Head-to-heads
Each winner next to the company it beat, traced from its first innovation to its moat. Open one to see every step, what the rival did instead, and the sources.
66 head-to-heads
Quark owned one application, page layout, that printed through Adobe's PostScript and placed images and drawings made in Adobe's tools. Adobe owned the page model and the designers' daily tools, so it could sell layout inside a suite designers already bought, and the trained teams working across that suite stayed.
Afterpay offered four free payments on small baskets, earned part of its income from late fees and sold itself to Block in 2022 as its losses rose. Affirm quoted the full cost of each loan with no late fees, underwrote it in real time for purchases from $50 to five-year loans, became Shopify's and Amazon's lender, and turned that volume into the funding base of a standalone business with 23 million active consumers.
HomeAway organized second homes that were already rented by the week and charged their owners a yearly fee to list. Airbnb created new supply from homes people lived in, made it safe to book, and took its fee only when a stay happened.
Barnes & Noble answered each move online: it listed the same titles, matched the discounts and offered free shipping. But it kept its stores' prices and its own stock at the center, and barnesandnoble.com stayed a bookseller's website, while Amazon turned its catalog into a marketplace that outside sellers stocked and its own warehouses delivered.
Buildium put small landlords' books online first. AppFolio built the ledger for managers with hundreds of units, then ran rent and screening through it, so each unit earned more and was harder to move.
Why Buildium lost →Practice Fusion also sold cloud software to small practices, but gave its record away and earned money from the companies that wanted doctors' attention. athenahealth did the billing work itself and was paid on collections, so every claim it handled made its rules better.
Why Practice Fusion lost →Rally sold agile management to enterprises through salespeople and coaches at $35 or more per user a month. Atlassian sold Jira online to any team for a few dollars, let outside developers add the agile boards Rally sold, and let practitioners carry it from team to team.
Intergraph sold turnkey CAD systems whose revenue came from its own computers, sold them directly and wrote its own applications. Autodesk sold drafting software for other makers' PCs through dealers, let outside developers write the specialist applications and trained a workforce in schools, until DWG was the format Intergraph itself had to read.
PayPal Here arrived in 2012 with PayPal's brand and a slightly lower rate, but as a reader and an app. Square had already taken on its sellers' risk, removed the other fees and was putting a register and then loans on the same payment flow; PayPal bought iZettle in 2018 and closed Here in the US in 2023.
Iterable began as an email platform for smaller marketers and added channels one at a time. Braze began inside the app, with an SDK and a stream processor that every later channel reused, and sold to large brands whose spend grew with their users and channels.
Why Iterable lost →Shift rented or outsourced the steps Carvana owned and stayed a small West Coast operation whose costs outran its gross profit. Carvana started on DriveTime's plants, built a national network before demand arrived, kept it through the crisis and then filled it.
Why Shift Technologies lost →Simulations Plus launched GastroPlus first, as an absorption tool it sold and steered on its own, and reached the FDA's drug-interaction reviewers only in 2013. Simcyp had drug makers and regulators build its interaction simulator together, appeared in 80.5% of PBPK-backed FDA novel approvals in 2019–23 against 5.2% for GastroPlus, and anchored a company worth $5.15 billion at the end of 2020, when Simulations Plus peaked at $1.43 billion.
Akamai sold metered delivery to large enterprises and earned more as their traffic grew. Cloudflare gave its network away to the sites Akamai did not pursue, sold the same network's services up-market, and collected free traffic that Akamai's pricing gave it no reason to carry.
eXp built a larger network of agents, 83,060 at the end of 2025, but earned $4.77 billion of revenue that year and posted a net loss. Compass spent investor money to hire the agents with the biggest books of clients, then bought whole brokerages, and ended as the largest U.S. brokerage.
Why AGNT / eXp lost →AWS first pushed its own Kinesis, then hosted standard Kafka on AWS and left partitions and scaling to the customer. Confluent owned the standard's community, ran Kafka for customers on every major cloud and rebuilt its engine for the cloud, and AWS spent five years closing the gap.
Cylance convinced buyers to rip out antivirus with a single prevention product. CrowdStrike was already on their machines with a cloud sensor, so it supplied the replacement and then sold everything after it through the same agent.
Why Cylance lost →New Relic started in the application code and was installed on the servers that ran it. Datadog started on every host, then put tracing and logs on the same agent and tags, so teams replaced several tools with one.
Why New Relic lost →ADP sold time and payroll as separate products feeding a service bureau, and kept that bureau and its float income until its rebuilt platform reached general availability in 2024. Dayforce calculated pay continuously on one employee record, and Ceridian stopped selling its own bureau in 2012 to sell it.
Linode sold Linux administrators a $20 server they found by referral. DigitalOcean sold beginners a $5 one they found through its tutorials, and the tutorial library grew into a search presence Linode never caught.
Why Linode lost →Sermo gave doctors an anonymous room they visited occasionally and sold what they said there. Doximity gave them a verified name and daily tools, so colleagues and then paying customers could find them.
Why Sermo lost →Cerner built its software department by department and later bought breadth, while Epic put care and billing on one clinician-shaped database and never made an acquisition. Epic's flagship customers, certified staff and exchange protocol then made each new hospital cheaper to run on Epic, and Cerner lost Kaiser, then Mayo, then a stream of merging health systems.
InVision owned the review layer on top of other tools. Figma owned the file itself, then let everyone into it free.
Why InVision lost →CancerLinQ pooled feeds from record systems it did not own and could not afford to clean them. Flatiron owned the record, paid clinicians to curate it, and sold the result to every drugmaker, which paid for more curation.
SingleCare arrived in 2015 with one card and prices it negotiated directly with pharmacy chains, and bought its reach with TV ads and cards in doctors' offices. GoodRx showed every network's price at each nearby pharmacy, let PBMs pay for the prescriptions it sent them, and grew on unpaid referrals into a profitable company valued at $12.7 billion at its 2020 IPO.
Foundation Medicine built its testing around tumor tissue and offered blood as a fallback. Guardant made blood the first sample, with a sequencing method accurate enough to trust, built the evidence, coverage and approval that made ordering it routine, sold the resulting data to drug makers, and by 2018 had about three times Foundation's share of liquid biopsy, a lead Foundation now pays patent royalties to compete in.
Ro launched the same service days earlier, raised far more money and was the larger company in 2020. Hims made its subscriptions profitable first, then spent the margin on a louder brand and its own compounding pharmacies, and by 2024 its revenue was about two and a half times Ro's estimate.
Omada built human coaching for diabetes and other chronic conditions and added MSK later. Hinge built MSK care that software could deliver, so its costs grew more slowly than its membership.
Why Omada Health lost →Marketo sold marketing automation that ran on top of Salesforce's CRM and moved upmarket to enterprises. HubSpot sold small businesses a complete suite through its own content and agencies, then gave away its own CRM, so the customer records lived in HubSpot.
Mailchimp was a newsletter tool for every small business, organized around lists of subscribers; Klaviyo was a store's customer database that could send email, so abandoned carts, repeat buyers and attributed revenue came standard. When Shopify asked partners to share merchant data, Klaviyo agreed and Mailchimp left the app store to become an all-in-one suite for all small businesses, and growing online stores consolidated on Klaviyo.
Monster sold employers a database of resumes posted by people already looking for work. LinkedIn gave employed professionals a profile worth keeping, spread it free by invitation, and sold recruiters seats to reach them, so the recruiting budget followed the passive candidates.
Oracle Data Cloud matched customer records in order to sell Oracle's own data and marketing software, so its business depended on platforms that later shut third-party data out. LiveRamp matched records for anyone and sold nothing else, so competitors, publishers and platforms connected to its key.
Rethinkdb spent years perfecting its database and arrived after developers had chosen. MongoDB shipped early with drivers for every language, became the interface developers' code and skills assumed, and then sold that interface as a managed service.
Why RethinkDB lost →Blockbuster earned its money from store rentals and late fees, so each move Netflix made, from the fee-free subscription to streaming inside it, asked Blockbuster to give up income it depended on. It tried each one late, pulled back, and was sold by the time Netflix's programming costs were spread over 20 million streaming members.
Itaú sold cards with fees through branches on a mainframe core, and answered Nubank with side brands, Credicard Zero and iti. Nubank ran a no-fee card from the phone on its own cloud core, let customers recruit the next customers, and sold an account and loans into that base until it served more Brazilians at a lower cost than any bank.
OneLogin had the same product and more apps in 2011, but it sold to smaller companies on a fraction of the money, competed on price, and stalled after its 2017 breach. Okta spent its funding lead on large enterprises and the security record they paid for, then grew inside them until vendors built to its network.
Bright Health sold the same low-priced narrow-network plans, but raced into new states on equity and saw its 2021 claims only after the year had ended. Oscar paid its own claims, retreated where it lost money and shared its risk with reinsurers, so it was solvent when Bright left the exchanges.
Why NeueHealth (Bright Health) lost →PlanGrid put blueprints on the field crew's iPad and sold each subscription separately. Procore gave every company on the job one shared record and let them all in free, so partners became customers and the record grew from bid to payment.
Surveymonkey won individual survey writers with a free tool. Qualtrics trained future corporate buyers at university, followed them into companies, and grew one research engine into company-wide programs that are hard to replace.
Why SurveyMonkey lost →Digg ran one front page that a few power users steered, and in 2010 rebuilt it around publishers to raise revenue. Reddit gave each interest its own community run by its members, kept its costs low enough to stay on users' side, and was ready when Digg's users left.
E*Trade charged about $7 a trade and bought its growth with a $200 million marketing budget. Robinhood let market makers pay for free trades and let customers recruit each other, reached first-time investors E*Trade had never served, and when E*Trade matched the price it gave up about a sixth of its revenue and agreed to sell itself within five months.
Siebel sold client/server CRM that each customer bought as a license and installed, through a large direct sales team built for big companies. Salesforce ran one service for every customer at a monthly price Siebel could not match without undercutting its own licenses, then built customization, outside apps and a trained workforce on that one system.
KeepTruckin gave drivers a free logbook and grew with small carriers under the electronic-log mandate. Samsara sold operators a live, self-installed gateway through its own reps and free trials, then added cameras and other products to the same hardware, so it reached large accounts years earlier.
BMC Remedy was a client/server help desk that each customer installed, customized and upgraded. ServiceNow ran one platform itself, used IT service management to win Remedy's enterprise customers, and then put their other departments on the same data model.
Jobber made software cheap and quick to adopt for the smallest crews in every trade. ServiceTitan went deep in one trade, grew with larger contractors into the system that runs their whole business, and took a share of the money flowing through it.
Why Jobber lost →Bigcommerce followed Shopify into hosting three years later, stayed neutral on apps and payments, and moved upmarket in 2015. Shopify let its partners earn more than it did from the same merchants, then earned on those merchants' sales through payments.
Why BigCommerce lost →Hipchat reached teams four years earlier, capped its free plan at five people, then spent its effort on enterprise editions and a rebuild. Slack let whole teams in free, charged for history, and connected each workspace to its partners and software.
Why HipChat lost →Redshift put the older shared-nothing warehouse on AWS and sold it by the node-hour; Snowflake gave each workload its own compute billed by use, ran on every major cloud and let accounts share live data. By the time Redshift separated storage (2019), shared data (2020) and billed by the second (2022), the workloads had moved.
Earnest priced the same graduates from better data but stayed a refinance lender funded by others. SoFi raised more capital, sold each borrower more products, and bought a bank so members' deposits paid for the loans.
Why Earnest lost →Rdio built the nicer app and charged from the first play, so friends arriving from Facebook met a paywall. Spotify made free listening feel like owning all music, used it to reach far more listeners, and turned their playlists into personal recommendations.
Why Rdio lost →Salesforce bought the leading listening tool in 2011 and the leading Facebook publishing tool in 2012, joined them only in 2014, and sold social as one module of its Marketing Cloud. Sprinklr started with the whole job in one workspace, built it with the largest brands, and rebuilt every later product inside it; Salesforce stopped selling Social Studio in 2022.
Wix gave its builder away and bought a larger user base with years of losses. Squarespace charged everyone, kept the creatives who valued a designed site, and made as much cash as Wix on less than two thirds of its revenue.
Qlik was nearly eight times Tableau's size in 2009 and also landed and expanded, but developers and partners built its applications. Tableau put the query in the analyst's hands, let those analysts spread their work and train in public, and passed Qlik in revenue in 2015.
BetterHelp sold the same messaging therapy and kept buying cash-paying consumers with advertising. Talkspace put a credentialed network inside health plans, so when ad costs rose insurers kept paying for its patients, and it became a profitable company that a hospital operator bought.
Amwell sold health systems and plans the software to run telehealth under their own brands, so each client still had to supply doctors and get members to use it. Teladoc sold the finished service for a fee on every covered member, so its revenue grew with contracts rather than calls, and in 2025 it was about ten times Amwell's.
Why Amwell lost →Foundation Medicine sold a reimbursed DNA test, obtained clinical histories through its partnership with Flatiron, and became part of drugmaker Roche in 2018. Tempus ran its own laboratory so every test produced a matched molecular and clinical record, let drugmakers' data fees pay for a broader test, and sold the same growing library to nearly every large drugmaker; it was worth $11.15 billion at the end of 2025, against the $5.3 billion at which Roche valued Foundation in 2018.
Revel sold an iPad system to every kind of merchant and left card processing to partners until 2017. Toast built only for busy restaurants on hardware it controlled, processed every card, and spent that margin on installs, service and more products in each restaurant.
Lonely Planet sold travelers one paid expert's view of a place, printed and revised every two to four years, and earned only when a traveler bought the book. Tripadvisor gave away millions of current reviews by travelers, had booking sites pay for the readers it sent them, and made its review pages the place travel searches landed.
Nexmo sold wholesale SMS routes to apps that already had volume; Twilio sold every web developer the phone network as code on a free trial, then went out to find those developers. Nexmo matched the API by 2013 but began recruiting developers seven years after Twilio did.
Sidecar opened the pool of everyday drivers first, but spread across cities it could not fill and gave up surge pay. Uber pulled the same levers with a written playbook and billions of dollars, one city at a time, until each city held.
Oracle sold drug makers the Siebel CRM it sold every industry, with life-sciences features added to a general release that each customer installed, customized and upgraded every few years. Veeva ran one validated pharma-only release for all its customers, took the largest drug makers first, and then sold them regulated-content products that Oracle never built beside its CRM.
Solutionreach sent reminders from the practice-management system and left the phone line with a carrier, reaching phones only through Jive in 2018. Weave became the carrier, put the patient record on every call, and sold phones, texting, reminders and later payments as one subscription that offices found hard to give up.
Oracle owned PeopleSoft's customers and earned more from their yearly maintenance fees than from new licenses. PeopleSoft's own founders sold those buyers one version with no upgrade projects, and by the time Oracle's cloud suite caught up, Workday already held their HR and finance records.
MYOB sold an installed file that the owner and the accountant passed back and forth, and it held its online product back to protect that desktop base. Xero put both on one live ledger fed by the bank, and accountants moved their client books onto it, then kept them there.
Freshdesk copied the easy hosted help desk and the free trial four years later and came close to Zendesk's customer count. Zendesk had already turned small first purchases into expanding departmental accounts, so the same kind of customer base earned it several times the revenue.
Why Freshworks lost →Trulia built a better search for people already shopping and sold that traffic first to brokers, then to agents. Zillow gave everyone a reason to visit, the value of their own home, turned it into a brand people typed in, and used the larger audience to outspend and then buy Trulia.
Why Trulia lost →Netskope built the same kind of cloud service, so architecture did not separate them. Zscaler reached the web gateway, the largest companies and private access years earlier, and Netskope has been catching up to a larger base ever since.