> So 16 years ago it has been shown that advertising for your own brand name is worthless.

Chanel and every fashion brand that has a reall business in selling stink, Coca Cola, evian, Mercedes, JP Morgan ... We could list them all day. In fact just about all companies that rely on brand perception. All of them Beg to differ with that. And they have market cap that backs that view.

Glossy magazines, television, fancy Cinema ads, Point of Sale, event sponsorship. Nothing online. Ever. At least that I'm aware of and who can miss their advertising without which they have zero business.

Especially consider the stunning success that is Chanel selling $100+ bottles of stink that literally has a manufacturing and delivery cost of cents each. Cost of goods sold is a little higher than that when they include the brand marketing. Without it, not doing much business. Elon Musk has achieved the same, not stupid fashion shows but he is a media event and it reinforces Tesla's brand - that's brand marketing. He's brilliant at it. It's been weeks since a huge Musk media circus, he's hired the onion writers too, must be one worth millions coming up in your favourite news source soon...

Otherwise Google will put an ad for some competitor, that many people will take for a search result, in that slot, right?

> Yet still in 2019, when I google for "Amazon" I see an Amazon ad?

It's funny, I tried the same experiment and saw no ad. Also tried ebay, no ad. I wonder how they're preventing other people from bidding that search term up.

I think your answer is in your comment. Why would the digital marketing guys at Amazon say that their advertising doesn't work and their budget should be cut? They want a bigger budget with more responsibility, not less!

Unlike most companies, Amazon can easily track ad display or click to purchase. While they still can't establish causality (you might have bought it even without the ad), it's still fairly easy for them to calculate if the ads are worth it.

When you have Amazon's scale, you can easily do geo-controlled lift studies like the ones mentioned in the articles to get a pretty good estimate of the incremental ROAS.

Increased click-through I guess, and because they have unlimited money.