Google (Alphabet) has its own eponymous app for both iOS and Android devices called “Google”. Sometimes it is more colloquially referred to (especially inside the company) as Google Now. The premise is a smart app that trolls your emails, location data, maps history, calendar, and other information that Google knows about you to offer you recommendations and information that Google thinks you would want to know.


Each snippet of information shows up in a little card. There are cards for weather, traffic, public transportation information, news, sports scores, airline reservations, your calendar schedule, stock prices, restaurant reservations, and more. Google wants to predict what you might be thinking and when and offer you specific answers to your internal questions based on a host of information already saved somewhere inside your personal Google account. It also wants to keep you updated on your life events that might be happening via your emails and calendars. Users can click cards for more information, swipe away cards when they are deemed unnecessary or not useful, and scroll to find more cards.

Furthermore, the app features Google voice search command that one sees so often on TV commercials. Open the app, tell your phone your search query, and most of the time it’ll give you a nice little list of results with some snazzy narration.

If one is an avid Google user and funnels his or her email, calendar, internet browsing, video watching, and social life through Google properties, then the app can actually be quite useful. I especially find it useful regarding upcoming flights—the app goes through your email confirmations received from airlines and displays all the relevant information for you in a little, easy-to-digest card format. Another cool use case is for traffic. Presuming Google knows where you work and where you live, it can give you real-time traffic information (taken from standard Google Maps app) and suggest when you should start heading home (via a push notification). Another nice touch is that it includes calls-to-action with some cards, whether it’s to buy tickets to an event, book a hotel, reserve a table at a restaurant, or buy a product. So far, no ads have shown up in the card feed—probably because the usage is so low there are no advertisers willing to spend money to reach minimal eyeballs. Another plausible explanation is that Google doesn’t want to clutter this product with ads as it has with so many of its other products.


The service has been out for quite some time (I have been using it since 2013 and only because I was working at Google) but an informal HBS straw poll of 10 students showed that not one person had even heard about this app offering from Google. I can confirm that sentiment elsewhere in the tech-hub and tech-forward Bay Area, California nonetheless. Few outside of Google actually know Google even has an app, let alone what it does or how it can be useful. Despite massive TV advertising, among the generation of 20-30 somethings this Google app has been far from a hit.

One of the more amazing but downright eerie features of the Google app is the location history. If you have the app and have your location services on (which I presume many users do by default but there is definitely an option to turn it off), Google will track your every movement via your phone and GPS. Look at where I went on December 29, 2014. I can confirm I did play golf that day out in Eastern San Diego and then returned home.


How about Sunday September 28, 2014? I clearly made a full day of Harvard. I wonder what I was doing on Linden Street?


I flew home on Thanksgiving Day in 2014 to surprise my whole family who thought I wasn’t coming home for the holiday. Good thing they have no idea this even exists on Google.


It even tracked me during FIELD II all the way in Brazil on January 4, 2015. Don’t tell mom I went to Bar Tize in Belo Horizonte!


One doesn’t have to think very hard about all the possibilities of how this information could be useful, harmful, interesting, and disastrous all at the same time. I am sure this would take relationships to a new level if only people actually knew about this. For Google and for me, it’s scary—Google basically knows everywhere I go and when I go there with a good degree of accuracy. Should I be upset or should I be thankful in case this information actually comes in handy one day? I think that’s up for an individual to decide. But regardless, Google has done a poor job marketing this feature—I wonder why…

 In summary, if you are a Google maniac and run your life through Google products, Google Now could be very useful for you. I actually really, really like the idea of smart recommendations and trying to predict what you are thinking and/or the information you need. I think it’s the future of the connected world—having technology think for you. And I believe Google is the one company in the world best positioned to do this due to the extensive data they have on all of us. However, without adoption or a solid marketing effort (a problem that is pervasive throughout Google for most of its products and services), this app is nothing but a fun little app for Google employees, ex-employees, and those who love their TV commercials or happen to stumble upon it one day. Perhaps Google doesn’t care to advertise it because they are still working on making the cards relevant and smart and don’t want mass adoption of a less-than-optimal solution.

Finally, the location history is an interesting one. I love showing my friends the technology and most are very intrigued when I do, wondering if they can figure out their own location history. Most are disappointed that they can’t because they don’t have the Google App installed. But then they have no real desire to install it, so they don’t.

Overall, it is definitely a fun little exercise to think about what Google can do with Google Now and what the future might hold.

By: Danny Belch

read more

The Meta-Search Paradox and Value of Online Aggregators

Priceline is an online hotel booking service with inventory that is also directly available from Marriott, Intercontinental, Hilton, and many others. Priceline receives a commission for every room reserved, a commission that is less than the profit that the hotel reaps from the booking. However, in online search ad auctions,  Priceline’s ads are somehow able to appear higher than the ads of any of the individual hotels. Considering that Priceline makes less money per booking, how is it able to pay more per visitor than the hotels?

AdWords is Google’s search ad marketplace and is the most prevalent ad auction online. It uses a formula to determine ad position based on the following inputs: advertiser cost-per-click (CPC), a “quality score,” and the click through rate (CTR) of the ad for a particular search term. The quality score is based on ad relevance (to the search term), landing page experience, and expected click through rate (CTR).

Assume that the auction formula inputs for a Priceline ad and a Hilton ad on a search query for “hotel room in Boston” were the same – same CTR, same ad quality, etc. It seems impossible for Priceline to be able to outbid Hilton in this case, but paradoxically, they they are able to outbid individual hotels on a huge variety of keyword queries.

Suppose a simplified scenario where there are only 10 hotel providers in Boston: Hilton, Marriott, Starwood, and 7 others. Each of the 10 have 10 rooms available for a particular night that a user is interested in. Imagine a user who clicked the Google Ads of each individual hotel before making a decision. Each of the hotels would have to pay for the user to click their ad, even though 9 out of 10 got no benefit. Thus, each one has only a 10% chance of attracting this person’s business, given that he is committed to buying a hotel room in Boston. If a hotel has to pay Google for 10 ad clicks to make one room sale, those ad payments start to add up and eat into the profit margin of the room sale. In contrast, an aggregator like Priceline who has deals with each of the 10 direct players, will have 100 rooms in its inventory. It has a much higher chance of actually gaining a conversion since the consumer is likely to find exactly what they are looking for on Priceline. The aggregator has a higher conversion rate and therefore it can pay more than a direct advertiser to attract a user to its site, despite a lower profit margin for a sale.

In addition to this structural benefit that aggregators benefit from, users often appreciate aggregators because they simplify the buying process. The CTR of an aggregator and is often higher than those of direct players which, coupled with the higher purchase conversion rate, means that aggregators can bid even less and still maintain the top ad positions in search. Aggregators are valuable to users as they minimize the need to shop around across a highly fragmented industry. It is much easier for the user to compare amenities, prices and location in one place. The breadth of listings gives consumers the further confidence that they are making an informed decision.

Furthermore, aggregators can seem like an independent third party and thus foster trust. A user that visits a hotel’s website directly, isn’t likely to have full trust in the reviews that are displayed since they might be biased. A third party that aggregates review information is likely to be more trustable. This isn’t to say that a third party is necessarily trustworthy, but rather that consumers are more likely to trust their reviews and/or recommendations.

You might ask why the hotels would continue to support this structure and fund a channel that undercuts their margins. The answer primarily lies in the fragmented nature of the marketplace. The aggregators would only be threatened if a large majority of hotels in an area colluded to pull their contracts. It is not in an individual hotel’s best interest to turn off this marketing channel and they are unlikely to be able to come to a deal with their direct competitors. The result of this fragmented structure is that the aggregators continue to thrive online.

This model works in a surprising number of fields. Flights, hotels, rental cars, apartments, home services, online courses, and many others.

In addition to providing deep breadth of listings for their users, aggregators can further differentiate themselves by helping users in the decision-making process. This is becoming especially important as online consumers seek the simplicity associated with being served an answer on what to do by a trustworthy expert. Unfortunately, some companies exploit this user desire, by, for example, promoting products that bring in the highest affiliate commission rather than the ones that are truly the best.

There is an opportunity for new companies in many fields to step up and take their role as quality raters and custom recommendation seriously. For example, the company is focused on getting real expert input on specific products and recommending products in all categories on the basis of the combination of expert opinions and public reviews. Similarly, users will value, the well-researched science-based supplement recommender.

As the online economy continues to grow and evolve, the role of online aggregators will also have to adapt if they want to stay competitive. I believe that the aggregators will further differentiate by creating more value by providing complimentary services such as deep research and trustworthy recommendations.

By: Liza Yermakova

read more

I must confess, dear reader – I absolutely love wine, and even my blog post in the Online Economy blog will be about wine. Nonetheless, I’m pretty sure we can find common ground, as I’m willing to bet you enjoy wine too. In fact, I have never met a person who told me they hated wine. Prefer beer or liquor? Sure. Hate wine? Nah.

However, for a person walking into a store to buy a bottle of wine for [insert occasion], there exists a fundamental mismatch between the segmentation that retailers are offering (grape / origin / vintage / nothing at all) and what the customer is seeking (price range / gastronomy / occasion). Why haven’t brick-and-mortar stores figured this out yet is a topic for a whole another blog post, but meanwhile, 23% of shoppers were classified as “Overwhelmed” in Constellation Wines’ Project Genome wine consumer typification study. [1]

The issue is further exacerbated by the choice the customer is facing. Range of SKU’s offered differs wildly depending on the type of outlet, and can be as low as 100-150 at Costco (who, nonetheless, tends to rotate the selection quite often in order to encourage the treasure-hunting effect), to a more typical 1,000-2,000 or more at an upscale supermarket or a specialized liquor store. [2] How many wines of those available in the store have you tried or even heard of? Ten? Twenty?

At best, our choices are guided by the grape type and, for the enthusiasts, production region, while we remain almost fully blind to the specific attributes of the bottle we chose – that is, until we open it. We can use the help of the store assistant (when there is one available), but where’s the guarantee that they (a) know what you like, and (b) are not just trying to sell you their highest-margin bottle?

The role of the Internet in facilitating the wine purchasing process seems to have vast potential here. However, when we start to dig into the details, things are not so simple. For starters, most of existing Internet wine stores (including the market leader, are cluttered and ugly, have a complicated wine choice process and are plagued by phantom inventory that appears to be in stock, but in fact, as you learn at some point after ordering, is not really available. [3]

The largest issue by far, however, is that of delivery timing. While I don’t have hard numbers on hand, out of the $30B worth of wine purchased annually in the U.S., the majority is purchased for same- or next-day consumption, i.e. going to a house party and bringing some wine with you, or dropping by the store after work and grabbing a red to go along with those pork chops your partner is cooking tonight, etc. Those scenarios do not lend themselves to purchasing wine over the Internet to be received in several days’ time. Furthermore, because one normally only buys one or two bottles under those circumstances, the economics do not allow for same-day delivery.

Existing wine recommendation sites, of which Snooth is probably the largest, share the same criticisms of segmentation not matching customer needs, clutter and unavailability. Additionally, they have their own challenges of wine descriptions not being linked to sites or stores where those wines can be bought.

Finally, wine flash sales sites like Lot18 and Gilt Taste Wine – the last group of participants of the online wine sales ecosystem – target the smaller deferred consumption segment and can only substitute the local store if the customer is willing to stock up on multiple wines to cover every potential occasion, which is rarely the case.

Having discussed the limitations imposed by the purchasing behavior, as well as the challenges of existing online and offline solutions, we can formulate the requirements to the “ideal” wine recommendation engine:

  1. Is simple and easy to use, segments wine into same categories as the consumer
  2. Is unbiased and personalized to the taste palate of the individual
  3. Recommends wine that can be purchased at the store near the user NOW

Of the three criteria, partial solutions exist for #1 and #2 in the form of various half-baked mobile apps, subscription clubs, etc., whereas #3 is by far the most difficult one to satisfy. Any start-up aiming to resolve this issue, beyond the obvious need to compile a vast, Pandora-like database of wines and their characteristics, will need to have expertise in hyperlocal commerce in order to provide the seamless transition from online choice to offline purchase.

One company that aims to resolve the discontinuity between online and local shopping is Milo, now owned by eBay. Their product, Fetch, integrates with POS systems such as QuickBooks, RetailPro and Comcash POS used in local stores in order to generate fully-synced in-store inventory that is then made searchable online. Ideally, if Milo was ubiquitous, it would resolve the local requirement for the wine engine.

For some product categories where retail is highly concentrated in only a few national chains, such as electronics, Milo is already a viable solution. Wine retail, however, is still dominated by mom-and-pop shops: top 50 companies combined only form less than 20 percent of the sales in the United States. [4] This means that the scaling of Milo (or similar, alternative solutions), which seems to be going in the logical top-down direction, i.e. signing on the largest national chains (Target, Sears, Best Buy, Barnes & Noble, etc.), is unlikely to reach sufficient penetration of the mom-and-pop shop segment anytime soon.

Meanwhile, alas, any mobile wine recommendation solutions will have to focus on large retailers, most popular wines and deferred consumption.







read more