish Malayalam. If someone is quoting you a Plus upgrade to add a language, that's a bad reason — the real triggers are checkout customisation, B2B and volume.
Can I just run the auto-translation and skip the editing? For navigation, policies and shipping FAQ, yes. For anything with a product claim, no. That's where the embarrassment lives.
Start with ten products. Your best sellers going into Vishu or Onam, translated properly by someone who writes Malayalam rather than someone who speaks it, with the checkout left alone. Tag the language switcher event, run it through one peak, then decide what else is worth the money. If you want the current state of your store checked before you spend anything, our free audit covers the Markets and language configuration along with the usual speed and tracking issues, and the Kerala page has more on how we work with merchants here.