Comment by jasode | Hacker News Reader

Comment by jasode | Hacker News Reader

>Advertisers pay Facebook on the assumption that the people viewing and clicking their ads are real. But that’s often not the case.

Actually, advertisers already know there are lots of fake (Facebook/Twitter/Snapchat) accounts. Likewise, advertisers also know that newspaper & magazine circulation numbers are inflated (even though the circulation #s are "audited"). Ad buyers also know that tv audience sizes are inflated as well.

What matters in the end is if there's a positive ROI on the ad spending. The advertisers can measure the uptick on sales and if the ads worked, they renew their ad spend on Facebook. The majority of Facebook revenues come from repeat business of advertisers who already know about fake users. In contrast, if the majority of Facebook revenue were to come from 1st time ad buyers that were easily fooled by fake accounts, that's when the false user count would drastically affect revenue.

I'm the last person to defend Facebook but just wanted to highlight how advertisers think. For the Facebook ponzi scheme to fall apart, the ads have to stop working. This has happened before. In the 1990s, advertisers were buying Yahoo banner ads. But after the initial novelty of naive web surfers clicking on them, advertisers quickly realized banner ads were worthless. As a result, Yahoo revenues plunged.

"What matters in the end is if there's a positive ROI on the ad spending."

This would make logical sense. BUT! it turns to not be the case. Larger accounts: Pepsi, Johnson&Johnson, Tmobile, etc tend to just take a massive budget and blow it on ads. They don't necessarily care about the ROI or click tracking. They just want to shove banners in front of people's faces before the quarter ends. What is nuts -- is that these large accounts (we used to call it dumb money) are almost all of the ad revenue. Accounts that care about click ROI tend to be low budget -- that is why facebook/google will push these people into self service ad portals.

I started out working as a senior Facebook / social app developer at digital agencies. Saw it firsthand.

Agencies would hire me late in the process, when most of the campaign was specced out (as some silly app on a page). I saw many times that the campaign wouldn’t really be that viral etc. But the design had already been nailed down and it was my job to make it happen.

So what they would do is just buy 50,000 likes and maybe get some organic action. They showed the results, the guys at the brand showed the results to the giys upstairs, everyone seemed happy.

It was one of the things that made me realize current social media platforms were about fake or shallow online engagement, and gave me the impetus to start Qbix

Is that really true for Facebook ads? My feed contains relatively few ads from companies like Pepsi, J&J, T-Mobile, etc trying to do brand building. Most ads are from somewhat smaller companies trying to get me to buy a specific product right now.

Those brands usually try to connect with existing brands so events/people sponserships are more rewarding and global vs facebook ads who generally want you to click now while targeting a smaller segment.. the smaller the better

That highlights something I noticed about advertising - it is ironically meta in that what is needed is to convince the buyer that it is the source of their success.

Many have "totemic" advertising where they think slapping their name everywhere will be what gives them success when they are already well known and widespread enough that people wouldn't forget about them if they stopped advertising tomorrow. Brand awareness is a thing but it seems overhyped - especially when it forgets the downsides of obnoxious advertising.

If as the old joke the moon was painted a cocacola logo that would give brand awareness in perpetuity but it would piss off a lot of people rightfully.

Pissed off people are still more likely to buy that brand because to do otherwise would require a thought override. These ads speak to the subconscious your anger is more conscious and will ebb and flow.

It wasn't a joke and it was Pizza Hut, not Coke.

> Larger accounts [...] don't necessarily care about the ROI or click tracking.

Well, sure. They are going for brand awareness. There's an entire field of study around it.

Large accounts are not enough to keep FB and the like afloat. The small accounts do care about conversion rate, etc.

This is wrong to the point of being backwards (I've worked on building these systems). The most demanding clients from an ROI perspective are _always_ the biggest clients, since they're the the only players who can support an employee whose full time job is to care about this ROI. Seriously, look at the features of the ad portals - they're obviously built for major advertisers. Features are prioritized based on how much business they will bring in, thus large advertisers drive the product roadmap.

Small businesses care about ROI in the abstract, but are generally so time strapped just running their business that they barely have time to even set up attribution or measure ROI beyond "I spent this much and it kind of seemed like I got more business".

Nah, it’s true. Pepsi doesn’t care much about click thru rates on their banner ads.

click through rates != ROI

Pepsi cares about making you read the word "Pepsi" over and over until you buy one, but you don't buy them from pepsi.com, you buy them at every restaurant/store you go to.

That's true, because click-through rate is not at all a good measure of ROI.

Pepsi cares about sales rates of Pepsi products in $localRegion when they run an ad campaign in $localRegion.

Major multinationals are sophisticated enough to correlate their ad campaigns with product sales. And they have the resources to run numerous ad campaigns simultaneously in multiple regions and iterate on the campaigns that work. Smaller players are less able to do that, and are forced to rely more on metadata like click through rates because they can't see the bigger picture.

Yeah - apparently it's called brand awareness - you see the logo, that's enough.

> What matters in the end is if there's a positive ROI on the ad spending

From what I've seen, the big brands don't really know the ROI of their ad spending. They allocate X% to digital, Y% to outdoor, Z% to TV, etc. Then they sub-allocate across channels within each category. It's not really about ROI.

And smaller advertisers don't have much in-house capability to measure ROI. Instead they rely on Facebook's own analytics.

Only a handful of companies I've seen actually build their own analytics, measure customer acquisition cost, and match acquisition cost to lifetime value for specific customer cohorts.

The smaller firms that do their own analytics invariably discover that 95% of the traffic from "paid social" Facebook ads is low quality, high churn, with deeply negative ROI. And it's impossible to isolate the 5% high quality, low churn, positive ROI traffic and just pay for that.

Basically everything about this is wrong. Big brands invest far more in measurement than small advertisers. Smaller companies who do their own analytics and seriously invest in Facebook ads tend to find that they perform better than all but their most core search keywords.

It's true that "seriously investing" in Facebook ads is much more involved than the equivalent investment in search ads, but it's absolutely false that the traffic from Facebook ads is definitionally low quality.

I don't know about the experience with others, but I've built a few different online businesses over the years and I am always tracking my ROI on digital spend intensely. I have almost always found Facebook traffic to be extremely poor, especially when compared to Adwords. Obviously display vs. search advertising naturally have big differences in conversion rates, but the difference was always way more extreme than I had anticipated, to the point where I've just completely given up on Facebook entirely.

what kind of attribution model did you use?

In general, last-click methods will substantially under-value FB and over-value Google. This occurs because you see an ad on FB, don't click, but search to get back to the website, and click on a Google ad.

I've found, that for a lot of b2c businesses, FB provides extremely good ROI. Google is much better for b2b though.

> perform better

If by "perform better" you mean "generate traffic" then I agree with you. Facebook ads really do result in a lot of clicks.

And if you're running a political or awareness campaign, maybe that's what you want. But it's hard to measure ROI for this kind of campaign except by comparison with other traffic sources.

On the other hand, if you're selling a product or service, and you track narrow cohorts from prospect through conversion, repeat buy, defection and churn, you can compare total acquisition cost of the cohort with LTV of the cohort.

Regardless of the product or service, I usually see CAC >> LTV at the cohort level.

Within a cohort, usually < 10% of customers have CAC <= LTV, but it's nearly impossible to target this subset in advance, so negative ROI dominates the cohort.

Ads that "work" are more a case of "can we show numbers that the people upstairs will consider ok?" than "can we prove these ads have contributed to our bottom line?".

The old "I know half of what I spend on advertising is wasted but the problem is I don't know which half" still applies, despite all the promises of digital advertising.

But there's a gap between what the advertiser knows and what the intermediary knows.

Often even when the advertising account is directly held by the advertiser there is an intermediary managing the ad spend.

And the intermediary is incentizived to spin the numbers as best they can.

>For the Facebook ponzi scheme to fall apart, the ads have to stop working.

I am taking the Ponzi scheme here refers to numbers of users? Because for Facebook's Client, it is obvious that Ads is working and bringing a positive ROI. And these ROI are not paid out directly by Facebook ( which could easily distort the number ) but results measured through other channels. Which means from Facebook's clients perspective it is not a Ponzi scheme at all, as there is nothing to lose from its Client.

I think you have some confusing typos here, perhaps some missing negations?

Reply on news.ycombinator.com