> For most businesses today, Google ads are not cost-effective. This applies double when you are buying ads for your company's name: you would have gotten the clicks even if you weren't paying!
Source?
It’s always extremely funny to read engineers on hacker news to make claims about ads, with no understanding of how the business work. And the underlying assumption that companies throw away billions of dollars away, with no results to show for it.
I wonder if someone from marketing chimed in on a testing topic, to point out that for most lines of code written today, unit tests and integration tests aren't cost effective, and that companies are wasting billions of dollars on writing and running tests that pass 99% of the time, anyways...
If it was a marketing person with serious stats chops using the data to make the case I'd absolutely listen to their case.
Yeah marketing serious stats chops, never seen it yet. Show us the flipping data if you want to make the case for something otherwise the /only/ option is to update your prior with bayes.
xyz corp & co spends billions on it = must work. Reasonable.
every small co. I've heard of loses money at it- looks bad.
Businesses built on brand advertising, chanel, coca cola, etc don't use google at all. Why not? Looks bad.
Modern, internet marketing maestros who have built major brand names on the internet did not pay google a cent. Tesla - rivals mercedes with no ads at all, Trump maga etc. (no I don't like them not that it's relevant). - Looks bad.
Google, advertising to /internet/ /users/ don't use google to do it, they have tv ads! Because google advertising works so effectively?
I dunno, I guess it could work. It could work badly but fashionably for large cos? It could be a group-think crock.
The last 2 shouldn't be as plausible as they are though. Why has google let them be that plausible if their stuff works and they can produce the numbers? Or maybe they, I dunno, can't?
The world of marketing is much deeper than you make it out to be. Brands spend millions - obviously if someone smart came in and did testing to prove it was ineffective thus move that money elsewhere.
You can easily test branding dollars with things like brand lift studies, but I’ll skip over branding.
The majority of small business and good chunks of large companies ad spend is spent on performance marketing. These campaigns have strictly enforced goals of return on ad spend (or sometimes net profit return on ad spend). When a new client starts running big budgets they run incremental alit y tests with their internal data science team, a neutral third party vendor (paid regardless of the outcome), and with the platform their spending money on. Each team shares their analysis but brands usually just ignore the platforms numbers. The advertiser can determine in their own, with legit statistical analysis, whether purchases would’ve occurred even if they hadn’t seen an ad.
I don’t work on one of those teams, but I think it’s basically you spend 80% if your budget like normal on 80% of users. Then on 20% of users you use the same targeting but serve 1 single blank ad. Then you compare how the test and control groups make purchases, and how much of a lift in purchases the advertising drives. They also compare the user buckets before even doing the test to ensure users are evenly split so each group was making the same amount of purchases on average before the campaign begins. They remove outliers, etc.
It’s not a bunch of rich old established men in a room high giving themselves, it’s data scientists who don’t at all trust the guys spending their money. So they test the shit out of it over and over.
All of the raw data can be seen by the advertiser, and they track each individual impression, click, etc. that data can also be filtered through vendors to identify potential fraud, brand safety, etc.
I’ve had more than a couple client run an incrementality test and then stop spending on the platform because the results were too low. But I’ve had many many more be happy with their own internal analysis and increase budgets, then run another test at that budget level. I have clients who only pay us if we spend $1 and they get $2 in incremental revenue from users who their data science team would not have purchased without the ad. Those teams don’t care if they spend money, and they even get bonused on increasing incremental revenue so they constantly test new vendors and move spend to whoever performs the best.
Nothing there anyone can check which makes it indistinguishable from "I know more than you and I can tell you you're wrong." That may even be correct for all I know, it's just not at all useful.
Which non-internet company is on record claiming they see better returns from google ads on racist blogs than on say, tv ads for example? Must be a few, surely. Where?
What brand ever advertised on google? Pick a good one. I want to see the campaign.
I've had this discussion before and it's usually at this point of wanting any data or examples that it sort of melts away like morning fog in the sun or gets needlessly nasty. Hopefully this time will be better. It's not like I'm trying to push something here.
I’m not sending you client data, just like any dev on here would not send their company’s data. Campaign architecture is kept secret as competitors/etc can use that info to their advantage.
I’m explaining how an entire industry works and you’ve just closed your eyes and said everyone is lying. It doesn’t make sense to assume everyone in an entire industry is lying - then that industry would be dope for disruption as any brand not spending marketing dollars would be hugely more profitable than competitors. Show me a startup that doesn’t have an advertising team/agency.
Your client data? No never asked for it. Any data, from anywhere relating to this, ie showing google ads success.
Google publishing success stories maybe? Anything.
Conversation works better if you don't claim people said things they didn't say. I guess we're going the second route? Seems a shame.
But sure for disruption we have Tesla - brand strength to rival Mercedes. Ad spend $0. We have Trump, became president. I strongly doubt google ads were the key part of his presidency push. Both of those marketing campaigns were pretty disruptive.
Never seen a google internet ad for the google pixel. Seen plenty of tv ads. That seems like an interesting data point? Lots of tv ads for chrome in my sport among the gambling ads too.
I'm looking for data as evidence of the claim made by google that paying them is worthwhile which I don't see much of.
So yeah you can update your prior with "people pay for it so it must be good" It has some weight. I mentioned something pretty similar right up top when I went through the anecdotal evidence we have.
a nobel prize for it where he announced there were never any negative side-effects or outcomes(!). To as room full of serious and successful scientists. Meh.
Group think really does happen. We find out if it has when there is direct evidence, which we don't yet have here, which continues to be a shame. When it comes up I'm always hopeful someone will say "Yes we do!" and link some good data.
Source?
It’s always extremely funny to read engineers on hacker news to make claims about ads, with no understanding of how the business work. And the underlying assumption that companies throw away billions of dollars away, with no results to show for it.