The wise understand Google’s predicament. Google’s growth is slowing as the customers who matter the most monetarily to Google are now online in some form via a PC, tablet, or smartphone. Monetizing this customer base is how they will grow meaningful revenue. Adding another billion people using their services is not the short term answer to Google’s revenue problems. So what do they do? This was the question underlying the I/O Conference for many of us who study the industry. After watching the keynote and then having the luxury of being at an event with many smart analyst colleagues to kick around ideas, I’m both skeptical and optimistic for Google.
My skepticism has always been tied to their business model. My conviction is free-with-ad-supported business models can only take you so far. I feel Google has reached the limit of that model. Google may sense this also, even though they may still believe the free-with-ads model can take them farther in the consumer space. In an interview with Adam Lashinsky of Fortune Sundar Pichai noted Google can afford to be patient. My interpretation of this statement and other parts of this interview is he says it because Google has a growing business, cloud platform, and enterprise apps segment. This business is not free — using their enterprise apps and cloud platforms are solutions businesses pay for. Similarly, Pichai made this point in his interview:
“First, if we help users get information, a lot of which is inherently commercial, monetization opportunities arise. Second, history always shows that if you build something millions or billions of people end up using, that builds a lot value too.”
The underlying tone of this statement is essentially where the bull case for Google can be made. They are ultimately a machine learning company, arguably the best in the world. Leveraging their machine learning capabilities and monetizing data in many different ways for both commercial and consumer customers is key. Much of the focus on Google has been on their consumer facing stuff like Android. While Android is not a money making platform for Google as an endpoint, it has helped them feed their machine learning engine. However, more emphasis on Google analysis must move toward the commercial side of their big data story.
How this story manifests itself is the tricky balance. You can see the strategic imperatives all over Google’s moves with Android. A good example is their newest feature which gets users out of the app and into the browser when clicking a web page link. Google is essentially trying to un-bundle the web from the app with this tactic. Google is not collecting data so long as people are in non-Google apps. This is why they want to do everything they can to get you out of the app and back to their browser. Their business model incentives are all over this move. Arguably, this is also not a bad move, user experience wise, but it does showcase how their model drives endpoint experiences in things like Android. But there are times when this can and will hurt the user experience as well. Which is where my skepticism in free-but-ad-supported business models lie. They are simply not always aligned with the best interests of the customer in mind.
With a basis of Google more effectively leveraging their machine learning expertise, optimism can be found in their willingness to be more horizontal than they previously were. Bringing their new Photos app to iOS is a good example of this as is bringing their updates of Google Now with Now on Tap to iOS via their Google search app. But these moves may ultimately be a challenge on iOS since tight OS integration is the key to a better experience. The vast majority of consumers will use what is integrated more times than not over what they have to install and set up themselves.
Google Now on Tap was perhaps the most interesting thing I took away from Google I/O. I’m extremely bullish on artificial intelligence engines (I call them “anticipation engines”) and what they bring in terms of a much deeper “assistant” experience to our smart devices. However, I also believe Apple’s and Microsoft’s ability to develop an anticipation engine are being underestimated. A number of use cases Google pointed out for Google Now on Tap are being done on iOS now. For example, iOS will already keep track of where you parked your car and help navigate you back. It also pro-actively shows you your boarding pass the moment you reach the airport. I experienced this for the first time yesterday while traveling home from Austin, Texas — my boarding pass showed up on my lock screen the moment I got to the airport. Now on Tap certainly goes deeper than this, but my point is building a comprehensive anticipation engine will not be exclusive to Google.
But the greater point is Google Now and the new On Tap features are a step closer to the future of smart devices. Underlying artificial intelligence, residing on both the local hardware and the cloud, will help predict, anticipate, and surface value for us before we realize we even need something. Perhaps a good example of this is where Google Photos can go. While no explicit advertising or data collection is specified from Google at the I/O conference, you can imagine in the future Google may notice a picture I took with my car in the background, see my tire is low on air, and recommend I get it fixed or looked at by a professional — even offer local tire shops who have available deals. To many, this will sound or seem creepy, but it is also convenient. There are certainly many security conscious people out there who will hate this but, for the mass market, I’d argue usefulness trumps creepiness in many cases. This may also be one of those things were we need to opt-in to if we want these predictive features or not.
The IoT platform is one I’m very skeptical about. Google does not have the influence over customers the way SoC companies do to drive IoT standards. Apple is truly in the driver’s seat here since they have the customers to drive connected IoT products while the market is still immature. In 3-5 years, we will see where these IoT standards are but it is still early days.
After WWDC, I’ll do a greater ecosystem contrast between what Google and Apple are building as foundations for the future. But the consumer facing stuff from Google I/O was not particularly earth shattering and even the things they are doing with Now are way off from being mass market. But the moves Google is making in deepening their machine learning and monetizing that in commercial segments is a source for optimism.
I can’t find any attempt at trying to quantify the potential size if the online ad market . I’d assume one could take a gander at it, starting from the mature TV ad market (it’s all about brain-time), adding a bonus for better targeting, a malus for being more disruptive, and a correction for marketing spend macroeconomics. Opinions seem to range for “doomed, that market is doomed” to “a thousand years of prosperity will blossom”, with no theoretical background. Logistics curve, anyone ? Oh, that one’s more complex ^^
I don’t understand the statement “Google is not collecting data so long as people are in non-Google apps”. Say I’m using Yelp. Google still knows where I am, that I’m doing stuff on my phone, and that the stuff in question is restaurant-related. It seems to me Google does know pretty much all there is to know (except whether I’m looking at Chinese or Italian, what budget, and who goes there with me… but those blanks could be filled in later, or are being filled in as we speak by the ad code in the Yelp app ?).
I laughed at the car tire example. Next up “Hey lady, you need your roots re-did !”; promotional offer on “How to fix your low self-esteem”; etc. That can’t go wrong :-p
The car example made me laugh. Although my laughing may end when I suddenly find myself targeted (based on my photos) with weightloss pills, dental services and plastic surgery recommendation.
Thinking about photos; The one type of photo that is uniquely valuable to marketeers are photos of newborn babies.
If Google successfully identified those photos then they would be able to target me for the next two decades with offers for baby formula, baby shoes, back-to-school offers, 529 plans, savings accounts, Xmas presents, bday present and college enrolments. I’m sure my happiness with Google products might be wearing a bit thin after the first decade of that barrage.
That would cause a public outrage at least in Japan I would think. We are very sensitive with the information that gets out about our children. Our PTA’s PCs aren’t even allowed to connect to the Internet.
Ben’s point about the limits of the free-with-advertising business model are important. For example,
1) advertising in the US is typically around 1.3% of GDP (ie. it is not a growth industry, so you grow by stealing market share from others), although it may bounce up and down a bit.
2) ad spending in Asia is well below 1% of GDP (eg. China 0.5%) (ie. the ad-supported model will not get you far in Asia for the moment).
These observations suggest that broadening its range of business models may not be an unnecessary luxury for Google.
http://www.bloomberg.com/bw/articles/2014-03-03/advertisings-century-of-flat-line-growth
http://www.statista.com/statistics/304134/ad-spend-share-gdp-asia-countries/
Yes and no.
1- I’m not sure how much of the wwide ad market “online”, then Google, has. There’s probably a lot of room for growth still. And then consolidation ?
2- Maybe new ad channels partially add to older ad channels, instead of just substituting for them
3- especially if the new ad channels are better -offer better ROI, which should make a point for increasing “ad investment”-.
4- by your own figures ad markets besides the US are underdevelopped, ie probably an overall growth opportunity
That’s before lateral expansion (payments, sales,…)
I agree that estimating the potential market size of online ads using the current market as a reference point is probably not the best approach.
From the demand side, there is no shortage of companies that would want to purchase ads if the incremental gross profit earned exceeds the cost of the ads. If online ads can prove that they can do this in all product categories, then the only upper limit is that dictated by the total gross profits of the nation (or some similar calculation).
On the supply side, ad inventory is dictated by the total time one spends in a certain media, which is increasing as devices get more personal and are always with you. Smart watches for example would increase as inventory through notifications.
The real problem as I see it is that traditional ad formats are losing their appeal and are no longer effective enough. Mobile is definitely an issue. Ad blockers are also getting quite popular.
It all boils down to the effectiveness of the ads, and Google for example, hasn’t been as innovative in this area as they have in others. Facebook on the other hand seems to be doing quite well.
The interesting thing about the statistics that I referenced is that they start before mass adoption of radio, TV or internet. In other words, given that ad spending has not changed dramatically over the past 90 years in the US as a percentage of GDP, I would need some very compelling evidence (that I have not seen yet) before I believe it will change much in the next 10 years. So my default is to take the current market as the best indicator for its future size.
Economically, China and the Far East is at least comparable in wealth to the US in the 1920/30s (ie. the beginning of the time series I referenced). The fact that their advertising spending is lower than the US at that stage of development may mean, as you say, that they are underdeveloped and about to take off. Alternatively, the process of connecting buyers&sellers may be different for cultural reasons and may not converge (or converge slowly) to the US way of doing things. A 10-20 year time line for convergence is not much on the scale of history, but from a corporate point of view it would be an eternity.
Do you have and data points on how many people actually use Google Now, and how many thing that it is useful? I agree that machine learning is a good idea in concept, but I have yet to see actual proof that people are consistently getting value out if it.
A lot has been made of Google’s machine learning. The application where machine learning should make the most financial sense, is deciding what display ads to show on network sites based on the content of that site and the preference of the viewer. However, we are seeing many people getting annoyed that they are simply being shown ads for stuff that they just purchased yesterday, etc.
We are also seeing people (like a recent tweet from Jan Dawson) who are saying that Google Now doesn’t give them useful information.
Although machine learning is fantastic in concept, I am not yet convinced that most people find it useful. I would be grateful if you could share some data on this.
From my anecdotal evidence, Google Now needs a lot of input to have enough data to meaningfully chew on, but if it has that, it can really shine.
People happy with gnow are 1- fully Google-ized (phone, tablet, gmail, gcalendar,gsearch, gdocs…) 2- have a lifestyle that mostly fits the ideal case (lots of meetings/outings, a fair bit of travel, a fairly active online life).
People who satisfy these 2 requirement and who are willing to sign up for some AA (artificial assistant, or should that be gaaaah for google-amplified artificial assistant at home ?) seem to be getting a lot of value on their phone and watch, with timely and relevant reminders/alerts/suggestions.
People who don’t check all the right boxes get exponentially (quadratically ?) less value. I’m not very peripatetic nor meeting-y, so though I’m fully google-ized, the most I get is a handful of article recommendations per week, most of which I’ve read already. Plus I’m anal retentive so I’ll set my own reminders and alarms, thank you; and I’m old fashioned so I’ll know what the weather is by looking out the window ^^
A Google Now utilization metric would be nice indeed, though it’s probably very much in flux still, and for a while yet.
Very interesting thank you.
You don’t sound like too many people are getting beneficial information, but then again, maybe we shouldn’t go too deep into anecdotes.
Machine learning is interesting in improving the world, but I would say that there is a mismatch in Google’s case. They only monetise through advertising and the application of machine learning to advertising is quite banal (and often annoying). Of course they could apply machine learning to other things, but for the moment they do not have the business model or experience in making that pay for itself (and seem to have no discipline in doing so).
My view is a bit different. I am skeptical of machine learning in general for the prediction of personal human behaviour or desires. The thing is, statistical methods are great in getting the answer right more often than random. However, they are not necessarily effective in getting it right >90% of the time, which I think would be necessary as a personal assistant.
Traditional ads on the other hand are random. This is the hurdle that internet ads have had to clear. It is a ridiculously low bar to clear.
Ads can have a low hit-to-miss ratio because nobody expects them to hit anyway. On the other hand, personal assistants would quickly become annoying even if their miss ratio was even as low as 10%.
Of course with more contextual info, it should be possible to significantly increase the hit ratio. The problem is, will that be good enough? I’m not sure. Maybe if Google could observe your facial expressions, your blood glucose levels etc., all the time.
Whenever this discussion comes up I am reminded of Brian Hall’s article awhile ago when he discussed for all that companies like Google and Amazon know about a person, their success rate for targeting ads is pretty miserable.
Predictive engines may be the future (personally, I don’t look forward to that, shades of Minority Report have clouded me), but we are a ways away.
Joe
Yes.
Actually, there is a better and non-creepy way to put up ads for tires.
If your car had an accelerometer that could detect whether you had a flat tyre, and if your smartphone had an app that would be notified of the condition, then it would a lot of sense for that app to list nearby tyre shops.
In this case, there is an explicit opt-in to be notified of any malfunctions in you car, and to be given a list of nearby shops. There is a perfect understanding of what information you are giving and to whom. It’s still an ad that understands the context, but you know you aren’t giving it more context than is necessary.
That’s the kind of predictive engine that I want.
That’s exactly the case with Google Now too.
Maybe I should have been a bit more detailed.
The reason why Google Now and in particular the retargeting display ads are so creepy is because they prove that Google is tracking you all over the Internet. The issue is not that they are showing ads. The issue is that they are spying on your actions, and they are doing this without explicit, opt-in consent. In fact, they are ignoring the “do not track” setting on browsers.
If you had an app that connects to the accelerometer in your car, you are explicitly giving the app access to your driving data. This is not the case when Google is analysing your photos.
The opt-in is partially about not getting ads, but much more importantly about not being spied upon.
This distinction between displaying ads and collecting data is very important, but something that Google has intentionally blurred in the past. The following link is about a class action lawsuit against Google where they reluctantly admitted for the first time that even when ads are turned off in Google Apps for Education, they were mining student emails etc. for ad targeting purposes. Parents were very opposed to this (in the 80-90% range).
http://safegov.org/2014/1/31/google-admits-data-mining-student-emails-in-its-free-education-apps
Ignoring the fact that such an app doesn’t exist, to make your analogy work, this “app” needs your car’s accelerometer history to properly work. Otherwise you get yet another generic app that doesn’t work half the time.
Google is actually a good engine because they know you and what you want. Have you considered the fact that you need personalised results to actually build a decent search engine?
I don’t know what you are talking about. All you have to do is not log in and you won’t be tracked. Or at least, they won’t know who you are.
Everyone does that, including open source software. Because it’s stupid. If they don’t, it’s deliberately done so as a PR stunt.
You seem to be saying the following;
1.
Google search needs to track you to give you good results.
2.
Logging in even once on Google includes an explicit opt-in to have your behaviour tracked around the Internet, including non-Google properties.
Are you sure this is what you intended to say?
1. yes, definitely sure about that
2. when you registered for a google account, there’s a privacy statement, read it. and no, there’s no non-google properties involved.
1.
Regarding Google Search, this might be like Google Now. Some people think it’s much better when you give it your personal information, while some people don’t notice anything beneficial. I’m certainly in the latter. I’ve cleared all Google cookies (and I do from time to time) and never noticed a difference in the quality of the search results. Which means that I see no purpose in giving Google access to my personal information to enhance my search results. I am perfectly happy with the non-personalised results.
The Wikipedia article doesn’t seem to think too highly of personalised search either. http://en.wikipedia.org/wiki/Google_Personalized_Search . Although it may be beneficial for some users, it doesn’t seem to benefit everybody.
2.
Whether having a link that leads to a technical and confusing privacy statement constitutes an explicit opt-in is debatable. And yes, Google does track you around the Internet, even on non-Google properties. They can track you on any site that displays their ads. The technology is often referred to as third-party cookies. This is the issue over which Google was successfully sued by the FTC when they were found circumventing the code on Safari’s third-party cookie blocker.
1. I don’t know what you’re talking about. Duckduckgo’s entire existence is based on the premise that you get different results when you are logged in or out. Are you saying they are wasting their time?
As an example of personalised search that I think it’s very useful, I sometimes find myself trying to remember something I read on some blog post that I can’t remember where it was that I visited a few months ago, type the few sentences I remember, and voila! my history related links are always towards the top. Just an example of how personalised results are hugely beneficial.
People may not find it clearly obvious of why Google results are so good, but in part at least, it’s due to personalisation. A feature that many people don’t know it exists so they just assume it’s wouldn’t useful. But they use it all the time without even knowing and acknowledging how good the results are!
You may not notice it. But I do believe you are simply oblivious to it. It doesn’t happen on most queries you know, it’s a bit rarer than that. Which makes all the difference between a good search engine, and a “meh” one.
2. Sorry, my mistake, I did not consider websites that use Google Ads as non Google property. Silly me.
I’m also not familiar with the trial you are referring to (some cookie mishandling or something?).
Yet all you have are anecdotal evidence. And when it comes to proving the opposite, anecdotal evidence suddenly isn’t enough. Don’t be a hypocrite.
Google Now works because it is actively extended and improved by Google. I am sure they have the numbers.
If people really didn’t find it useful, it would have been slowly hidden away and never talked about, like Google+
I’m not sure what your point is.
Is there any harm in asking for a quantitative survey of Google Now usage? Do you think that the fact that Google is not hiding Google Now away, is sufficient to prove that it is useful?
That’s true. So why do you believe it is not found useful on anecdotal evidence alone?
Of course. Nothing is 100% but the fact that it’s continuously work on and improved, while being pre-installed on all Google search apps (Android and iOS) means it’s working to me. Heck, even the main Google search page shows up Google now cards when you search specific keywords.
Question is, why do you think differently?
If you read my comment, I wrote that “I am not yet convinced that most people find it useful”. I did not say that nobody finds it useful. Nor did I assert that most people do not find it useful.
My hypothesis is that the majority of people are not finding it useful, but I do not have substantial evidence to back this up. That is why I am looking for quantitative data.
It seems that you have simply misread my comment.
Ah fair enough, good call.
I just think that because Google Now is actively developed on and pushed so hard is evidence that it is useful to most people to various degrees. I guess that’s not enough for you, which is fair enough.
Perhaps Google needs to extend the Chrome OS to phones and tablets and stop developing Android – leave that to the “Open Source” folks.
Superb post however I was wanting to know if you could write a litte more on this topic? I’d be very grateful if you could elaborate a little bit more.
Today, I went to the beachfront with my kids. I found a sea shell and gave it to my
4 year old daughter and said “You can hear the ocean if you put this to your ear.” She
put the shell to her ear and screamed. There was a
hermit crab inside and it pinched her ear. She never wants to go back!
LoL I know this is entirely off topic but I had to
tell someone!
Great website.ots of useful information here. look forward to the continuation.stream cnn news live free
Great post Thank you. look forward to the continuation.Live TV
Great blog here! Also your site loads up very fast!
What host are you using? Can I get your affiliate link to your host?
I wish my site loaded up as fast as yours lol
Some really excellent info Sword lily I detected this.LAPGEAR Home Office Lap Desk with Device Ledge Mouse Pad and Phone Holder – Silver Carbon – Fits up to 15.6 Inch Laptops – Style No. 91585 – Hot Deals
You have noted very interesting details! ps decent web site. – hey dude mens shoes
As I website owner I believe the content material here is really good appreciate it for your efforts. <a href="https://www.larazondechivilcoy.com.ar/a/click2.asp?id=1354
But wanna say that this really is quite helpful Thanks for taking your time to write this.