Do Not Advertise On Facebook Until You Read This | Hacker News Reader

Do Not Advertise On Facebook Until You Read This

wahanegi.com

123 pointsslaven47 comments

I wrote this blog posting. I am not claiming that these are bots or that Facebook is defrauding us in the strict legal sense. After a bunch of research, I believe they are real people, as Facebook asserted to me and to the BBC among others. However, I believe Facebook is misleading advertisers and behaving unethically by not disclosing this fairly widespread, unusual and costly user behavior, one that hits the campaigns of less sophisticated small advertisers particularly hard.

The problem has two main parts:

1) The click behavior of these users is _extremely_ unusual and unexpected according to the beliefs of the advertisers and advertising experts I have shared these finding with. Most people think that 'no one clicks on FB ads.' So when they learn there are profiles of real people that click 5-6 likes in a minute, dozens of times a day adding up to thousands of times a year, they immediately exclaim 'no one does that, they must be bots, you must be wrong.' But they are real people. And unless you are extraordinarily careful (see here: http://wp.me/p2dKfK-7x), Facebook naturally charges the same for their clicks as for everyone else, even though their clicks are nearly meaningless from an advertiser perspective.

2) These real people were taught this behavior by the new Facebook timeline interface, which a) makes the number of 'Likes' extremely prominent, b) encourages rapid liking of ads by refreshing the ads immediately as they are clicked and c) uses these likes to change the content that appears in the stream of stories.

Perhaps a better one of my postings to read is this one, where I estimate how widespread the problem may be from an advertiser perspective http://wahanegi.com/10-percent-of-fb-revenue/. To save you time, here is the main bit, a quote from one of these users after I asked them why they like ads in batches of 5-6 throughout the day:

"Yes, I do remember why I liked things in batches. Facebook suggests things for you to like in the right column on some pages. As soon as you click to like one of them, it replaces it with another suggestion. I’m quite happy to like thousands of things on Facebook as it improves the kind of stories and ads that come up in my news feed and again in the right column. I would rather see things I am interested in than things I’m not."

Makes sense for him. But not so much for a small advertiser who just paid Facebook a few hundred or a few thousand bucks so that Facebook could use the ads like click-bait for: people aiming to hone their Edge Rank content preferences, people who are bored and enjoy choosing which ads they like best in the list, or people who want to get their Likes 'score' higher because it gives them a sense of accomplishment.

Trust me, once an small advertiser learns that there are two types of people on Facebook, and they get charged the same amount for clicks from each, they are not happy. Especially after Facebook customer service gives them the Heisman.

;)

I don't know if things are as clear-cut as you posit, but I definitely appreciate the research and attention you've put into this.

Whether it's ethical or not, everyone is coming to the realization that Facebook ads suck. There have been many explanations, mostly of the hand-wavy variety surrounding the intent of users on a social network vs a search engine, but this is the first theory I've heard about how the mechanics of Facebook itself are creating specifically low-value classes of users.

Now obviously you're never going to get an honest response from a customer service rep as their hands are tied. However, if Facebook are smart, their top ad guys are deep at work investigating your claims. Being able to increase the real value of clicks is absolutely essential for the future growth of Facebook's ad platform. If they jealously guard the information gap, then it will hurt them badly as everyone comes to their own conclusion that Facebook ads are worthless. They need to make sure that doesn't become the conventional wisdom.

I am confused: this seems like a separate issue. If you post an ad for a Page, you get charged if the user Like's it, even if they don't really pay much attention. However, the whole point of a Page is to be able to target users with content later: those Like's from people trying to train their newsfeed are actually the very people you were trying to find. Meanwhile, for ads that have a normal landing page, you should be charged only if the user actually clicks the link in the ad and is taken to the destination: that is not a "Like" action.

This is a new spin on an old phenomenon. For banner advertising, a tiny percentage of the population (< 10%) contributes a massive percentage of all display ad clicks (> 80%). She's disproportionately a lower-middle to lower-income older American female in the Midwest who plays the lottery and subscribes to Reader's Digest. Totally not joking. A word often used to describe this segment is "compulsive clicker."

Online publishers are well aware of this and, well, do not go out of their way to highlight this fact on their rate cards. Instead they'll provide their audience information and often do not mention that they know their typical click is not representative of their audience. Given that they're the "whales" of the CPC business model, they're frequently not quality-priced down to zero, unless they so dominate an advertiser's campaign that the advertiser complains.

There are related issues with almost all online advertising. One should be very attentive to the default settings at one's major advertising network of choice, because the default settings almost always group the prime rib cut of the attention market with the pink slime and sell it at prime rib prices. (Examples: there are strong differences in pricing with regards to user location. Many advertising platforms, ahem, make it very straightforward for American advertisers to advertise their wares to users in developing markets. All you have to do is not opt out.)

What is the reasoning behind this type of behavior? Is it that she is very susceptible to the techniques used in banner ads (animated dancing mortgage people, "one weird trick", etc.)? Or does she have some OCD behavior of clicking all the links on a page?

In the OP's case, I can see the motive of the Like-collecting mindset, but with the compulsive clickers, I'm imagining someone who clicks on ads for no reason other than a compulsion to "complete the page".

Could simply be that they are less familiar with the internet, people who have been using it for a while learn to filter out the banner ads to a certain extent. If you are less familiar with using search engines etc, clicking big obvious links on pages might seem like the easiest way to "surf".

I imagine if you looked at one of these people's PCs you'd find at least 6 toolbars installed in their browser.

My wife grew up in Oklahoma, and there are several women in her extended family who fit the profile. I've been watching how her family interacts with technology for over a decade now. The behavior cited in the article makes a lot of sense to me.

First, a warning: it would be a big mistake to think these women are technologically backwards. They _love_ computers and the internet. They were installing wifi routers years before my own East Coast elite-educated family. My wife's sister has 5 iphones for 4 family members, half of them are jailbroken.

Part of what they love about the net is that it makes them feel _modern_ and _included_. They're no fools. They see the money, the power, the TV all emerging from NYC and LA. They know that the most their town has produced is a country singer with a chip on his should (Toby Keith).

Going online, by contrast, feels cosmopolitan, sophisticated, like they're "in it," in the modern world. (There's a similar phenomenon with chain restaurants: Oklahomans _love_ chain restaurants, because it's the same meal you could get anywhere.)

So having heard all the buzz about the net, they go online and ... well, here the paths diverge. The younger and/or savvier become sort of power-consumers. Like my sister in law with the jailbroken iPhones. She can quote you all the specs and all the industry-celebrity gossip, and there are at least a dozen USB devices hanging off her computer.

But the older and/or less savvy, they equate the net with buzz, so when they go online, they go to its buzziest parts: banner ads, scams, get-rich-quick schemes, "empowerment" classes, etc.

They click the banner ads because they think that's what you're "supposed to do" on the net, and doing so makes them feel a part of the modern world. They're not stuck in a fly-over state, they're right in front of the busy modern world, right there on their computer screen.

This makes sense to me, I delete all my cookies and sessions frequently in my browser and usually browse the internet logged out.

At least 50% of adverts I see are for either weight loss products or anti-ageing products.

Personally, on the rare occasion that I do see an ad that I find interesting I won't click it (due to an ingrained fear of losing the page that I am on) and will simply open a new tab and google search for the product name. Not sure if most "internet savvy" people do the same?

I also wonder how well your average holds across websites. If you are selective on the websites you advertise on, say advertising a programming product on Stack Overflow then I wonder if the majority of your clicks still come mainly from the midwest moms who ended up on SO by mistake?

on the rare occasion that I do see an ad that I find interesting I won't click it

With Google, they at least give you a URL below the ad. Sadly, about half the time it goes to a non-existing page so you have to search around anyway.

I refuse to intentionally* click on ads because I ran some adwords campaigns years ago. Seeing Google charge you multiple whole dollars for a "click" feels like one of the largest scams in the world. You have to trust that: they are never lying to you, they aren't running a shadow click-happy network, and they monitor and refund bot clicks. Each click gives google money essentially for free. I'm not fond of working for Google for free.

* Google keeps decreasing the contrast difference between the top row text ads and actual search results. If my laptop screen is tilted the wrong way, I can't see the shaded "THIS IS ADVERTISING" box. Each mistaken click there gives Google $3 to $25+. Just for one click of the mouse. The mind boggles.

This reminds me of the ethical dilemma I face whenever I search for something and the site I'm looking for comes up in both the organic and paid search results.

Do I like them enough to save them a few pennies or not?

Flipside: Do you like Google and all their ad-supported services enough to click and help pay for them?

I've worked on many SEM campaigns over the last 5+ years spending millions on Google and calling it a "scam" seems a bit extreme for the following reasons:

1) Very few categories have CPCs upwards of $3. If that's what you were paying, it's either that you had a poor quality score (hence high opportunity cost to show your ad according to Google) or it was a category that was competitive and had the economics to justify a $3+ CPC (simple economics).

2) Google has extensive bot traffic and fraud prevention to ensure ad clicks are not coming from fake people. You tend to find much lower quality clicks when you opt into the content network but from the search network, I've never seen much fake activity (looking at logs/analytics, etc.).

3) It doesn't even matter what is being charged by Google, unless an advertiser is completely naive, they should be paying the cost that makes sense for their business. I don't care if I pay $25 per click if I can make that back overall. Similarly, even if a portion of the clicks are fraudulent or bots, you should be adjusting your bids downward to reflect that.

I am so fascinated by the idea that these people exist. I would love to read an interview with one of them.

Odds are that you have one in your family tree.

Here is a quote from one of them:

A little different than the typical idea of a Reader's Digest granny who doesn't understand the internet. Unlike the problem with display ads, in this case Facebook's UI purposefully encourages clicks based on user intent that is highly unrelated to the ad itself. Plus, as a FB advertiser you can actually see the profiles so you can see that these folks are heavy-duty clickers...and if you check the numbers carefully, you can also see that FB invalidates some of the clicks from the folks who like your page, since at some point the behavior is clearly indefensible. However, FB still leaves them attached to the page, since in the end FB can't tell if they really meant to click on your ad or not. And they don't appear to go back and invalidate previous or subsequent clicks from that invalidated user. They just trim off a few.

I consider this unethical business behavior on the part of Facebook. And since I also believe in caveat emptor, I wanted to make sure to get the caveats out there for any emptors who are interested.

I haven't advertised on Facebook myself, but this is is an interesting concept. If it's broadly true and that Facebook essentially has two "classes" of users, then it's up to them to price them segment and price them effectively for advertisers.

This reminds me of Paul Graham's essays about Viaweb's acquisition by Yahoo[0][1], and how Yahoo seemed mostly uninterested in PG's algorithms for Revenue Loop. To quote: "In 1998, advertisers were overpaying enormously for ads on web sites. In 1998, if advertisers paid the maximum that traffic was worth to them, Yahoo's revenues would have decreased."

I remember when AdSense ads used to be images and Google replaced them with text links, essentially reducing the clickable surface area of the ads. This essentially slashed revenue in the short term with the theoretical long-term benefit of reducing accidental clicks and increasing the value of the clicks they did send to advertisers, who should then be willing to pay more. As a now public company, it will be interesting to see which path Facebook follows.

[0] http://paulgraham.com/6631327.html [1] http://www.paulgraham.com/yahoo.html

I seem to be greatly misunderstanding the issue here, and thus I apologize in advance if this seems like a massive, and possibly dismissive oversimplification but the complaint here is that there are overeager users who are clicking on ads, not knowing what they're clicking, and because of this the author/campaign manager is owed a refund?

Unless Facebook, using the logs they've requested can conclusively prove that these are bot profiles aimed at disrupting the Ad Platform (which is something I would assume Facebook would already be keenly aware of and taking excruciating steps to stamp out, as Ads play a big role in their bottom line), why exactly does the author feel he's entitled to a refund just because people click the ad who might not be interested in the product?

Again, unless Facebook can show these were malicious bots or that their targeting algorithm has malfunctioned and displayed ads to profiles that do not meet the demographic criteria setup by the campaign manager and thus, ads are being displayed to users they shouldn't be and resulting in erroneous clicks, what's gone wrong?

If Facebook has developed a culture of rampant ad-clicking to further some other cause unrelated to your advertising then those ads should be valued accordingly.

Imagine if in some Zynga game they had ads but rewarded users for clicking on them. Not interacting with, but simply clicking. Would you find much value in advertising there at all?

Imagine if in some Zynga game they had ads but rewarded users for clicking on them. Not interacting with, but simply clicking. Would you find much value in advertising there at all?

You've actually brought up a very valid point; I kind of hinted on this in my OP but it deserves some clarification: has Facebook proven so far that these ad clicks are in some way spurious? Granted, knowing what motivated the clicks is an impossibly tall order; however I don't see much difference in what you've proposed versus the many Pages that encourage users to promote more followers in promise of some sort of a rebate/discount/incentive with a brick and mortar establishment.

The primary and most obvious difference here is that there isn't a cost simply for putting up a Fan page. However and ultimately, my original premise I think deserves just as much consideration as incentivizing ad clicks to boost leads: unless Facebook can show that these clicks originated from a source and can sufficiently extrapolate evidence that these sources are motivated to disrupt the platform, I do not understand the author's complaint.

Because the clicks by these "clickants" are worthless, and shouldn't be sold. Somebody blindly clicking on ads has no value, and you shouldn't be selling their clicks. It's akin to selling advertising on TVs that happened to be fast forwarded/skipped on Tivo. If the advertisers are aware from DVR logs that fewer people saw the logs, they need to reflect that in reduced costs to the advertisers by reducing bill-rates over time.

The Tivo analogy makes sense. Thanks for clearing that up, I figured initially that I was misunderstanding the point of the post and your analogy fixed it.

I don't think the author is saying the clicks are malicious, just that they are lower 'quality'. Based on the authors experience he seems to be saying the intent of some facebook clickers is different than regular web link/banner clickers. He seems to imply the Facebook clickers have some other incentive to click other than to see the advertises website; and that incentive might be to collect 'likes'. And that actually makes sense to me, reminds me a bit of the incentivized clicks of the eagerly 00s, not worthless just worth a whole lot less than organic clicks. If Facebook has a lot of these clickers their inventory value becomes worth less.

Of course they were of lower quality, they were intentionally running their ads to people from the "Philippines, India, Pakistan and Bangladesh".

I suspect at least some of these "booklicant" users are people attempting to build up fake profiles that look real.

They do this by liking a lot of advertisers, which makes their profile look more populated.

Thank you, I couldn't think of a scenario where individuals would want to automate ad clicking on Facebook, but this makes complete sense.

Very good point, I think this could also explain why a bot would spent time and resources "clicking" ads (I believe there was another post HN where it was suspected 80% of the click came from bots).

If you post an email chain, please, post the messages in timestamp order.

It's a little annoying when you need to scroll all the way down to identify the beginning of the discussion, and then have to read from bottom to top to see the exchange as it unfolds.

Sorry about that. At the time I posted this, I just wanted to get it out in the world and move on. I felt (and feel) that Facebook had already wasted enough of my time. If I knew that ten thousand people were going to read it in a single day, I assure you that I would have spent more time editing it to give you a better experience.

I read though this missive - and the first thing that came to my mind was the definition that "more than six clicks a minute is considered invalid" - and I tried to recall if there was a time in the past year that I clicked on six display ads in a day.

I then realized that, when I'm searching to purchase something on Google, I may click on half a dozen ads in a minute - but realistically, each of those has a chance at getting $20-$40 worth of revenue (and the potential, as in the case of monoprice, of hundreds, if not thousands of dollars of revenue after I discover the site)

I'm wondering if anybody has done a definitive study on the conversion value of a "click" on google versus Facebook, both in terms of brand recognition, immediate conversion (purchase), and long term conversion (delayed purchase) - that study, if definitive, would probably adjust GOOG/FB by 10s of billions of dollars - clearly a leading indicator of future revenue.

Probably not, but it sounds like a fun weekend project. I think you just gave me something to do this weekend :D Thanks

That is a great point. If anyone does want to do the calcs ;), it's important to keep in mind the difference between what you do on Google when you are clicking around 'searching' for content, and what happens with these folks on Facebook. On Google, your intent is relatively clear since the advertisers bought ad space against _search_ terms...not perfect, but pretty good. On the other hand, Facebook is claiming that they can do an amazing job of targeting customer segments...which may be true, but it is also mostly irrelevant in this case since the intent behind the behavior is the only thing that really matters in cost-per-click advertising that costs more than a fraction of a penny per click.

I believe the market will take care of the price once it understands the product that Facebook is selling - at least I hope so. In the mean time, Facebook is fleecing a bunch of advertisers, especially small ones.

> "when I'm searching to purchase something on Google, I may click on half a dozen ads in a minute"

People really do that? Not search for some specialized site and ask/search there but actually click on ADs to buy useful things?

Absolutely -a company that actively wants to pay for me to see their existence when I'm searching for somthing is a company that actively wants me to buy it there; that's often a valuable indicator that they're worth considering as a supplier.

I'd say when I search for something I'm intending to buy on google, somewhere between a quarter and a third of the time I buy it via an adsense link rather than an organic search link. Anecdata worth every penny you paid for it, as always, but I've said this conversationally to a fair few people and the majority have said something more like "yeah, me too" than "people really do that?"

My brain algo is something like this:

If I know exactly what I want I search in Google and might or might not click on the first result (the ad) since sometimes the ad is exactly the same as the first organic result (and sometimes with a discount because you go to a landing page instead of, say, the home page of that company's site).

If I want something but have no idea which company I want to buy from I usually do look into the ads first and click on those that seem directed at me (because, as you said, they actively want me to buy from them).

Just as another anecdata point, I'm in the 'people really do that.' group.

When I'm researching something to buy, I will usually do searches on google for broad scope brands/products, then move to Amazon as my first option for purchase, provided the item is available there and within a few % of the lowest online price I've seen elsewhere.

My research may include reading blogs/forums, and possible asking questions there. If its not something too esoteric, I might ask my FB friends for an opinion or suggestion.

At no point in my normal buying processes does FB ads, or 'Likes' play any role.

My wife follows roughly the same pattern, but may be partially influenced by my approach, though I will say her methods differ slightly enough that they are her own.

People do this all the time - there are whole advertising optimization companies built around doing it automatically.

You won't usually see public numbers for it (because it is pretty site/product specific), but the IAB has some public studies on the display/brand recognition side.

It'd be an interesting study, although I think you're comparing apples and oranges - Google is direct response whereas FB is banner advertising. The level of intent in Google queries is high. The level of intent in a user seeing a FB ad is still largely an unknown - with FB arguing that the information they have about a user indicates intent - but to me it is the difference between targeting based on intent and profile.

I have a feeling if this thing does not stop spreading, then expect Facebook stock to drop by 50% on their next earning report...

Advertisers will cut their budget and that will kill Facebook revenue. Well, maybe start a short position?

This is how you do it:

(a) Estabish what you spend on ads per visitor.

(b) Establish what you earn per visitor coming from ads.

However, Facebook should have a look at this issue. It is the reasong why Google's AdWords is so successful, and most other self-service ad programs fail. AdWords is very good in filtering everything that does not represent a value for the costumer (the advertising web site).

Can anyone here tell me if this bot problem is common on other ad networks, or is the 80% bot click rate much higher on Facebook? Because maybe it's just a common hard problem to solve for ad networks. Or maybe it's Facebook who specifically have a problem. If it's the former, then advertising somewhere else wouldn't solve the problem for advertisers. But I cannot know for sure without knowing if this 80% bot rate is common in others networks too or not.

I do know that many people reported lower conversion rates on FB, but they usually attribute that to lower purchasing intent from real FB users, not from higher bot rate. Or are they simply mistaken?

It's not common on other networks or on Facebook. This is likely an isolated incident. I've been advertising on Facebook for months and >80% of the clicks show up in my own stats, not <20%. If advertisers were regularly being billed for 5x the ad clicks they actually received, this wouldn't just be coming out now.

I've done some minor advertising on Facebook too, and I've never seen anything like this either.

This article isn't about bots. The 80% bot stuff was from a different article posted yesterday.

That first email is 2,500 words. I'd be really impressed if anyone from Facebook actually read through the entire thing. It's appropriate for a blog post, but not for an email.

When it comes to the advertising giants, it's not a bad strategy to write a blog post but send it as an email first. Realistically, no matter how long or short your email is, you're unlikely to get a helpful response from either google or facebook to your problem, especially if that problem is on the scale of "You need to change your entire ad algorithm (and also give my money back) - unless you somehow manage to turn the issue into a PR problem for them, by, for example, getting it on the front page of hacker news.

You're more likely to get on the front page of hacker news if you can show that FB is aware of the problem but doesn't care to make any changes, which you can do by posting the email exchange. And you're also more likely to get the HN readers interested in an email that's basically a meaty blog post than in a short exchange that goes "Your algorithm sucks and I want my money back. :("

I don't know how FB currently charges, but if it's a flat rate per click, I'd think a reasonable solution would be make it proportional instead: this user spent 100% of their clicks during time period X on you, whereas that one spent only %10.

Then you would price based on the share of the user's attention you got, not on individual clicks.

Seems a lot simpler than retroactively invalidating clicks from users who pass some arbitrary threshold.

I doubt many businesses have real direct measured ROI from their Facebook ads. Right now people are still happy to experiment with brand/ community building. Eventually though without that direct ROI and the social hype everyone is going to give up.

There are a lot of businesses in the world to churn through but eventually, as a company like groupon for example are probably seeing, this drys up.

Reply on news.ycombinator.com