← All notes
30 Sept 2026 · E-commerce

Do keyword ads still work?

Originally written in Chinese. Translated with AI, reviewed by Gary. Read the Chinese original →

Recently, while reviewing marketplace ads for a brand, I came back to a very interesting question: at this point, do keyword ads still work?

My answer: yes, they do.

But it's no longer the game it used to be.

When we used to do digital ads, especially search, a lot was in our own hands. Choose keywords, set bids, watch CPC and CTR, adjust placements. Put more money behind keywords that converted well, switch off the ones that burned cash.

A great media buyer often won on exactly these details. But in the last few years the whole advertising industry has clearly been changing, and not just on Shopee. Google has Performance Max, Meta started leaning more and more on its algorithm a long time ago, and marketplaces are now heading in the same direction.

It used to be "I tell the platform which keyword I want to buy". Now it's increasingly "here's my product, here's my budget, here's the ROAS I want, you work out the rest".

Shopee's GMV Max is a very clear example. Keywords haven't disappeared; it's just that the decision over keywords is gradually being taken from the advertiser and handed to the algorithm. We used to be able to adjust keywords, bids and placements. Now, much of the time, what we can really control keeps shrinking, usually down to budget and target ROAS.

Labu-Labu comparison graphic, From keyword buying to algorithm buying. The old way: keyword, bid, placement, click, conversion, with advertisers manually managing keywords, CPC bids and placements. The new way: SKU, product signals, algorithm, traffic, conversion, learning, with brands setting goals and the platform's algorithm deciding delivery. Key takeaway: the optimisation layer is moving from media buying to commerce

So what then? I think the most important things have actually started moving outside the ads dashboard: product assortment and SKUs. In performance marketing, people used to love asking "which keyword should I bid on?" In future, the better question may be: "which SKU deserves the budget?"

Think about it. Give GMV Max two products. One has a good price, good reviews, a high conversion rate, enough stock and a clear listing. The other is pricey, with poor reviews, low conversion, ordinary images and not enough stock. However smart the algorithm is, it can't turn a product that doesn't sell well into a hero SKU.

Ads can bring traffic, but ads can't solve every commerce problem. So now performance marketing is no longer just media optimisation. It's becoming commerce optimisation.

Your pricing affects your ads. Your reviews affect your ads. Your stock affects your ads. Your product title affects your ads. Your vouchers affect your ads. And your margin affects whether you dare to scale at all. In the end, merchandising, inventory, pricing, content and advertising are all starting to connect.

Many people may read this change as: "OK, no need to learn ads any more, just switch on GMV Max."

I think it's exactly the opposite.

Once the platform automates things for everyone, the real gap moves somewhere else. Because when everyone uses the same algorithm, the algorithm itself is no longer your advantage.

What you give the algorithm becomes the advantage.

So keyword ads aren't dead, and search isn't dead. It's just that people in marketplaces may need to start accepting a reality: the most important skill in future may not be how well you tune keywords, but whether you can pick the right products, build the right assortment, have the inventory ready, and then hand the algorithm a SKU that truly deserves to scale.

I recently turned this idea into a longer article on the Labu-Labu blog.

— Gary