Hacker News Reader

Comment by mft_ | Hacker News Reader

What a terribly frustrating article.

1. It doesn't satisfactorily make the case that targeting the audience rather than the topic-related websites doesn't work or make sense. (I'd argue that a deeper problem is that most simple online ads just don't work, for most people. And otherwise, if you're actually susceptible to online advertising, then actually getting served ads tightly targeted to your interests does make sense.)

2. I'm not sure that the data used (in support) supports the author's case that people stick to a small number of their favourite websites; rather, it looks like it shows that popular websites are popular. (Hard to be certain without examining the market research questions though.)

3. It then takes a tangent to show that lots of ad exchanges are unreliable; this may indeed be true, but is totally unrelated to the original premise.

View on HN

Judging from what kind of targeted advertising I get, pretty much any method would be better. Topic method would at least made sites I visit less annoying. I am totally wanting to investigate anti bed wetting method (or other kid related targeted ad) to buy when I am visiting sci-fi web or gaming web page page or ... get the idea.

Judging from experience if people I talked about it with causually, I am not alone nor special.

Absolutely! In the past year, I've been doing more computing on a tablet and I don't block ads on the tablet. I am logged in to a variety of sites and make zero effort to hide my activity.

I started paying attention to the ads about six months ago, because I'd manage to notice how poorly targeted they were when I did see them. They are terrible.

I think the best targeted ads have been from a company that sells barn doors. Really, they are in fact selling barn doors online. That's a very compelling ad for me. Except, you know, I already have doors for the barns. I am not actually in the market for new barn doors.

I don't actually need a bra, free for thirty days. I don't have breasts. The missus doesn't use my tablet and actually has her own network connection as I tend to saturate mine sharing Linux isos and stuff like that.

I don't need a used car. I have more than adequate automotive sources already. I'd buy a used car, but it would not be a Toyota that has just come off lease. Maybe try selling me a 900S Turbo from 1992?

I don't have any major medical problems, trying to sell me drugs is pretty silly. I don't need insurance, I'm already well insured. I'm not concerned about paying the least amount for insurance, I'm concerned with the best coverage and service. An ad for, "We are dirt cheep but just shoot you if you're involved in an accident insurance" is not very interesting.

They have vast amounts of information on me. Why not try to sell me a new plow? I'm kinda in the market for one. Why not something to do with hunting, fishing, gardening, or food preservation? Not one ad has has any of those things. They know I own a Kubota, why not advertise some new implements?

I don't get it... I'll buy stuff. I'm more than happy to buy stuff. I love buying stuff. I buy stuff I don't need - all the damned time. Yet, literally no ads have compelled me to click and make a purchase. I have clicked, I just haven't found anything worth buying. I've never bought anything from a placed Internet ad, except when I was searching specifically for that product.

Something is broken.

>Something is broken.

Perfect summing up.

Have you tried to go on your ads preference from Google Ads?

I never had bad targeted ad but I did update that page maybe 3 times in the past decade, that's maybe why.

I have not. I will ponder the implications. I kind of want to see what it does without my prompting it.

Ah, retargeting. Do a search once, be followed forever...

>Ah, retargeting. Do a search once, be followed forever...

Yep, what I do normally first thing on monday morning is to do a google search for "drone" (not that I am interested in buying one, mind you, but they are actually nice little thingies).

This way - at least up to Wednesday - I see mostly ads for drones.

On Thursday I make a search for "led lamp", just like KGIII reported for barns doors, I have no actual need of any led lamp, but it is nice to have (mostly) "known content" ads.

Cool idea, thanks!

> It doesn't satisfactorily make the case that targeting the audience rather than the topic-related websites doesn't work or make sense.

I've been saying this needs investigating for a long time. From my understanding of things like triggers, priming, threshold for action, etc -- you're probably doing yourself a disservice having out of context advertising.

When you advertise in context, you get constructive interference between the context and your message, leading to a more powerful effect -- your buyer was ready to hear what you wanted to sell; when you advertise out of context, you get destructive interference between the context and your message, leading to muddled or even negative effects -- advertising roast beef on a porn site is probably bad for all three parties! (No matter how much the person viewing porn likes roast beef.)

"Targeted" rather than topic based advertising ignores basic facets of human psychology (eg, that we're very context based animals) in order to pump up ad stock and sell more volume of a diluted product -- because all of the "targeting" is fundamentally a worse replacement for the signal shown by visiting a topical website. It's about inventory fluffing, not efficiency.

That said, the article didn't do the best job supporting that point.

Here's my view: advertising industry, being mostly about manipulating and scamming people, is even better at doing that to itself than to people outside. Between customers and the company that seeks them, there's so many ad-industry providers now that it's easier for them to manipulate tooling and metrics than to actually deliver in an efficient manner.

The ways I've personally seen this happen mostly boil down to marketing people with absolutely no understanding of statistics showing "metrics" to their customer, which "show" that they're indeed doing a great job. But since the customer has no fucking clue about statistics either (why should they, that's why they're hiring a marketing company, right?), they believe that - and they have no direct way to verify the effectiveness anyway. I.e. incompetent party selling bullshit to another incompetent party, but as long as both parties believe things are going in a good direction, money changes hands, and everyone is happy.

And that's just double incompetence. I suppose you can excuse people doing divination from whale charts. But how many similar things happen when the marketing provider party just straight exploits their clueless customers?

Also some thoughts on this topic from the times of Optimizely debacle: https://news.ycombinator.com/item?id=10873226.

Reply on news.ycombinator.com