gift shop scientist

Content acquisition strategies

Sun, 12 May 2013 19:54:37 -0500

Content acquisition strategies:

1. Aggregation (retrevo) - can be a good starting point but not an attractive destination. Scambait.

2. Skilled Labor (cnet) - easy to get started and usually highest objective quality but doesn’t scale.

3. Unskilled Labor (Turk) - easy to get started and scales some but hard to provide anything meaningful beyond simple classification. Techniques to get more discerning info push cost to unattractive levels.

4. Machine Generation - hard to start, hard to produce meaningful and flexible results. Generally poor human consumable results.

5. Crowdsourced (pinterest, Wikipedia) - impossible to get going, hard to maintain, easy to break. Fantastic subjective quality. Magical if achieved.

6. Vanity (Facebook, dribbble) - hard to get going, subject is generally fixed, hard to monetize due to ownership of content, privacy concerns.

These lines are not fixed. Google successfully combines many of these categories. Most top500 websites use more than one.

Thoughts?

Reviews as a driver of change

Fri, 10 May 2013 03:34:40 -0500

While I’m in Beijing I have been using travel advisors Beijing guide. It’s essentially a downloadable dump of their reviews for Beijing (good idea for lark long term btw).

It’s interesting because if I ask a guide or a hotel or a taxi where the best place to get Peking duck or a massage I get pointed to the restaurant or brothel/masseuse that is paying them.

However just looking in the tripadvisor app I can see that the democratized voting by review system that they have created is surfacing restaurants that are very different than the ones I’m being recommended by locals. As it turns out they are fantastic (and frequently cheaper).

So in this way tripadvisor disrupts the kickback system and nepotism that always evolves in these tourist destinations (just like Las Vegas).

I think there is a similar opportunity in the product review space. It seems to me that in order to make this happen you need a few things:

1) A system of trust that increases the barrier to faking reviews - reputation + voting
2) Strong social/psychological incentive to write reviews - traffic + reputation
3) The ability for users/businesses to add products

I could include things like objective surfacing and review quantity but I think those fall largely without the parameters of 1 & 2.

Tripadvisor has all of these in the travel space. This doesn’t exist in the consumer product apace.

What would democratized product reviews look like? Would products from up and comers in interesting product verticals do better against established brands if their product was better? Would products that typically would not even be found online receive interest because the manufacturer themselves can list the product independent of distribution?

I’m not sure but I really want to find out. I don’t think it can be anything but powerful.

ReviewLark: You're smart, shop like it.

Thu, 25 Apr 2013 15:30:13 -0500

ReviewLark: You're smart, shop like it.

Lark facebook group

Tue, 05 Mar 2013 18:11:13 -0600

Lark facebook group:

Reminder: we post updates to our lark facebook group!

“E-commerce retailers should also take notice of the results in the study, as Amazon won’t be the...

Wed, 27 Feb 2013 12:36:17 -0600

“E-commerce retailers should also take notice of the results in the study, as Amazon won’t be the only beneficiary of showrooming. By understanding where your customers shop offline, you identify opportunities to acquire new customers, and convert more offline spend online”

David Shim, Founder and CEO of Placed

Today we got search working, and there’s a simple register/login mechanism!

Thu, 21 Feb 2013 17:09:48 -0600

Today we got search working, and there’s a simple register/login mechanism!

Lark update video

Thu, 21 Feb 2013 08:42:00 -0600

I’ve put up a quick walkthrough of the functionality we’ve built over the last couple of weeks. I plan on updating this as we make more progress, love to hear your thoughts!

http://www.youtube.com/watch?v=G9-74W3SBzc

http://www.dropzonejs.com/ is awesome. We’re using it on...

Wed, 20 Feb 2013 08:04:13 -0600

http://www.dropzonejs.com/ is awesome. We’re using it on our review form and it looks/feels great!

Lark’s YC Summer 2013 application is in! Wish us luck :)

Tue, 19 Feb 2013 17:14:53 -0600

Lark’s YC Summer 2013 application is in! Wish us luck :)

Help us pick a domain name!

Fri, 08 Feb 2013 11:15:33 -0600

Help us pick a domain name!

Are you a Fit Loyalist?

Wed, 06 Feb 2013 11:10:24 -0600

While doing user research over the past few weeks I’ve noticed a trend I’ve called “Fit Loyalists.” These are shoppers who would normally not be brand loyal but due to the inability to try clothes on while shopping online they become strong brand loyal shoppers.

A great example is a friend of mine who is a total researcher and displays no particular loyalties to either retailers or brands, except in the case of clothing.

Asking her about purchasing a new lamp or a DVD player she tells me she’ll “start at Amazon, then google for reviews before making a decision”. However, asking her about buying a new blouse and all she has to say is “I go straight to H&M.com and if I don’t find anything I like, I go to BananaRepublic.com. If I can’t find anything at those two places, I give up or wait until I can drive to the mall.”

She displays none of the normal Brand Loyal behaviors, except when shopping for apparel at which point she becomes extremely Loyal.

Fit Loyalists.

Epinions recognized this in 1999

Tue, 05 Feb 2013 16:19:24 -0600

Word of mouth is far and away the way people prefer to receive their recommendations and reviews for products. Depending on who you ask you’ll see numbers like 62% higher conversions [1], 10x purchase rate, or this one is good - 433% increase [2]! We may not agree on how much it helps but social recommendations make people buy. That much is obvious.

What I think is REALLY interesting is that Epinions recognized this network effect of reviewers and recommendations in 1999 [3, 4]. Before there was a facebook or a twitter to watch and notice these behaviors - the epinions founders were able to reason out the need for a network of trusted reviewers. That’s fairly admirable. Epinions as an e-bay business is going out of style faster than Crocs, but those founder guys are smart.

References:

Check out this infographic we put together on showrooming

Mon, 04 Feb 2013 16:48:04 -0600

Check out this infographic we put together on showrooming

Great article on how MakeupAlley uses a community to provide trusted reviews

Mon, 04 Feb 2013 06:08:16 -0600

Great article on how MakeupAlley uses a community to provide trusted reviews

Favorite buzzword: Crowdturfing

Thu, 31 Jan 2013 10:23:00 -0600

Crowdturfing (or crowd-turfing) is a combination of crowd sourcing and astroturfing.

Crowd sourcing is a phenomenon popularized over the last half decade or so on the web. It means using an unknown group of free or inexpensive labor to solve a problem.

Astroturfing is a phrase that originated in political circles. It was originally used to describe cultivated grass-root political movements. In other words, fake movements designed to look organic. It has also come to be used as internet parlance for abusing social media or consumer reviews to provide the appearance of an organic popular movement in support of a product or idea. Generally when I say this I’ll be referring to merchants or manufacturers attempting to manipulate reviews to promote their brand in a way that is designed to seem organic.

So then crowdturfing means using unknown groups of cheap labor to artificially promote your brand in a way that seems organic. An example would be posting jobs on a crowd sourcing site such as fiverr.com offering to pay for positive reviews on sites like Amazon.

There is a great read on crowdturfing from UC Santa Barbara @ http://www.cs.ucsb.edu/~gangw/crowdturf-www12.pdf

The problem with online reviews: Reviewer Agendas & Bias

Wed, 30 Jan 2013 14:55:27 -0600

Here’s a shocker: online consumer reviews are polarizing. I’ll pause to give you a moment to recover from your shock.

Really though, they are. This phenomenon is generally referred to as a J-Shaped distribution of reviews and is well documented in academic research [1]. Some studies have found as many as 50% of the reviews posted on sites like Amazon are 5-star reviews [2]. There are many 5-star reviews and many 1-star reviews, and not a lot in between. Online reviews suffer from (at least) two types of biases.

Purchasing bias, also known as Self-selection bias [3] is simply stated as “I buy stuff I like.” Essentially, if I’m a reviewer of items on the internet and I’m choosing the things I’m going to buy; I’m probably buying things I *want*. That should skew my reviews in a positive direction, since I’m only reviewing the items that I have already selected as likely to be enjoyable for myself.

Underreporting bias, or as I like to call it “brag or bitch bias”, is the tendency for reviewers to only write reviews for items that they have a strong emotional reason to review. Thinking about it, this makes sense. I know I’ve personally only made one review on Amazon. It was to trash an induction charger I bought for my mouse (It’s bad, don’t buy it [4]).

It gets worse though. There are also reviewers with hidden agendas. These aren’t the same as the ones that are posting positive or negative reviews for profit as I discussed in my previous blog. These tend to be people who are indirectly incentivized, like: competing brands, jilted consumers trying to cause pain, aggrieved individuals with no claim against the product, and sometimes even expert reviewers with conflicts of interest.An interesting example of an aggrieved group of people attempting to bomb a product is the 2012 biography of Michael Jackson by Randall Sullivan [5]. Hundreds of Michael Jackson fans flocked to Amazon and posted 1-star reviews.. Expert reviewers are not immune either. Greg Sandoval of CNet resigned for ethical reasons when his employer refused to let him consider DISH’s DVR for an award due to an ongoing lawsuit between DISH and CNet’s parent company CBS [6].

Fundamental causes

There are three problems that are working together to produce this biasy-soup (™).

No ongoing reviewer incentive. There is no real reason contribute reviews on a consumer review site or retailer as an end user. The thing that will tend to drive most people in the absence of an ongoing incentive is an extreme emotional response to a product, a hidden bias or some other agenda - to brag…or to bitch.
No diversity incentive. There is no strong incentive to contribute reviews about items that are not particularly interesting to me as a consumer. I bought my backpack on Amazon, it would never really occur to me go to review it because I don’t really care too much about backpacks. It’s a nice backpack though. It probably deserves 4 stars. Without an incentive that brings me back for each purchases, I’ll self-select consumer electronics and computer parts.
No insight into bias. There really are no good tools to help discover biases as a consumer interested in learning about a product. I can try to discern based on the writing style of the review or other circumstantial evidence [7], but lacking CSI type investigatory skills I just guess.

The solution

Lark solves all of these problems, and then some.

Lark’s reputation system creates an ongoing incentive to contribute reviews, even if I’m not particularly excited about the item. My backpack would get some love on Lark because I am investing in my ongoing reputation on Lark. I’m establishing my presence; I view this as something I’m doing for myself.
Biased reviews will be buried, and subsequently biased reviewers will struggle to achieve a strong reputation - causing their future reviews to also be surfaced lower or buried. This will make it harder for biased reviews to be seen, and it creates an incentive to post only believably unbiased reviews. I want to post a review that others can believe is a real review because I want my reputation to climb. It’s the same reason I want to craft thoughtful responses (or questions) on StackOverflow, because it affects my reputation.
The social aspect of the site provides a strong disincentive to be mean. I don’t really want all of my friends to see me in a 1-star flame-war on some book. If it were anonymous I may not care, but since it’s hooked to my Facebook profile I’ll be at least passingly civil. Plus, I don’t want to get unnecessary reputation dings for being rude in my responses.
Lark empowers lurkers to decide for themselves if a review is biased. The reputation number is a badge of honor, or shame, depending on it’s value. This can be used as a primary litmus for the trustworthiness of a review. I can also easily click through and see the review history of a user and make a judgement call - Hey why does this guy always review LG stuff? The social nature of the site will also shine here when other users start calling out the bias in the comments of a review.

Lark is the death of the oh-so-entertaining Amazon review flame wars, and it couldn’t come soon enough. I love building something that I want to use.

Make sure to sign up for our beta @ http://www.lark.io. You can also follow me on twitter for the latest news: @amirpc.

References:

[1]: http://dx.doi.org/10.1145/1562764.1562800

[2]: http://businesscommunication.org/wp-content/uploads/2011/04/06ABC07.pdf

[3]: http://ericchaing.org/files/Li_2008_ISR.pdf

[4]: http://www.amazon.com/review/R397HTWA49FA9B/ref=cm_cr_rdp_perm

[5]: http://consumerist.com/2013/01/21/michael-jackson-fans-attempt-to-crush-book-with-negative-amazon-reviews/

[6]: http://arstechnica.com/gadgets/2013/01/cnet-reporter-resigns-over-cbs-censorship-of-dish-hopper-review/

A survey of review sites on the internet right now.

Tue, 29 Jan 2013 14:09:29 -0600

A survey of review sites on the internet right now.

The problem with online reviews: Fake reviews (pt1)

Mon, 28 Jan 2013 11:50:00 -0600

The Problem

$5. That’s the cost of a 5 star review on major retailers right now [1]. This is a big problem that’s getting bigger. Gartner thinks that by next year 10-15% of all reviews will be fake [7]. Some people think it’s already worse than that. Dr. Bing Liu of UIC estimates that one-third of all consumer reviews on the internet are fake! [2].

You don’t have to go too far to find evidence of companies engaging in this practice. Check out this WSJ expose on a paid reviewer for DeLonghi Espresso Makers. Then there’s the hilarious incident in which a Belkin representative was hiring workers on amazon turk to write reviews on amazon.

Consumers Perspective

Consumers are starting to take notice. One peeved reviewer on Sephora wrote a forum post called “fake reviews” in which she said

I get the feeling that 70% of the reviews on Sephora.com are fake. [4]

At lark we did a little informal survey (< 100 people) in which 31% of respondents claimed to trust reviews not at all or only sometimes. This mirrors the results of the crowdtab survey which also found that 30% of respondants “don’t trust [reviews] much or at all.” [5]

Consumers are working to solve this problem themselves, the same way they do everything: socially. With posts such as this one

“I noticed what appear to be clearly fake reviews from carolineops (submitted 5 reviews for the same product [sp.] all saying that it’s the best thing ever. and no other reviews of other products. really??).” [4]

There are a glut of articles advising users on how to spot fake reviews with tips like this one

“[beware of] Marketing speak. Normal people do not write in marketing speak.” [6]

The outrage has spread to the New York AG office where they settled a lawsuit with a cosmetic surgery company, called Lifestyle Lift, to the tune of $300,000 for faking reviews on popular cosmetic review site RealSelf.com [3].

Retailers Take

Many would ask why aren’t the retailers themselves doing something about this? Well, that’s not entirely fair. Some are. Amazon is trying with their “owner verification” program, in which they provide a badge to people who have purchased the product. Many retailers will delete a review if they receive complaints about it being fake. Amazon, Newegg and others limit the number of reviews that can be posted from a single IP address. The problem is that these retailers don’t make money from policing reviews, they make money by selling product. It’s a hard sell to a retailer living a life of 3% profit margins and a continuous onslaught of a competition to invest their time and money into monitoring user reviews.

The Solution

The demand for a solution has never been hotter, and that’s where lark comes in. We’re solving the trust problem in reviews the same way you are: socially! Our reputation engine, owner verification and real-identity requirements for reviewers are going to transform the reviewsphere (that’s our buzzword, but you can use it).

Make sure to sign up for our beta @ http://www.lark.io. You can also follow me on twitter for the latest news: https://twitter.com/amirpc. If you’re interested in lark as a business you should check out our angel list profile.

Stay tuned for part 2!

References:

[1]: http://www.nytimes.com/2011/08/20/technology/finding-fake-reviews-online.html

[2]: http://www.nytimes.com/2012/08/26/business/book-reviewers-for-hire-meet-a-demand-for-online-raves.html

[3]: http://www.nytimes.com/2009/07/15/technology/internet/15lift.html

[4]: http://community.sephora.com/t5/Sephora-com/Fake-reviews/td-p/135315

[5]: http://www.marketingtechnews.net/media/influence-how_we_are_influenced-infographic_1.jpg

[6]: http://consumerist.com/2010/04/14/how-you-spot-fake-online-reviews/

[7]: http://www.gartner.com/newsroom/id/2161315

I open sourced our custom pacman themed wedding invite

Sun, 27 Jan 2013 17:14:12 -0600

I open sourced our custom pacman themed wedding invite:

Please don’t crash my wedding

Review Traffic Charts

Sat, 26 Jan 2013 07:16:42 -0600