DropBox it just works

I was searching for a new opportunity that was much spill client computer software product to a Windows, Mac, or Linux PC or to an phone, pad, Blackberry, or Android mobile device, the software created a local twilight folder for accessing shoot downs of either size or sign via an encrypted Internet liaison from differentwise be sick-enabled devices or from whatsoever web browser. The client software tracked changes in real-time to any appoint in the lend oneselfrs local give the sack folder, then instantly synchronized a copy of the turn on on sweepsys servers, updating only the portions of he saddle that had changed, in order to save bandwidth and time.Likewise, within milliseconds, copies of the send were synchronized in local fox folders on all other devices connected with the affairrs account. We engineered push aside so it just worked, all of the time, Drew explained, We supported all the major operating systems and handled all kinds of obstacles, from f lakey wireless connectednesss to corporate firewalls, which was non an easy t use up. The party adopted a fermium business model, that is, it offered both(prenominal) free and premium accounts.Users got 2 gigabytes of storage for free and had the option to ay $10 per month for 50 gigabytes or $20 per month for 100 gigabytes. Industry observers estimated that 2% to 3% of flipsys users were paying guests, which implied a $10 gazillion to $15 cardinal annual revenue run rate in mid 2010. 1 At that time, the caller had 25 employees, most of whom worked in engineering or support functions. muster out had raised $7. 2 zillion in two rounds of venture capital funding from Sequoia Capital and Cell Partners.Market Overview eject was a new-fashioned entrant to the fiercely competitive online musical accompaniment and storage services space. The first firms in the space, which had splendid companies as clients, ere launched in the modern asses by startups offering outsourced s torage at remote decanters. As prices declined, services as well as became available for consumers seeking to complement their data online. Most early users were technically adept, for example, college students downloading music from peer-to-peer filing cabinet sharing services.Few firms in this first wave of services survived the dot. Com crash, still by late 2006 the market was crowded again with new competitors. In July 2007, the tech blob Amassable published a total of much than 80 online backup and storage services. 2 Market research vendors like DC cut the hype by predicting that the worldwide market for online backup services would grow to $71 5 cardinal by 2011. 3 Investor interest in online storage surged when Muzzy was acquired by EMCEE for $76 cardinal in late 2007.Houston was confident that Drop could succeed in the face of intense competition. He cogitate that Drop would be able to collect revenue from well-nigh users, because consumers generally understood that storage court money, whether it came in the form of a physical snub or an online service. When challenged by endure capita arguings to explain why the world needed another cloud backup company, Houston prayed them, How many of those services do you personally use? The answer from Vs. was almost invariably, None of them. 4 Houston asserted that direct convey with rival services, which often give outed to beam data crosswise firewalls and slightlytimes balked with big files or large numbers of files, was helpful in innovations that contri furthered to these advantages 2 The first generation of cloud storage services was based on a simplistic model, where file accesses were redirected over the Internet instead of to your computers hard rive. Your operating system and all your applications assume that accessing your hard focalize near is cheap and fast, but when these requests are instead routed to a server thousands of miles away, they can take an order of order of ord er of magnitude longer.This subtle but critical distinction explains why when working remotely, even simple actions like browse a directory can freeze your computer for seconds at a time. We needed to take a all in all different approach by storing files locally and updating the cloud copy in the background victimization a number of time- and vindications optimisations. Launching Drop Its hard to imagine Tom Cruise in minority Report sending himself files via Gamma or lugging around a USB thumbprint. ? Drew Houston After his frustrating check on the bus, Houston started working on Drop full time in late 2006. He tell I needed it badly. I worked on multiple desk slip aways and a lap covering and could never remember to note my USB drive with me. I was drowning in e-mail attachments trying to share files for my previous startup. My home desktops power impart literally exploded one day, killing one of my hard drives, and I had no backups. I attempt either(prenominal)thing I co uld find but each product inevitably suffered problems with Internet latency, large files, bugs, or adept made me think too much. To help with the project, Houston recruited Rash Overdose, who dropped out of MIT and by and by became Dropsys co-founder and chief technology officer. The yoke spent the scraggy four months coding a proto type in a tiny Cambridge apartment. With a working prototype in hand, Houston came up with an innovative approach for sieveing demand for a borderline viable product. He had produced various recruiting television receivers for his college fraternity with this know-how he created a three-minute crassest of a product demo and uploaded it to hack News, a popular forum for developers. l did this out of necessity.There was no way I could ask for peoples files sooner we were 100% sure our code was reliable. But I had a prototype that showed off the products scoop features. 7 Houston used the screens to recruit important shewers and to solicit fe edback on features that Drop might include. He added, Not entrance is painful, but not acquire can be fatal. We got a lot of feedback through that movie, so we were learn while we were building. Houston had another reason for posting the video on literary hack News he hoped to ND selective Y Combinatory seed fund and incubator program.He recalled, l had only submitted my application to Y Combinatory and as a gambit to get their attention, I submitted the video to Hacker News. I hoped it would work. 8 It did in April 2007, Drop received $15,000 in funding from Y Combinatory (see Exhibit 1 for excerpts from Dropsys Y Combinatory application). In exchange for a low-toned pctage of a startups common equity-?usually 2% to 10%-?Y Combinatory provided up to $20,000 of seed capital as well as mentoring, workspace, and introductions to other advisors ND investors over a three-month period.Many startups applied to Y Combinations program, which had a track record for matching upstand ing technical police squads with elite venture capital firms. 3 Upon conclusion of the Y Combinatory program in September 2007, Drop raised $1. 2 cardinal of convertible debt from Sequoia Capital. We fit into Sequoias sweet skin senses we were two young technical founders, working out of an apartment, targeting a big market. It helped that we were ranked at the top of our Y Combinatory cohort, Houston recalled.He and Overdose moved to San Francisco to continue building the many, but despite the capital infusion, they proceed to run lean. Drop delivered its service through Amazons SO cloud storage platform, avoiding the need for infrastructure investments and lieu the company to scale rapidly. The co- founders created a private beta program for a limited group of users who registered through a simple landing page. The page contained a short description of Drop and requested an e-mail address from visitors interested in participating in the beta test (Exhibit 2).Houston commented Theres a spectrum of well-informed opinions c recidivate when to launch your product. At one end, Paul Graham tells entrepreneurs, Launch early and often to accelerate learning. At the other end, respected software guru Joel Spooky says, Launch when your product doesnt comp permitely suck. We were managing peoples files, and its a big deal if you pull back or ruin them. That meant moving toward Spooky end of the spectrum and observeing our beta test small. Next, Houston devised ways to spawn demand for the beta service.In a guerilla selling move, he produced another short demo video and posted it in March 2008 on Dig, a site that showcased web content deemed popular by Digs users. Houston felt it was essential to communicate in an authentic manner with the tech enthusiasts who frequented Dig. He sprinkled east wind eggs into the video, for example, references to Chocolate Rain (a Youth phenomenon), TIPS reports used in the movie Office Space, Mitts Gillian Hall, and the 09 IF key for decrypting woolly-headed disks (dissemination of which, in the face of movie studio legal threats, was a hacker crusade).With this tongue-in- cheek nod to its tech-sway audience, the Drop video soared to the top of Dig, few days. Overnight, the list for Dropsys private beta Jumped from 5,000 to 75,000 Ames, far exceeding the groups expectations. Building the corporation Make something people want. -? Y Combinatory motto Based on consumer response to the second video, it appeared that the counter behind Drop-? It comely works-?resonated with potential early adopters, especially those who were familiar with the performance limitations of existing online backup/ storage services. Houston shifted his focus to product development.The Drop team was comprised almost entirely of engineers during the first two eld of the firms existence. Early on, board members tasked Houston with hiring a reduce manager to help coordinate engineering efforts and prioritize features. Houston reflected If you ask ten people what a product manager is, youll get ten different answers. They tend to pin on a continuum with the end points being poet and librarian. A librarian is focused on blocking and tackling, coordination, and facilitating communication. This type of PM is inherently organized and follows up relentlessly.A poet PM listens to the voice of the customer during usability tests and focus groups and based on that insight formulates an aesthetic vision, a grand strategy, and a product roadman. Our first product manager was 4 more of a librarian than a poet, because we needed a librarians discipline even today we dont concord enough of that DNA in the company. But he except drove people nuts. It was painful, but we had to permit him go subsequently six months. For the abutting year, until Drop hired another product manager, the company relied on Houston and Overdose to drive the product roadman.Development proceeded more slowly than Houston had originally expected. In his April 2007 Y Combinatory application, Houston had communicate availability of a version that he could charge for thin 8 weeks, but entry Drop to the public actually took 18 months. Houston give tongue to, As a result of doing a few things well, we left a lot of other things behind. We had no business people, we were terrible at getting mainstream PR, and running fast and tripping didnt make for the most predictable engineering organization. 9 Public Launch Drop opened its beta to the public in September 2008 at Outstretched, an annual competition showcasing high-potential startups. Drop was one of 50 startups selected to present at the event from a pool of over 1,000 applicants. ND withal provide a product development deadline for the team. Houston mused that since Drop was following a move-and-true blueprint for launching a consumer Internet service, his next step would fuddle to be devising a merchandise plan. Drop retained an online marketing adviser to h elp with this task.Houston said, What do most web companies do? Apply to Outstretched, check. Buy Towards, check. Get real marketing people, check. 10 Early on, Drop attempted to acquire new customers through gainful search advertising. However, incumbents had weigh up the cost per click for obvious search keywords. As a result, it cost Drop more than $300 to acquire a paying customer (Exhibit 3). This was not sustainable, since an annual subscription for 50 KGB service was priced at $99. Drop had tweaked its sign-up process to increase the conversation rate from free user to paying customer.The company also experimented with hiding the free service option for visitors who arrived via search ads. Houston recalled, Our average acquisition cost per paying customer went from thousands of dollars to hundreds, but we still had a problem with our economics. And we didnt feel good about doing sneaky things to our users to get them to pay. 11 Sequoia Capital and Cell Partners subsequently led a $6 million Series A round of financing in October 2008, but even with additional capital in the bank, relying on compensable search would not be a viable long- term option.In addition, the team had experimented with display ads and affiliate programs, but these efforts also yielded unacceptably high acquisition costs per paying customer. Houston realized that with a fermium strategy, optimization of marketing messages and pricing would be critical to Dropsys success consistent with this priority, the company hired an analytics engineer as its eighth employee. Inspired by the Backbone growth team dedicated to user acquisition and engagement, Houston posterior assigned 30% of engineering resources to optimizing customer acquisition efforts.This team conterminously tracked metrics across Dropsys conversion funnel by cohort,a for example the percentage of landing page visitors who registered as free users the percent of registrants who still were active free users later on X months and the percent of free users who upgraded to pay subscribers after Y months. Houston said, We run our business based on the Startup Metrics for Pirates framework developed by investor Dave McClure. He says firms should a A cohort was a slew of prospects or users acquired at the same time and/or via the same marketing method. Closely track metrics around the acquisition of landing page visitors energizing of those visitors into users retention of users referral of new visitors by satisfied users and revenue earned from users. The team used A/B testing to fine tune page layouts free storage given to users. Analytics showed that gigabytes were not necessarily the take up measure of value for Drop users. We had all kinds of people paying us for Drop but not even bumping against their quota, Houston said. Analytics likewise revealed that few users were accessing past versions of their files, all of which-?including deleted files-?were being permanently stored by Drop at a s ignificant and rapidly growing cost. The company modified its policy, offering 30 days of give away history free of charge and making unlimited let out history a premium option. Houston said, Just a tenth of a percent improvement in conversion rates, or a small decrease in the cost of serving a customer can have a ample impact on profitability. Premium is a spreadsheet game-?one you win with lots and lots of little moves . 13 Fourteen Months to the EpiphanyDespite improvements through analytics, Houston and his colleagues struggled to make the company marketing programs profitable. Nevertheless, the service grew rapidly, reaching 200,000 users ten days after launch and 1 million users seven months later. The vast majority of these users were acquired through word-of-mouth referrals and viral marketing efforts, rather than gainful advertising. A relentless focus on ease of use and reliability had paid dividends in the form of loyal users who encouraged friends, family, and co-wor kers to try Drop. Houston commented, The power of focus cant be understated.If you look at a feature matrix of Drop versus everyone else, we would never come in first. We would rather do a few things well rather than present Drop in a confvictimization way. 14 To detect ways to improve ease of use, the Drop team tracked support forums closely. Houston said, We get feature requests for things we already have. These are particularly bad because it means that even though weve implemented something, our users cant find it. We pay close attention when that happens. 1 5 The company also maintained a Vote on its site, allowing users to choose and comment on treasures they would like to see added.Since the team gained insight on users preferences through support forums and the Vote, the company did not conduct regular consumer surveys, but it did conduct occasional usability tests. In one instance, the entire team watched as not one of five typical consumers recruited from Scraggliest cou ld successfully install and interact with the application. Houston recalled Watching them exit was excruciating. Imagine if your coffee maker Just spit coffee all over the counter every third time you used it or your car stopped in the middle of the road. Thats the computer welcome for a normal person.The PC is always conspiring against you to lose your stuff or break in some weird way. You have no idea what happened or what you did wrong. Watching those five consumers struggle to try to effigy out how to use our product was probably the most painful day we ever had as a team, but afterward, we created a list of 70 things to fix. B A/B tests divided a set of similar individuals into a control group that screwd a status quo product and a test group that experienced a product with one modified element, to determine if the modification yielded a statistically significantDropBox it just worksI was searching for a new opportunity that was more Drop client software to a Windows, Mac, or Linux PC or to an phone, pad, Blackberry, or Android mobile device, the software created a local Drop folder for accessing files of any size or type via an encrypted Internet connection from other Drop-enabled devices or from any web browser. The client software tracked changes in real-time to any file in the users local Drop folder, then instantly synchronized a copy of the file on Dropsys servers, updating only the portions of he file that had changed, in order to save bandwidth and time.Likewise, within milliseconds, copies of the file were synchronized in local Drop folders on all other devices connected through the users account. We engineered Drop so it just worked, all of the time, Drew explained, We supported all the major operating systems and handled all kinds of obstacles, from outlandish wireless connections to corporate firewalls, which was not an easy task. The company adopted a fermium business model, that is, it offered both free and premium accounts.Users got 2 gigabytes of storage for free and had the option to ay $10 per month for 50 gigabytes or $20 per month for 100 gigabytes. Industry observers estimated that 2% to 3% of Dropsys users were paying customers, which implied a $10 million to $15 million annual revenue run rate in mid 2010. 1 At that time, the company had 25 employees, most of whom worked in engineering or support functions. Drop had raised $7. 2 million in two rounds of venture capital funding from Sequoia Capital and Cell Partners.Market Overview Drop was a late entrant to the fiercely competitive online backup and storage services space. The first firms in the space, which had small companies as customers, ere launched in the late asses by startups offering outsourced storage at remote decanters. As costs declined, services also became available for consumers seeking to backup their data online. Most early users were technically adept, for example, college students downloading music from peer-to-peer file sharing servi ces.Few firms in this first wave of services survived the dot. Com crash, but by late 2006 the market was crowded again with new competitors. In July 2007, the tech blob Amassable published a list of more than 80 online backup and storage services. 2 Market research vendors like DC fuel the hype by predicting that the worldwide market for online backup services would grow to $71 5 million by 2011. 3 Investor interest in online storage surged when Muzzy was acquired by EMCEE for $76 million in late 2007.Houston was confident that Drop could succeed in the face of intense competition. He reasoned that Drop would be able to collect revenue from some users, because consumers generally understood that storage cost money, whether it came in the form of a physical drive or an online service. When challenged by endure capitalists to explain why the world needed another cloud backup company, Houston asked them, How many of those services do you personally use? The answer from Vs. was almos t invariably, None of them. 4 Houston asserted that direct experience with rival services, which often failed to tape transport data across firewalls and sometimes balked with big files or large numbers of files, was helpful in innovations that contributed to these advantages 2 The first generation of cloud storage services was based on a simplistic model, where file accesses were redirected over the Internet instead of to your computers hard rive. Your operating system and all your applications assume that accessing your hard drive is cheap and fast, but when these requests are instead routed to a server thousands of miles away, they can take an order of magnitude longer.This subtle but critical distinction explains why when working remotely, even simple actions like seek a directory can freeze your computer for seconds at a time. We needed to take a whole different approach by storing files locally and updating the cloud copy in the background using a number of time- and vindic ations optimizations. Launching Drop Its hard to imagine Tom Cruise in minority Report sending himself files via Gamma or lugging around a USB thumbprint. ? Drew Houston After his frustrating experience on the bus, Houston started working on Drop full time in late 2006. He said I needed it badly. I worked on multiple desktops and a laptop and could never remember to keep my USB drive with me. I was drowning in email attachments trying to share files for my previous startup. My home desktops power supply literally exploded one day, killing one of my hard drives, and I had no backups. I tried everything I could find but each product inevitably suffered problems with Internet latency, large files, bugs, or Just made me think too much. To help with the project, Houston recruited Rash Overdose, who dropped out of MIT and later became Dropsys co-founder and chief technology officer. The pair spent the next four months coding a prototype in a tiny Cambridge apartment. With a working prot otype in hand, Houston came up with an innovative approach for testing demand for a token(prenominal) viable product. He had produced various recruiting videos for his college fraternity with this know-how he created a three-minute crassest of a product demo and uploaded it to Hacker News, a popular forum for developers. l did this out of necessity.There was no way I could ask for peoples files ahead we were 100% sure our code was reliable. But I had a prototype that showed off the products best features. 7 Houston used the screens to recruit beta testers and to solicit feedback on features that Drop might include. He added, Not launching is painful, but not learning can be fatal. We got a lot of feedback through that video, so we were learning while we were building. Houston had another reason for posting the video on Hacker News he hoped to ND selective Y Combinatory seed fund and incubator program.He recalled, l had Just submitted my application to Y Combinatory and as a gambi t to get their attention, I submitted the video to Hacker News. I hoped it would work. 8 It did in April 2007, Drop received $15,000 in funding from Y Combinatory (see Exhibit 1 for excerpts from Dropsys Y Combinatory application). In exchange for a small percentage of a startups common equity-?usually 2% to 10%-?Y Combinatory provided up to $20,000 of seed capital as well as mentoring, workspace, and introductions to other advisors ND investors over a three-month period.Many startups applied to Y Combinations program, which had a track record for matching fond technical teams with elite venture capital firms. 3 Upon conclusion of the Y Combinatory program in September 2007, Drop raised $1. 2 million of convertible debt from Sequoia Capital. We fit into Sequoias sweet home we were two young technical founders, working out of an apartment, targeting a big market. It helped that we were ranked at the top of our Y Combinatory cohort, Houston recalled.He and Overdose moved to San Fran cisco to continue building the many, but despite the capital infusion, they act to run lean. Drop delivered its service through Amazons SO cloud storage platform, avoiding the need for infrastructure investments and localisation the company to scale rapidly. The co- founders created a private beta program for a limited group of users who registered through a simple landing page. The page contained a short description of Drop and requested an email address from visitors interested in participating in the beta test (Exhibit 2).Houston commented Theres a spectrum of well-informed opinions about when to launch your product. At one end, Paul Graham tells entrepreneurs, Launch early and often to accelerate learning. At the other end, respected software guru Joel Spooky says, Launch when your product doesnt completely suck. We were managing peoples files, and its a big deal if you lose or ruin them. That meant moving toward Spooky end of the spectrum and keeping our beta test small. Nex t, Houston devised ways to retrovert demand for the beta service.In a guerilla marketing move, he produced another short demo video and posted it in March 2008 on Dig, a site that showcased web content deemed popular by Digs users. Houston felt it was essential to communicate in an authentic manner with the tech enthusiasts who frequented Dig. He sprinkled easter eggs into the video, for example, references to Chocolate Rain (a Youth phenomenon), TIPS reports used in the movie Office Space, Mitts Gillian Hall, and the 09 IF key for decrypting wooly disks (dissemination of which, in the face of movie studio legal threats, was a hacker crusade).With this tongue-in- cheek nod to its tech-sway audience, the Drop video soared to the top of Dig, few days. Overnight, the list for Dropsys private beta Jumped from 5,000 to 75,000 Ames, far exceeding the teams expectations. Building the partnership Make something people want. -? Y Combinatory motto Based on consumer response to the second video, it appeared that the portend behind Drop-? It Just works-?resonated with potential early adopters, especially those who were familiar with the performance limitations of existing online backup/ storage services. Houston shifted his focus to product development.The Drop team was comprised almost entirely of engineers during the first two historic period of the firms existence. Early on, board members tasked Houston with hiring a reduce manager to help coordinate engineering efforts and prioritize features. Houston reflected If you ask ten people what a product manager is, youll get ten different answers. They tend to revert on a continuum with the end points being poet and librarian. A librarian is focused on blocking and tackling, coordination, and facilitating communication. This type of PM is inherently organized and follows up relentlessly.A poet PM listens to the voice of the customer during usability tests and focus groups and based on that insight formulates an aes thetic vision, a grand strategy, and a product roadman. Our first product manager was 4 more of a librarian than a poet, because we needed a librarians discipline even today we dont have enough of that DNA in the company. But he Just drove people nuts. It was painful, but we had to let him go after six months. For the next year, until Drop hired another product manager, the company relied on Houston and Overdose to drive the product roadman.Development proceeded more slowly than Houston had originally expected. In his April 2007 Y Combinatory application, Houston had intercommunicate availability of a version that he could charge for thin 8 weeks, but launching Drop to the public actually took 18 months. Houston said, As a result of doing a few things well, we left a lot of other things behind. We had no business people, we were terrible at getting mainstream PR, and running fast and gentle didnt make for the most predictable engineering organization. 9 Public Launch Drop opened i ts beta to the public in September 2008 at Outstretched, an annual competition showcasing high-potential startups. Drop was one of 50 startups selected to present at the event from a pool of over 1,000 applicants. ND also provide a product development deadline for the team. Houston mused that since Drop was following a tried-and-true blueprint for launching a consumer Internet service, his next step would have to be devising a marketing plan. Drop retained an online marketing adviser to help with this task.Houston said, What do most web companies do? Apply to Outstretched, check. Buy Towards, check. Get real marketing people, check. 10 Early on, Drop attempted to acquire new customers through paid search advertising. However, incumbents had attender up the cost per click for obvious search keywords. As a result, it cost Drop more than $300 to acquire a paying customer (Exhibit 3). This was not sustainable, since an annual subscription for 50 KGB service was priced at $99. Drop had tweaked its sign-up process to increase the conversation rate from free user to paying customer.The company also experimented with hiding the free service option for visitors who arrived via search ads. Houston recalled, Our average acquisition cost per paying customer went from thousands of dollars to hundreds, but we still had a problem with our economics. And we didnt feel good about doing sneaky things to our users to get them to pay. 11 Sequoia Capital and Cell Partners subsequently led a $6 million Series A round of financing in October 2008, but even with additional capital in the bank, relying on paid search would not be a viable long- term option.In addition, the team had experimented with display ads and affiliate programs, but these efforts also yielded unacceptably high acquisition costs per paying customer. Houston realized that with a fermium strategy, optimization of marketing messages and pricing would be critical to Dropsys success consistent with this priority, th e company hired an analytics engineer as its eighth employee. Inspired by the Backbone growth team dedicated to user acquisition and engagement, Houston later assigned 30% of engineering resources to optimizing customer acquisition efforts.This team closely tracked metrics across Dropsys conversion funnel by cohort,a for example the percent of landing page visitors who registered as free users the percent of registrants who still were active free users after X months and the percent of free users who upgraded to paid subscribers after Y months. Houston said, We run our business based on the Startup Metrics for Pirates framework developed by investor Dave McClure. He says firms should a A cohort was a set of prospects or users acquired at the same time and/or via the same marketing method. Closely track metrics around the acquisition of landing page visitors activating of those visitors into users retention of users referral of new visitors by satisfied users and revenue earned from users. The team used A/B testing to fine tune page layouts free storage given to users. Analytics showed that gigabytes were not necessarily the best measure of value for Drop users. We had all kinds of people paying us for Drop but not even bumping against their quota, Houston said. Analytics likewise revealed that few users were accessing past versions of their files, all of which-?including deleted files-?were being permanently stored by Drop at a significant and rapidly growing cost. The company modified its policy, offering 30 days of undo history free of charge and making unlimited undo history a premium option. Houston said, Just a tenth of a percent improvement in conversion rates, or a small decrease in the cost of serving a customer can have a large impact on profitability. Premium is a spreadsheet game-?one you win with lots and lots of little moves . 13 Fourteen Months to the EpiphanyDespite improvements through analytics, Houston and his colleagues struggled to make the company marketing programs profitable. Nevertheless, the service grew rapidly, reaching 200,000 users ten days after launch and 1 million users seven months later. The vast majority of these users were acquired through word-of-mouth referrals and viral marketing efforts, rather than paid advertising. A relentless focus on ease of use and reliability had paid dividends in the form of loyal users who encouraged friends, family, and co-workers to try Drop. Houston commented, The power of focus cant be understated.If you look at a feature matrix of Drop versus everyone else, we would never come in first. We would rather do a few things well rather than present Drop in a confusing way. 14 To direct ways to improve ease of use, the Drop team tracked support forums closely. Houston said, We get feature requests for things we already have. These are particularly bad because it means that even though weve implemented something, our users cant find it. We pay close attention when that ha ppens. 1 5 The company also maintained a Vote on its site, allowing users to suffrage and comment on treasures they would like to see added.Since the team gained insight on users preferences through support forums and the Vote, the company did not conduct regular consumer surveys, but it did conduct occasional usability tests. In one instance, the entire team watched as not one of five typical consumers recruited from Scraggliest could successfully install and interact with the application. Houston recalled Watching them fail was excruciating. Imagine if your coffee maker Just spit coffee all over the counter every third time you used it or your car stopped in the middle of the road. Thats the computer experience for a normal person.The PC is always conspiring against you to lose your stuff or break in some weird way. You have no idea what happened or what you did wrong. Watching those five consumers struggle to try to common fig out how to use our product was probably the most pai nful day we ever had as a team, but afterward, we created a list of 70 things to fix. B A/B tests divided a set of similar individuals into a control group that experienced a status quo product and a test group that experienced a product with one modified element, to determine if the modification yielded a statistically significant