Comment by tomrod | Hacker News Reader

Comment by tomrod | Hacker News Reader

Parent Full thread

> This is an industry that sells influence at scale.

---

>* Before the inevitable comments, yes advertising works, yes it works on you no matter how much you think otherwise

---

Show me the lift (in other words, prove it).

I'm extremely dubious about any claim related to digital advertising having direct effects over a baseline.

It is easy to make a claim that it changes the millieu, to appeal to second order effects. So, prove it. Show me the academic papers. Show me how those papers generalize outside of the specific economic/operation structure they studied.

I mean, look at Google Search. When searching, the top link is generally an ad. My (admittedly biased "no it doesn't work on me no matter how much you claim it does") experience is that I _always_ skip the ads. Even if it is for the company I am looking to purchase from. Because I am averse to digital ads. I can't trust what the ad is selling isn't skeasy. I am not the only one who does it. So is digital advertising being supported only by people who are easy to 419/Nigerian Prince scam or something?

On a tangential note note, this is Amazon's burgeoning issue with its Marketplace and third-party ripoff sellers, now that eBay isn't on top.

Your alternate hypothesis is that the companies collectively spending hundreds of billions of dollars per year on advertising are not getting anything for it? Why don't you take a swing at proving that?

Folks already have, at Microsoft, Yahoo, and Google.[0] The issue isn't showing lift, it's having sufficient power to know the test you ran isn't cherry-picked or BS.

[0] https://courses.cit.cornell.edu/jl2545/4550/materials/lewisr...

Also, the onus to prove the claim that digital marketing has an effect is not logically on the interlocutor, but on the claimant.

Knowing many marketing folks across many industries, their incentives aren't usually tied to incremental lift but rather gross channel volume. So if they can cannibalize from more effective sales or advertising channels they do so, without qualms.

I've heard the claim like grandparent comment has made before for years. Yet, when pushed, it is suddenly a golden cow. This is concerning, precisely due to the figure you cite of billions of dollars.

What he is saying is, the fact that hundreds of billions of dollars get spent on advertising IS the proof.

That's the evidence. Companies don't waste money for no reason. And so the only reason that they could be spending this amount of money is because it works.

So yes I'd say that this type of argument counts as a "proof".

It's not really any kind of proof. I don't want to clutter up the thread with identical content so I'll just link you to my reply.

https://news.ycombinator.com/item?id=16674883

I agree entirely that it's a total waste from the societal perspective (which is why I say we should ban it). But it can be very effective from the individual advertiser's point of view. That's true of any arms race.

I believe that tomrod is claiming the latter is also false.

> Also, the onus to prove the claim that digital marketing has an effect is not logically on the interlocutor, but on the claimant.

No, not really. You're making the extraordinary claim that it has no effect, that millions of businesses are independently making the same mistake.

No, I'm asking for proof that it works over alternative baselines. If people already decided to purchase a product offered, the ad sets up a costly, unnecessary channel.

As was stated by another in this thread, prisoner's dilemma leads to unintuitive economic outcomes.

If you're saying it's an arms race, sure, nobody's denying that. But that doesn't mean that it doesn't work, it just means that in general and over time particular markets are mostly at equilibrium.

Advertising is a classic case of the prisoners dilemma.

In short: Advertising is very effective and worthwhile if you're the only one with a product in that space. As soon as others enter it just becomes an arms race backed by mountains of money with questionable results.

https://en.wikipedia.org/wiki/Prisoner%27s_dilemma#In_econom...

Why spend all the money on questionable results? At risk of coming off as glib, perks. I worked in a field tangential to advertising and all of the money is very useful for throwing parties, signaling success and status, being able to invite clients to large events thatyour advertising is involved with (think superbowl), etc.

For some reason HN at large prefers a more conspiracy leaning attitude toward advertising which you can see in the top comment right now: "Anyone with a credit card can start changing how people think and act." This perspective believes that advertising is like a brain ray that persuades people to do whatever they say - that targets have limited agency over their own thoughts. It's very appealing to assume everyone is a gullible sheep (except yourself, naturally) but that's not the reality behind advertising spending.

That doesn't seem like a sufficient story to explain most kinds of advertising. What perks do all the local grocery stores get from mailing me leaflets about what's on sale?

That just falls into the standard example with the prisoner's dilemma. My perks example was just explaining what the billions are buying instead of influence/persuasion.

> signaling success and status

You already have the insight. "Signaling" is just as much advertising as anything else. It's well-studied and the money spent in maintaining status-quo is just as important as growing a new product.

Thank you, this is a fantastic encapsulation of my point.

The proof that digital advertising does anything but set up a costly lead channel is still a bit lacking, and your plausible assertion shows why that may be the case.

What's the "baseline" you claim? You can prove it yourself by buying some ads. 24 hours and a few thousand dollars and you can sell pretty much anything (which is a much greater signal than just views or clicks).

The obvious explanation to that is that it works, unless you want to call it magic. Also you probably do not realize how much advertising you are exposed to everyday, far beyond some banners or search results. It's never that simple.

Baseline: sale of product sans _digital_ advertising.

I'm not saying digital advertising never works. I'm simply asking for what it incrementally offers over other channels generally.

By digital advertising I'm being very specific: purchased ad space, be it real time advertising networks, sponsored content, and similar. Not "influencers" or other branding efforts for which proving effect is unlikely to be possible.

Further, I can see a difference from launching a new product using digital means versus the (much larger) established companies migrating or updating their marketing portfolio to include digital advertising. But a new launch may be choosing a sub-quality channel. Knife company sales aren't going to be great on digital versus going to a trade show, say, so the margin is what rules the day.

My request is not unreasonable. Show me the proof, the diff-in-diff, the actual assessment of incrementality generalizable to the channel for established companies. Otherwise there are competing theories which equally explain the effectiveness or lack thereof of the digital advertising channel, such as FOMO/prisoner's dilemma.

It either works or it doesn't, I didn't speak about efficacy which varies greatly depending on 1000s of factors. If you accept that it does work then I'm not sure what you're disagreeing about.

Does it improve over the baseline? Yes. It's very easy to see with modern analytics pipelines that report sales in real-time. Ask any growing venture-funded company selling a product or service. Stop the digital ad spend, see what happens. Start it again and see what happens. This has been done by channel, format, campaign and more to do media trials for the largest companies. These tests may not tell you exactly which ad was best but you can derive that this campaign works, or this format works, and can definitely see that the entire medium works. That's before getting into advanced referral tracking and econometric attribution modeling, or even simply asking customers where they last saw the ad.

All these media buying companies might waste money but they aren't in the business of losing it. FOMO/prisoner's dilemma from the other poster is just an interesting perspective regarding signaling, but it is not new. Advertising works but that's completely different from "it works for opposing sides or competing companies to create a net zero influence" which is just a cost of playing the game and a perfectly acceptable strategy. Maintaining market share is just as important as growing a new product and brands spend to signal that they can spend, thereby showing quality and success. Here's more on that: https://en.wikipedia.org/wiki/Signalling_(economics)

I'm not sure what you would like as proof but the Nielsen company has been doing this for decades and is accepted by the industry. Digital advertising isn't a new concept, it's just exposure like it was in TV, radio, print and signage since the beginning. It's just much faster, more personalized and much more pervasive in today's media and devices, and also much cheaper to run.

> I'm not sure what you would like as proof

As mentioned in my prior comment:

>> Show me the proof, the diff-in-diff, the actual assessment of incrementality generalizable to the channel for established companies. Otherwise there are competing theories which equally explain the effectiveness or lack thereof of the digital advertising channel, such as FOMO/prisoner's dilemma.

From your comment, you agree with FOMO/prisoner's dilemma even if you don't agree with the context of the jargon. As you mention, signaling theory is a more proper term for what we are describing, but signaling is efficient if we get to a separating equilibrium (e.g. quality differentiation of product) which clearly we don't see. Ergo FOMO/prisoner's dilemma is likely a better fitting model (though with some admittedly post hoc bias).

So the proof I am asking for is not existence of digital marketing spend, because that isn't proof of effectiveness of digital impact at all and has competing theories sufficient to cast shade on it as full explanation, but incremental improvement over other existing channels (and for established companies, doing nothing at all).

It either works or it doesn't, and you seem to agree that it does, so you're asking how it's better than existing channels? That's completely up to implementation and 1000s of factors as mentioned, and it doesn't need to be better than any other model to be useful for the bottomline, only that it returns a positive ROAS (return on ad spend).

A company selling to younger people might only advertise online while another only sells well through TV commercials. I don't see how that tells you anything useful other than to use the right media mix for your marketing needs.

That being said, whether a medium works is very easily proven by changing spend in isolation and seeing the impact on sales, and it's done by every major company. If spend exists, then campaigns exist, and so a outcome on sales can be measured as compared to the lack of spend (and thus campaigns). I'm unsure sure why that isn't evidence for you.

Reply on news.ycombinator.com