Hacker News new | comments | show | ask | jobs | submit login

Very well-written response. But I disagree with the answer to question number 3. Facebook can predict what you will like in the future based on what you like in the past. You are a predictable person. What you liked last week, you will like again next week.

Let's say there is a particular friend who you always like their posts or comment on their status updates. This friend posts once per day, and you comment on that post every time.

Facebook will show you that friends post each time. That friend is obviously important to you.

Let's say another friend posts 20 times per day, and you never like or comment on their posts. Facebook will hardly ever show you those posts, except when they think a friend that you follow closely has liked or commented.

To take it to another level, let's say you use the Facebook Check-In function to check into coffee shops and you Like Starbucks. Do you think Starbucks posts will appear more often in your feed? You betcha. You've shown interest in coffee.

And if you make a status update using the word coffee?

You get my point. Facebook knows more about you than Google does. They are doing a pretty good job of making the comment about search intent (which Google owns) moot.

To continue on with this train of thought, the greater point here is that not all likes are created equal. Facebook's system will understand the importance of a like you make.

Let's say you continue liking posts on your favorite news site, they can learn a lot about your interests. However if you arbitrarily like a page due to an ad, it may not be as much of a signal. Facebook can test your interest again by subtly displaying a post related to that like you made and see if you respond to it. Enough non-responses cancels out the value of that initial like.

The point is: to think that Facebook is only using a "like" on the surface to optimize the feed is significantly underestimating the company.

I am not saying it is only using likes like that :)

But to think that they have somehow solved the problem of intent is significantly overestimating the company.

And they still haven't solved the problem of giving me the right advertising in the right moment. They are just ads based on my interest.

It's a very very different thing than what google are able to do.

Facebook hasn't totally solved the problem of intent, but it has gotten damned good at solving habit, and people are creatures of habit. Bear in mind that Facebook's algorithm can detect a lot more than just the topics you're interested in, and the sources you're most interested in receiving those topics from. It can also detect your general susceptibility to advertising, and your purchasing habits based on your uptake rate of ads, your outbound traffic from sales links and ads, etc. From there, it can put various pieces together.

Is this the same thing as capturing in-the-moment intent, the way Google can with AdWords? No. Is it as powerful? The jury's still out. What Facebook does certainly isn't as elegant as what Google is able to do. It's more of a brute force solution to capturing your intent, whereas Google goes right to the point of action. But Google's approach has its weaknesses, too: it can't influence you upstream of your search. (Facebook can, at least in theory).

Traditionally, people have thought about online advertising as very transactional, situational thing. You're served an ad, and either you click it or you don't. If you click it, either you're converted to sale or you aren't. Done. Facebook is playing a different game. It's trying to influence your behaviors before you're served the ad -- both to serve you ads you're more likely to bite, and to stimulate your demand for ads in the first place. It still faces the problem of conversion, and Google is much better at conversion. But Facebook is operating from a fundamentally different strategic equation.

To put it broadly: Google is great at selling you the things you want; Facebook may become great at selling you the things you didn't realize you wanted.

I hear ya but habits are hard to break. So what does that give Facebook that it knows my habits?

I think that is what I am struggling with understanding the value of.

If I know you go by Petes Coffee every day then great I know something about you. Now how is that relevant for starbuck in any way that isn't already solved with knowing I drink coffee.

So StarBucks can try and get me to become a customer at their store. Fine but then we are back at how do they reach me?

They will have to depend on me checking my feed at some point where it's relevant or start spamming my updates tab when I walk by.

That's possible but a very very dangerous game to play IMHO.

With regards to upstream. Google as a working mobile app store :) FBs isn't even close to having solved that issue.

Edit: Saw you edited your comment so here is my edit to your edit :)

With regards to selling me things I didn't know I wanted then I sure don't hope that is their strategy. Cause that is such a big trial and error it's not even funny.

A user is currently valued at around 4USD, it needs to increase quite significantly and with a much harder job then (selling you things your didn't know you wanted).

FaceBook might be working on things we don't know about, of course they are, but OpenGraph to the best of my knowledge is not able to transcend from knowledge about habits into intent. Adding to that the problem with not having the ability to serve me with answers when I want them (especially hard on the mobile unless they want to turn it into a spam channel) then I don't see that as being a viable strategy.

But sure I might be wrong.

Having managed huge sums in web advertising, I have few points to make. Yes intents increase relevance by a large factor for most products or services. Yes, Google currently dominates the lucrative part of the web advertising. But there is still place for ideas and disruptions. And the challenge is to solve the issues exactly what ThomPete fiercely explains.

(1) Some clients need a lot more customers than a first rank ad at Google search may offer in a given time period. Social web sites offer a huge capacity, albeit less relevant. Facebook and social web sites are challenged to innovate and increase relevancy in engagements for different actions they provide and will provide. Likes are a step in the right direction for Facebook impressions.

(2) Sometimes ads are crafted in a way to create intent, taking relevance factor into account, in case ROI calculations make sense.

(3) Search is already the most exploited channel by majority of businesses and costs are already too high especially for low-performing competitors with smaller lifetime values or low profit margins.

(4) In conversion-based ROI focused campaigns, the click prices ideally reflect the end results of everything you discussed here and much more, in an efficient ad market.

(5) Advertisers are interested in new channels. And they would be interested in targeting their FB pages likes and your FB pages likes equally.

I agree, very insightful thnx.

If I where Facebook I would rethink Calender.

Turn it into a "future timeline" and add goals, to-do's allow for easy calender integration.

That way they might now own actual intent, but future intent.

Going on a trip? (find cheap tickets or hotels) Loosing weight? (here are places, products & diets) Going to run a marathon (products, books on training, websites to join) Planning bachelors party? (events) Bought a ticket (add to calender and propose things to do before)

Then their data would begin to be interesting and they would own a type of intent that Google doesn't (Because they mostly lack the proper context).

Right now the calender is primarily around parties and conferences. Expand that then at least I think they would have a chance of providing relevance and would have better time finding out when to present ads.

I see a few attempts but they need to re-do it completely.

The problem is much bigger than that.

With google I do a search and get result. In other words there is a feedback mechanism that provides me with opportunities right there and I want the result (thats why I am searching)

FaceBook does not have this mechanism. Instead they have to find gaps where they can communicate to me. However, I might not be interested in any answers/suggestions at that time.

That I "like" StarBuck does not mean I like mcDonalds btw and it does not by any metrics mean that I want any of them at the time I read the ad.

So I don't believe they can predict and I know they can't serve the "suggestion" when I need it.

Google does not even need to know what I am intersted in. I can one day decide to get something completely out of left field and it can serve me answers based on that.

Facebook is still just interest based advertising it's not intent based.

And if they get enough 'Like' buttons on pages all over the internet, they can correlate against your browsing history. If you suddenly start looking at lots of pages about vacuum cleaners, the intent becomes pretty obvious.

That's sub-optimal compared to what google can do.

You have to realize that FB have to increase the value of each user each quarter (currently 4USD per user)

But they have to do it with an advertising model that is only slightly better than tv ads.

But the thing is, if I check in at Starbucks all the time, and I "like" their facebook page, etc, what's the point of showing me their promoted posts? I'm already buying their products.

How will the fact that I like Starbucks tell Facebook when I'm in the market for a new vacuum cleaner? I might search for vacuum cleaners on Google, but Facebook is going to have no idea that I'm even in the market for vacuum cleaners unless I go and "like" a bunch of vacuum cleaner pages (which I presumably found by searching on Google).

If you check in at Starbucks, it's not Starbucks that would pay to talk to you, it's McDonald's. Just like Google, if you are searching "Starbucks", it makes no sense to show a Starbucks.com ad.

Don't sell the upsell opportunity short either. If you buy a lot of things at Starbucks, you could probably be more easily convinced to spend more on your next trip.

Again you do not have the luxury of delivering the message in the moment when I am actually interested in an answer.

It might be better than TV adds but it's not that different.

An 'ad' isn't the only activity Starbucks would post. We're in a less traditional paradigm here. Starbucks can do ad-hoc market research, community building, announce products... Lots of things. Some of the other commenters build on this further through competitive analysis and predictive modeling. Over the next few years we'll see this go further as they better understand the cooperative nature of brands and likes.

They are already doing these things.

As long as they can't deliver "suggestions to my question" then they haven't really done anything other than given me a more targeted add.

But targeted is sub-optimale compared to the feedback mechanism that's build into google intent based advertising.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | DMCA | Apply to YC | Contact