Traction

The bottom line

Startups fail for lack of customers, not products, so find traction by testing several channels cheaply and then focusing everything on the one that works. Builders who start this alongside the product learn early where it leaks and grow fast once it is ready.

Key takeaways

• Getting customers deserves half your time from the first day, alongside building the product. Most failed startups had a product but too few customers. Ch. 3 · Traction Thinking
  • Dropbox tested search ads while still building. Each customer cost $230 for a $99 product, so it built a referral program instead, which became its biggest growth driver.
  • Marketo ran SEO and a blog before product development began. It reached market with a pipeline of more than fourteen thousand interested buyers.
• Set one traction goal in hard numbers and drop any work that cannot move it. Winning a channel is worthless if the channel is too small to matter. PREFACE: Traction Trumps Everything
  • DuckDuckGo ranked first for "new search engine," but the phrase drew about fifty searches a day. It needed about five thousand new visitors a day.
• Consider all nineteen channels, especially the ones competitors ignore. The best channel is hard to predict, and crowded channels are already contested. Ch. 2 · Traction Channels
  • SureStop, a bike brake startup, sealed one partnership at a single trade show. It led to more than twenty thousand sales.
  • In under a year, myZamana went from 100,000 users to over 4 million by relying on viral marketing.
• Test about three channels cheaply in parallel, then pour all effort into the winner. One channel usually dominates growth at any stage, and weaker channels drain focus. Ch. 4 · Bullseye
  • Mint aimed for 100,000 customers in six months and tested blogs, publicity and search ads. Blogs won and brought the first forty thousand. When blogs maxed out, Mint reran the process, switched to publicity and reached 1 million users within six months of launch.
• Keep testing inside your core channel, because every strategy decays. Competitors crowd in and users tune out, so traction has to be found again at each stage. Ch. 5 · Traction Testing
  • The first banner ads drew click-through rates over 75 percent. Once banners became common, those rates plummeted.

Nearly all startups that fail have built a product. What they lack is a big enough customer base.

Gabriel Weinberg, Traction Thinking

1Startups run out of customers, not product

Startups die from too few customers, and working on traction from day one exposes the leaks and traps while there is still time to fix them.

• Traction is the one result that shrinks every other risk. PREFACE: Traction Trumps Everything
  • Traction means sales for a paid product, or a growing user base for a free one.
  • With traction, technical, market and team risks fall. Fundraising, hiring, press, partnerships and acquisitions all get easier.
  • The bar for first traction depends on the market. In enterprise software a few paying customers can count. In consumer software it may take hundreds of thousands of users.
  • In 2006 Gabriel Weinberg sold an Internet company he had cofounded for millions of dollars. It never had employees. He wrote the code and ran accounting and support himself, and it grew first through SEO and later through viral marketing.
• The usual startup path ends with the money gone before customers arrive. PREFACE: Traction Trumps Everything
  • Teams build, launch and expect customers to show up. Then they try a few familiar tactics, such as Facebook ads, local PR and blog posts.
  • Many of these dead products were onto something. A better traction strategy might have saved them.
  • Marc Andreessen says the main reason his firm passes on founders it otherwise likes is a focus on product to the exclusion of everything else. Some even claim they need no distribution plan, or call having none a viral strategy.
  • Paul Graham's view is that customers do not flock to a better mousetrap: Startups take off because the founders make them take off. The most common unscalable early job is recruiting users by hand.
• Spending half your time on traction from day one makes an imperfect product better faster. Ch. 3 · Traction Thinking
  • Early spend leaks out of an unfinished product like water from a leaky bucket. The leaks show where the holes are, so the spend is not wasted.
  • Cold customers have no tie to the founders, so they give more honest feedback than friends.
  • Splitting time slows building in the short run. It speeds a successful launch because it avoids extra product cycles after launch.
  • In fall 2008 Weinberg posted DuckDuckGo to Hacker News as "What do you think of my new search engine?" It was not good enough to switch to. The replies still showed real interest in a Google alternative and pointed him toward search privacy, years before that became a mainstream concern.
  • Phil Fernandez's Marketo, which IPO'd in 2013, blogged about the problems it meant to solve before building. Reader feedback then shaped the product itself.
  • Cold-customer contact also shows which messages land, which niche to start with, which customers are easiest to win, and what distribution roadblocks lie ahead.
• Early traction work reveals the traps a good product cannot escape. Ch. 3 · Traction Thinking
  • Four traps can kill a wanted product: no business model, a market too small, customers too costly to reach, or a hypercompetitive market.
  • Products pass through three phases: make something people want, market it, then scale the business.
  • In phase I, send only a small flow of customers through, using unscalable tactics like talks, guest posts, emailing contacts and conferences. Check whether the product leaks less over time.
  • Scale only in phase II, once product-market fit means customers stay. Phase III needs big reach: adding 100,000 customers at 1 to 5 percent conversion means reaching 2 to 10 million targeted people.
• Investors want ever more traction, so look for engaged bright spots before pivoting. Ch. 3 · Traction Thinking
  • The traction investors expect keeps rising as the ecosystem gets more competitive.
  • Investors who know your space can extrapolate from small numbers. Sustained engagement growth is hard to ignore even from a small base.
  • A few deeply engaged customers may hold promise that the core metrics hide.
  • Naval Ravikant traced the swings. In November 2010 a daily deal startup could raise before any traction, though eighteen months earlier none could raise at all. A mobile app once needed ten thousand downloads. Later it likely needed a few hundred thousand plus fast adoption. Recent AngelList fundings show the current bar.
  • DuckDuckGo has made privacy its core difference since 2009. Growth was steady until the 2013 NSA leaks made privacy mainstream, then it exploded. Weinberg has stayed with it for more than seven years, while other search startups quit after two.
  • Being a year or two early can be a head start. Being a decade early is not, and early adopters should already be using the product eagerly. Ask whether your bright spots are early adopters in a huge market or outliers. With no bright spots at all, it may be time to pivot.

2Pick one traction goal that would change the company, and cut everything off its path

A startup chooses its work well only when every task ties to one numeric traction goal and a short list of milestones that are strictly needed to reach it.

• The right goal is one whose achievement would change the company's outcome, judged against your own situation. Ch. 6 · Critical Path
  • Sample goals: one thousand paying customers, one hundred new customers a day, or 10 percent of a market.
  • Hitting it should make the company profitable, easier to fund, or the market leader.
  • Two questions help set it. Are you optimizing for growth or profit? What traction must you show to raise money within a set number of months?
  • DuckDuckGo set its goals in sequence. First came real switching, meaning people used it as their main search engine. That marked product/market fit and the move from phase I to phase II.
  • Its next goal was 100 million searches a month, which brought it to about break-even. The current goal is 1 percent of general search, which would make it an entrenched player with better deals and publicity.
  • A 1 percent target is usually weak. It works for DuckDuckGo only because search is a huge market with very few companies. A goal that matters to one company can mean nothing to another.
• Break the goal into dated, numeric subgoals and put them on the product calendar. Ch. 6 · Critical Path
  • Work backward to targets like one thousand customers by next quarter or 20 percent monthly growth.
  • Dates and numbers create accountability.
  • Sharing one calendar with product work makes sure traction gets at least half the time.
  • The goal should state both a customer count and a growth rate, usually the level needed to raise funding or turn profitable.
  • What counts as moving the needle changes with scale. Adding 100,000 customers at 1 to 5 percent conversion means reaching 2 to 10 million targeted people.
• Before investing months in a tactic, check whether it could reach the goal even if it worked perfectly. PREFACE: Traction Trumps Everything
  • A tactic that brings real users can still be a blip against the target.
  • Standard link tactics like press coverage can saturate fast and cap a channel early.
  • DuckDuckGo skipped "search engine" because rivals had a decade of history and tens of thousands of links. It aimed at "new search engine" instead.
  • To get links, Weinberg built a karma widget that showed a person's social profiles and follower counts. A "new search engine" link back to DuckDuckGo sat at the bottom. It reached number one, then traction leveled off.
  • He survived the detour only because he could fund himself. Most founders cannot.
• The Critical Path works mainly as a filter for what not to do. Ch. 6 · Critical Path
  • It is the fewest-step route to the goal, written as a list of milestones.
  • Hires or features belong on it only if they are absolutely necessary.
  • Off-path work wastes scarce resources, and competitors usually have far more of them.
  • The rule is blunt: If it is not on the path, don't do it! Run this check as part of regular management.
  • For 100 million searches a month, DuckDuckGo's path was a faster site, a stronger mobile offering, and more broadcast TV coverage through publicity.
  • Users kept asking for images and auto-suggest. The team left both off, because loyal users liked other features enough to forgive the gaps.
• Reassess the path after every milestone, and rebuild it whenever the goal changes. Ch. 6 · Critical Path
  • The first milestone list is often wrong.
  • Use market feedback to skip or add steps.
  • Challenge every item on the list and cut anything that is not strictly required.
  • For the 1 percent goal, images and auto-suggest became required milestones. Winning mainstream users meant serving people who were much less forgiving.

3Brainstorm all nineteen channels, test about three, then commit to one

Bullseye counters distribution failure in three rings: list a plan for every channel, test the few most promising for data, then commit all effort to the winner.

• The outer ring needs one concrete channel strategy for each of the nineteen channels, written before any is ruled out. Ch. 4 · Bullseye
  • A channel is a broad category, such as social ads. A strategy is a specific move inside it, such as running ads on reddit.
  • Each idea should come from research on what worked and failed for companies in your space.
  • The stakes follow Peter Thiel's view. One channel is likely optimal, and most businesses never get any channel working.
  • For each channel, picture what success would look like. That makes a vague category into a plan you can rank.
  • Keep every brainstormed idea in a spreadsheet with educated guesses. Ideas that lose this round become candidates for the next one.
  • Talk to founders a few steps ahead, especially those who failed at what you are attempting.
• Founder bias hides good channels, so the brainstorm has to check for it on purpose. Ch. 6 · Critical Path
  • Some channels are out of sight, such as speaking engagements.
  • Some are disliked, such as sales or affiliate marketing. Others get skipped as schlep, such as business development and trade shows.
  • Every one of the nineteen has given initial traction to both enterprise and consumer startups.
  • Your distaste is not your customers' distaste: Just because you hate talking on the phone doesn't mean your customers do.
  • Jason Cohen of WP Engine and Smart Bear bets that many competitors will refuse to try certain channels. He sees that as a reason to try them, rather than fight rivals for the top AdWords spots.
  • No one is expert in every channel. Note which ones you know least about, and ask mentors to help brainstorm and rank.
• The middle ring runs small parallel tests that collect data, not growth. Ch. 4 · Bullseye
  • Promote ideas until excitement clearly drops, often around the third channel.
  • Each test should cost at most $1,000 and take at most a month. It should roughly answer three questions: what a customer costs, how many are available, and whether they are the right customers now.
  • Parallel tests save time because tests take a while to run. Too many at once cause errors from lost focus.
  • Premature scaling is the common mistake. Testing is not scaling.
  • Mint's spreadsheet estimated traffic, click-through rate and conversion rate for each strategy, then computed expected customers. Its cheap tests included sponsoring a small newsletter, contacting Suze Orman and placing some Google ads.
  • To test blogs, Mint got a few blogs, each representing a different customer segment, to write about it. That showed which audience responded.
  • A $250 AdWords test gives a rough first read on whether search ads deserve more attention.
• Tests should show the best case, so a channel that fails one has little hope. APPENDIX: MIDDLE RING TESTS
  • Use your most confident keywords and tightest audiences, even if they cost more.
  • Set up conversion tracking before spending, since clicks don't show customers. If tracking can't be automated, ask new customers how they found you.
  • Start local and niche. Those audiences are easier to win and cheaper.
  • The search ads starter test runs four ads in Bing Ads, which is often cheaper than Google AdWords.
  • The social ads test targets two very specific niches. On Twitter it targets handles of industry leaders or competitors, and it tries several images, because the image strongly affects results.
  • The sales test lists twenty local prospects and meets them in person through warm intros. A viral loop test treats a coefficient near 0.5 as promising.
  • These are starting points, not fixed recipes. Your own Bullseye work may produce better tests.
• The inner ring puts all effort into the one core channel and cuts channels that only somewhat work. Ch. 4 · Bullseye
  • A hypothetical startup tests search ads, trade shows and publicity. Search ads win clearly, and the other two work, but much less.
  • The right move is to spend everything on search ads. New tactics inside the winner will matter more than the weaker channels.
  • Bullseye is to traction what Lean is to product.
  • Other channels may feed the core one, as when publicity earns links for SEO. Running several separate strategies at once is the mistake.
  • Lean addresses product risk, whether customers want what you build. Bullseye addresses market risk, whether you can reach them sustainably.
  • Both should run in parallel, because the belief that a good product draws customers on its own is false.

4Earn attention where your customers already listen

Blogs, press, stunts, events and talks win a first wave of customers cheaply by reaching people where they already pay attention, even though most of these tactics stop scaling later.

• Small blogs and link-sharing sites are the fastest route to first customers, even though they run out. Ch. 7 · Targeting Blogs
  • Mint's prelaunch "I Want Mint" badge appeared on 600 blogs and brought fifty thousand signups.
  • Mint's emails titled "Can I send you $500?" bought small tracked ads on personal blogs. They drew more than ten thousand preregistrations.
  • Dropbox posted a video on Hacker News and got more than ten thousand signups.
  • The badge traded access for promotion. Waitlist users embedded it with copy-and-paste code. Those who drove signups got VIP access before other invitations went out. The hundreds of new links also lifted Mint.com in search.
  • Personal bloggers often have loyal readers but earn nothing from writing. Payment plus a useful product to show off made most of them willing partners.
  • AppSumo did not pitch Lifehacker on the startup. It first built a bundle of productivity webapps aimed at Lifehacker's readers, then reached out. Lifehacker ran it, and AppSumo later became profitable with more than 800,000 customers.
  • Filepicker.io posted a basic demo on Hacker News and held the top spot for nearly three hours. The result was 10,000+ visits, 460 concurrent users and 500+ developer signups. Results depend on fitting the community. The surest way to find niche blogs is to ask target customers what they read.
• Press filters up from small sites to big outlets, so pitch the small ones with real news and a divisive angle. Ch. 8 · Publicity
  • Tech stories typically climb from Hacker News and subreddits to TechCrunch and Lifehacker, then to The New York Times.
  • Pitch only milestones such as funding, launches or big partnerships. Bundle small news into one announcement.
  • Keep pitches short, with contact details and demo links. Choose an angle people will argue about, because Satisfaction is a nonviral emotion.
  • DonorsChoose.org spread among New York City teachers, then local press, then Newsweek, with modest effect. An Oprah staffer saw the Newsweek piece. Oprah named it a favorite thing for 2010, and Gates Foundation sponsorship and a big rise in donations followed.
  • Jason Baptiste's PadPressed email to TechCrunch offered an exclusive before a noon EST launch. It listed a few features, linked a video demo and a live demo site, and offered giveaways to readers.
  • Ryan Holiday pitched a book on high-frequency trading as proof that the market is rigged. It drew far more press than a typical stock market book.
  • Gabriel talked with a Time reporter on Twitter long before any pitch. That reporter later put DuckDuckGo on Time's top fifty sites of 2011. Many print reporters ignore PR firm pitches but listen to founders, so test the channel yourself first.
• Stunts and customer appreciation are an uncrowded channel with big but unpredictable payoffs, so plan for flops. Ch. 9 · Unconventional PR
  • Half.com got Halfway, Oregon renamed Half.com for a year. It cost $70,000 and two hires.
  • Dollar Shave Club's $5,000 launch video was the main source of over 12,000 customers in two days.
  • Smaller gestures, like Hipmunk's luggage tags and Zappos's support, build word of mouth even without press.
  • Half.com brainstormed for weeks and launched on the Today show with the town's mayor in late 1999. The stunt drew over 40 million impressions. eBay bought the company for more than $300 million six months later.
  • WePay put a six-hundred-pound block of ice at PayPal's developer conference while PayPal faced criticism for freezing accounts. Thousands of people signed up.
  • Grasshopper spent $65,000 sending chocolate-covered grasshoppers and a video link to five thousand influencers. Visits from Twitter and Facebook rose over 3,000 percent. Its March Madness promotions, ticket giveaways and dancing grasshopper videos all flopped.
  • Chargify skipped a $10,000 to $15,000 SXSW sponsorship. It paid $3,000 for a stuntman in its green bull costume and left with hundreds of customers.
• Events reach customers who ignore online ads or sit behind long sales cycles, and few startups compete there. Ch. 23 · Offline Events
  • Twitter spent $11K on hallway screens at SXSW 2007. Daily tweets rose from twenty thousand to more than sixty thousand.
  • Enservio could not reach top insurance executives any other way, so it ran its own summit.
  • An unproven event sells through relationships, not ads.
  • At SXSW, attendees texted "join sxsw" to 40404 to appear on the screens and auto-follow about six ambassador users. Evan Williams called it about the only marketing money Twitter ever spent.
  • Enservio held a multiday summit at the Ritz-Carlton in Dove Mountain, Arizona. Speakers came from consulting firms and hot startups, and nothing felt like a sales pitch. It became an annual event.
  • MicroConf's first year was hard to sell. Facebook ads and AdWords failed, and a pay-with-a-tweet e-book spread but sold no tickets. Once year one proved its value, price stopped mattering.
  • Eric Ries asked his blog readers before selling tickets, kept it to one day in San Francisco, and live-streamed to meetups. More than twenty cities still hold Lean Startup Circle meetups. Events fit poorly when customers are scattered, as HitTail's are.
• Speaking slots go to founders who ask organizers what they want and reuse one or two story talks. Ch. 24 · Speaking Engagements
  • Steve Barsh asks organizers which topics they want covered, then writes a pitch to match.
  • Dan Martell keeps two core talks, one for entrepreneurs and one for partners.
  • Free talks at coworking spaces and small events build the track record bigger stages require.
  • Proposal lead times differ by tier. Premier events need six to twelve months, regional two to four, local one to three.
  • Martell opens by noting he sold Flowtown and Spheric for millions. He never marketed himself as a speaker. Invitations came from people who had seen him talk.
  • Build each slide as a roughly seven-minute story with a beginning, middle and end. A sixty-minute talk can then shrink to twenty by dropping slides.
  • A great talk to 250 people reaches only 250 unless it is recorded. Rand Fishkin tweets his slides beforehand, so his later video lands with buzz already built. Get the attendee list and book meetings before you speak.

5Buy attention in small tracked bets and scale only what converts

Paid channels can be bought at almost any scale, so a startup can test them cheaply and keep only the ones whose cost per customer works.

• Search ads test positioning before you build, and near breakeven in a few weeks signals a possible core channel. Ch. 10 · Search Engine Marketing (SEM)
  • Search ads reach people already looking for a solution, so they suit companies selling directly to their customer.
  • Even $500 to $5,000 yields baseline conversion rates, cost per customer and early hints of lifetime value.
  • Judge every ad, keyword and page by cost per acquisition, not clicks. CPA equals cost per click divided by conversion rate, so $1 at 10 percent is $10.
  • Archives.com, a genealogy site run by Matthew Monahan of Inflection, pointed AdWords ads at landing pages that each tested one concept, such as census data versus family trees, or celebrity links. The first campaign broke even within weeks, with no optimized pages. The company later spent over $100,000 a month on search, and Ancestry.com bought it for $100 million.
  • Monahan's math: keywords cost 15 cents. At 13 cents of revenue per click, scaling only loses money. At 20 cents, margins reach about 25 percent, so a roughly 50 percent gain decides whether you can scale.
  • Average click-through is about 2 percent. Below 1.5 percent, rewrite the ad, test a new audience or drop the keyword, because low CTR lowers quality score and raises price.
  • Retargeting, the content network and Conversion Optimizer help only once campaigns are profitable. Before scaling, they are premature.
• Social ads work by indirect response: build an audience around content first, convert it later. Ch. 11 · Social and Display Ads
  • People visit social sites for entertainment, so send ads to your story or mission, not a sales page.
  • Spend only behind content that is already being shared. Paid promotion cannot rescue a message nobody wants.
  • On social platforms, success depends on the advertiser, not the platform.
  • CareOne, a debt relief company, compared lead sources in 2011. Social leads filled out the consultation form at a 179% higher rate and were 217% more likely to make a first payment.
  • Airbrake put $15 of Twitter ads behind an interview with Stripe's CTO. It sparked hundreds of retweets, submissions to reddit and Hacker News, and tens of thousands of visits.
  • Warby Parker mails glasses for free home try-on and asks customers to post photos and ask friends for feedback. The product experience itself creates social presence.
• Offline ads can't be predicted, so track every one and scale only the winners. Ch. 12 · Offline Ads
  • Every medium sells at local scale, from a $300 radio spot to a cheap billboard.
  • Check each outlet's ad kit for its audience, and ask customers what they actually notice.
  • Give each ad its own URL, promo code or offer, and ask sign-ups how they heard about you.
  • Smart Bear Software offered a free book on code review in its ads. A spike in book requests after a Dr. Dobb's Journal ad proved it worked. Its founder Jason Cohen admitted I have zero ability to predict what's going to work. Cheap magazines sometimes paid off hugely, and winners decayed after a quarter or a year.
  • A $7,000 one-month DuckDuckGo billboard in San Francisco's SoMa district contrasted its privacy with Google's. Wired and USA Today covered it, and users doubled that month.
  • Infomercials drove P90X to $400 million in sales. Snuggie needed two-minute spots to make its case and sold millions. InstaCab had cyclists hand business cards to people hailing taxis.
• Affiliate programs move the risk of bad traffic onto the affiliate, because you pay only for real sales or leads. Ch. 20 · Affiliate Programs
  • You define the conversion, so declined cards and invalid leads earn nothing.
  • Start on an existing network with a flat fee or flat percentage set below what a customer is worth.
  • Recruit your own customers first, then bloggers and other content creators.
  • Network setup costs vary: Commission Junction over $2,000, Pepperjam $1,000, ShareASale about $500, ClickBank $50 to list.
  • Blogger Maneesh Sethi loved RescueTime, joined its program and referred more than three thousand people. That reached a new audience with no wasted spend on leads that never converted.
  • Amazon pays 4 to 8.5 percent depending on monthly volume. Tiered payouts like this should wait until a startup knows its economics. RetailMeNot ranks for searches like "Zappos discount" and earns a cut when visitors buy with its codes.

6Build assets that keep bringing customers after you stop paying

Content, free tools, triggered email, viral loops and communities take effort once and then keep producing customers at near-zero marginal cost.

• Choose a fat-head or long-tail SEO strategy based on search volume and your real odds of reaching page one. Ch. 13 · Search Engine Optimization (SEO)
  • Short, popular fat-head terms make up about 30 percent of searches. The specific long tail makes up about 70 percent.
  • Only about 10 percent of clicks go past the first ten results, so a low ranking is worth little.
  • Links are the hard part because they depend on outsiders. Bought links work briefly, then get penalized.
  • A new category may have no search demand at all. Rand Fishkin's Uber example: Nobody searches for "alternatives to taxicabs that I can hire via my phone."
  • Patrick McKenzie paid $3.50 for an owls of East Asia bingo page. It earned about $60 over three years. Freelance articles cost him $10 to $20 per term, while a SaaS customer can be worth hundreds or thousands.
  • Names Database generated a page for each person from its own data. People hunting for old classmates landed on those pages, which brought heavy long-tail traffic.
  • HubSpot drew 45 to 50k leads a month, and 60 to 80 percent came from unpaid inbound. Those leads cost 50 percent less and closed 100 percent more than paid leads.
• A blog becomes a top customer source only after six months of hand promotion and posts readers can't find elsewhere. Ch. 14 · Content Marketing
  • A post keeps getting indexed and shared for years. Ad spend is gone once used.
  • Early traffic stays low regardless of quality. Ask influencers for feedback, answer Quora questions and guest post.
  • Posts spread best when they show readers a problem they didn't know they had.
  • Unbounce started blogging the day it began building, and its first hire was a blogger. At six months the blog had fewer than 800 monthly visits. At the summer 2010 launch it had 20,000 monthly visitors and an email list over 5,000.
  • Unbounce's infographics were shared roughly twenty times as often as its usual blog posts. The company's 2012 Noob Guide was downloaded tens of thousands of times and won thousands of paying customers. One year on, people were still sharing it about once every hour.
  • OkCupid launched in 2004 but only blogged seriously from 2009. It wrote long, data-driven posts that took a month each. About a year later it ranked first for "online dating" and drew coverage from CNN and The New York Times.
  • Unbounce's partnership pitches went nowhere until the blog had readers. After that, integrations followed, including Salesforce.
• Free tools capture qualified leads, and email triggered by each user's actions converts them. Ch. 17 · Engineering as Marketing
  • Build a single-purpose tool for your ideal customer's pain. Ask only for a URL, handle or email.
  • Define activation, then email anyone who stalls. Dropbox reminds signups who never upload a file.
  • Emails triggered by behavior, such as an abandoned cart or a plan limit, beat unprompted pitches.
  • Dharmesh Shah built HubSpot's Marketing Grader in a few days to automate prospect checks. More than 3 million sites have used it. It drives a large share of HubSpot's fifty thousand plus monthly leads.
  • WP Engine trades a free speed report for an email. It then sends a seven-email course over about a month and pitches hosting near the end. Prospects who don't buy move to a monthly content list.
  • Customer.io emails each signup thirty minutes in, signed by founder Colin Nederkoorn and offering help. About 17 percent reply, and the replies drove many product fixes.
  • Microsites on their own domains keep paying off. Codecademy's Code Year signed up more than 450,000 people, nearly doubling its users. RJMetrics' Querymongo drives hundreds of leads a month. DuckDuckGo's DontTrack.us gets fresh traffic with NSA news and Data Privacy Day.
• A viral loop's strength is K, invites times conversion, plus cycle time, and it multiplies other channels even below 1. Ch. 16 · Viral Marketing
  • K above 1 means exponential growth. Above 0.5 helps a lot. Doubling either factor doubles K.
  • At the same K, a shorter cycle compounds far faster.
  • Loops need a product people use repeatedly, plus constant seeding from SEO or ads.
  • Hotmail's default signature and YouTube's embed code turned every use into an ad. YouTube's loop could finish in minutes.
  • Dropbox gives extra space to both sender and friend. Airbnb, Uber, PayPal and Gilt give account credits instead.
  • Even expert teams need one or two engineers for two to three months. Only one to three of every ten A/B tests are likely to win.
  • Computing K by country or age reveals viral pockets, for example Indonesia growing far faster than Australia. Copy written in the pocket's language can feed it.
• Communities compound when they have a mission, members connect with each other, and strict standards are set early. Ch. 25 · Community Building
  • Evangelists win over friends, who then become evangelists themselves.
  • Quality tends to dilute and drive good members away, so post the rules prominently and let members police them.
  • Community fits products that connect people. It fits ad and retargeting tools poorly.
  • Joel Spolsky and Jeff Atwood shaped Stack Overflow with their blog readers. A naming vote drew nearly seven thousand submissions. The site had hundreds of customers in its first days and thousands in its first month.
  • Rules allowing only practical, answerable questions sat in the FAQ, and users enforced them harder than Atwood would have. His biggest mistake was having no Meta, which told the most devoted fans to go away.
  • Chris McCann started Startup Digest by emailing twenty-two Bay Area friends about tech events. Twenty-second pitches at events took it to the low thousands within months. It now has more than 250,000 members.
  • Alexis Ohanian emailed thanks to anyone who wrote about reddit. He also sent early users shirts and stickers and ran an open bar tour at reddit's expense.

7Borrow other companies' customers through partners, platforms and a sales funnel

Startups can reach customers through organizations that already have them, as long as every partner deal, platform bet, sales step and trade show is chosen for its effect on the traction goal.

• Business development trades value with partners, and it only works when deals serve the Critical Path and a large pipeline absorbs the many failures. Ch. 18 · Business Development (BD)
  • Google's early traction came mostly from two partnerships, Netscape and Yahoo!, not only from a better product.
  • Keep fifty to one hundred potential partners in a pipeline, then focus on ten to twenty picked by attributes like revenue, reach or inventory, not brand name.
  • Partners judge deals by their own metrics. Reach the person who owns the metric your deal moves, through a warm introduction.
  • Half.com named three prelaunch needs: technology partners, one million items of inventory, and distribution. It picked one million because the number sounded big. The team traced where unsold books from publishers and stores like Borders piled up. Chris Fralic even flew to Atlanta to work a day in a used bookstore.
  • Delicious built The Washington Post a custom bookmarking interface so the Post's IT team was not needed. Other sites rushed in after the Post signed. Mozilla then promoted Delicious during the Firefox 2.0 upgrade, which more than tripled its users. A pursued Wikipedia deal failed.
  • Kayak powered an AOL-owned travel search engine. The distribution deal sent traffic right away and gave Kayak its first customers.
  • Term sheets are often one page: lifetime, exclusivity, payments, commitment, guarantees and revenue share. After closing, write a memo on how the deal got done, covering timing, contacts and sticking points.
• A few hand-made deals must prove value before a low-touch API can scale partnerships. Ch. 18 · Business Development (BD)
  • Delicious opened its API only after the Post and Mozilla deals worked. It then integrated with thousands of sites.
  • A public API alone attracts no one. Early deals create the demand.
  • Low-touch partnerships run through APIs, feeds, crawling and embed codes instead of personal negotiation. Delicious paid for its API with up-front engineering work.
  • SlideShare makes every slideshow embeddable, Disqus makes its comments easy to install, and SoundCloud opens its music library. Each integration widens the pool of partners.
• Sales for expensive products scales only through a repeatable funnel built from the buyer's side. Ch. 19 · Sales
  • Marketing generates and qualifies leads. Sales closes them, because cold calling costs more and works worse.
  • Calls follow SPIN order: situation, problem, implication, need-payoff. Each close ends in a concrete deliverable, a timeline and a yes or no.
  • Answer the buyer's questions on the website, track drop-offs and simplify each one away.
  • JBoss had over 5 million downloads but no contact details. It traded its formerly paid documentation for contact details and got over ten thousand leads a month. Heavy support-page visitors went to marketing for qualifying. Inside sales closed about 25 percent, against industry averages of 7 to 10 percent, at deals above $10,000.
  • Guidewire offered insurers a free consulting study of about seven to ten man-days, ending in a peer benchmark. In return it asked what the ideal system needed. It worked with over forty insurers this way.
  • Ask only one or two situation questions, because more of them lowers close rates. Fixes for drop-offs include SaaS instead of IT installs, free trials, case studies, and entry prices under $250/month for SMB or $10,000 for enterprises.
  • Funnels shaped around how the company thinks buying should work perform worse: You have to design it from the customer standpoint inwards, as opposed to your standpoint outwards.
• Existing platforms deliver users fastest when you fill a gap they leave open or arrive early on a new one. Ch. 21 · Existing Platforms
  • YouTube supplied the video hosting Myspace lacked. PayPal won eBay sellers, and Airbnb pulled hosts from Craigslist.
  • Early arrivals ride the platform's own launch marketing and get featured more easily.
  • Socialcam pushed signups through Facebook and Twitter, promoted user videos on both, and reached 60 million users in twelve months.
  • Trainyard's free version became number one in Italy after one blog review, with over 22,000 downloads that day. It passed 450,000 downloads in a week. An Apple feature then lifted downloads 50x.
  • Evernote shipped on day one of each App Store launch. It built a full iPad redesign without ever seeing an iPad. Android featured it six weeks straight, bringing hundreds of thousands of customers.
  • Platform features can be cut off. Airbnb's Post to Craigslist tool was eventually shut down, after it had sent tens of thousands of users back to Airbnb.
• Trade shows pay off only with set goals, a break-even estimate, meetings booked ahead and staff working the floor. Ch. 22 · Trade Shows
  • Walk a show as a guest the year before exhibiting, or ask past exhibitors.
  • Compare attendance cost to product price to see how many sales cover the show.
  • Worked example: 10,000 attendees, about 30 percent fitting the customer profile, leaves 3,000 targets. A $10,000 show turns profitable around the third sale of a $5,000 product. At $50 a unit it probably never pays.
  • RJMetrics founders approached people on the floor. Only 28 percent of conversations came from walk-ups, and going outbound tripled or quadrupled the show's value.
  • SureStop met Jamis at a show with only a prototype and learned the specs Jamis required. It built to those specs and formed a manufacturing relationship.
  • Jason Cohen aimed for five meetings a day at Smart Bear. He mailed discount cards that could be redeemed only at his booth.

8Every channel decays, so keep testing and rerun Bullseye

A working channel is a temporary asset, so traction depends on constant testing inside it and on finding a new channel at each stage.

• Every strategy loses power over time, and your first ideas have probably already started to fade. Ch. 5 · Traction Testing
  • Growth adviser Andrew Chen named this the Law of Shitty Click-Throughs.
  • The ideas that come to mind first are the common ones, so others are already using them.
  • The law in one line: Over time, all marketing strategies result in shitty click-through rates.
  • One serious competitor using the same approach can raise its cost and cut its results.
  • Zynga, maker of FarmVille, dominated Facebook's ad and sharing features while competition was thin. A gaming company today basically cannot grow there the way Zynga did, because the platform is too crowded and expensive.
  • Even an untapped strategy buys only a few months or a few years of strong results, and it still needs a strong product behind it.
• Inside the core channel, steady optimization and fresh strategy tests keep the channel productive. Ch. 5 · Traction Testing
  • At least one A/B test a week can make a channel two to three times more efficient.
  • Inner ring testing has two jobs: tune the current strategy and find better ones in the same channel.
  • Other channels may feed the core one, but they must not become separate focuses.
  • The levers differ by channel. For blogs: which blogs, what content, what call to action. For search ads: keywords, ad copy, demographics, landing pages.
  • A/B steps: pick one change, build a control and a variant, split people randomly, track one key metric, and keep B only if it wins clearly.
  • A blog-centered company can write guest posts on other blogs and buy Twitter and Facebook ads to amplify its best posts. Less standard feeders, such as community building, speaking or offline ads, may reveal novel strategies.
  • Fresh strategies come from new twists on established venues, from young platforms, or from brainstorming how each of the other eighteen channels could feed the core one.
• Young platforms give a real but temporary edge, and some of those bets will fail. Ch. 21 · Existing Platforms
  • Early movers face less competition and fewer big brands.
  • If the platform fails, the effort is lost.
  • Evernote built early for Nokia, Windows and BlackBerry smartphones. None moved the needle.
  • On Android, Evernote was featured in the store for six weeks straight while it was far less crowded. That brought hundreds of thousands of new customers, and Verizon's national Android push helped too.
  • Alex Pachikov still judges the approach worth it, because one win like Android more than covers several failures. Size platform bets with that ratio in mind.
• Tactics that moved the needle early stop mattering at scale, so each stage needs channels big enough for it. Ch. 3 · Traction Thinking
  • Growth comes in spurts. A channel unlocks a spike, saturates, then flattens until the next one opens.
  • Community building and viral marketing gain value as the company grows, because they scale with the user base.
  • Tech news articles worked for DuckDuckGo in phase I. Later they could not convert enough people to change its search numbers.
  • DuckDuckGo went from roughly one hundred searches per day to more than ten million. Every 10x jump needed a fresh channel. First came content marketing, then social and display ads, then publicity, and lately business development.
  • Late-stage reach is large. Adding 100,000 customers at 1 to 5 percent conversion means reaching 2 to 10 million targeted people. Test budgets grow in phases II and III for the same reason.
• When the core channel maxes out, rerun Bullseye, and if repeated rounds find nothing, fix the product. Ch. 4 · Bullseye
  • Saturation or rising costs signal that it is time to restart the process.
  • Several failed rounds point to a leaky product, not a missing channel.
  • Each new round should draw on data from earlier tests. Check the messaging and where each channel fell short.
  • Keep every brainstormed channel idea in a spreadsheet with educated guesses, so later rounds start from evidence rather than from zero.
  • Talk to founders a few steps ahead, especially those who failed at what you are about to try.

9Put it to work

On Monday you stop guessing at channels: you write a hard-number goal, set up tracking, and start three cheap channel tests next to product work.

1. Give traction half your time starting now Run channel experiments alongside product work, not after launch. Send a small stream of cold customers through the product to see where it leaks. Ch. 3 · Traction Thinking
   • The trap is calling early, leaky spend a waste. It shows where the holes are while there is still time to plug them.
   • Marc Andreessen's firm most often passes on founders it likes because they focus on product and ignore distribution.
2. Write one traction goal and the shortest path to it Pick a number that would change the company's outcome, such as break-even or a fundable growth rate. List only the milestones strictly needed to reach it, and drop any work that sits off that path. Ch. 6 · Critical Path
   • For 100 million searches a month, DuckDuckGo chose a faster site, a stronger mobile offering and more broadcast TV coverage. It left out images and auto-suggest, even though users kept asking for them.
   • Those same features became necessary once the goal changed to 1 percent of the search market, because mainstream users forgive much less.
3. Brainstorm a real strategy for all nineteen channels Write at least one specific idea per channel before ruling any out. Research what worked and failed for companies in your space. Ch. 4 · Bullseye
   • The trap is skipping channels out of habit. Fast-growing companies often win on channels their rivals never tried.
   • Talk to founders a few steps ahead, especially ones who failed at what you are attempting. Keep every idea in a spreadsheet for later rounds.
4. Set up tracking, then test about three channels cheaply Keep each phase I test under $1,000 and under a month. Each test should show what a customer costs, how many customers exist there, and whether they are the right ones. Ch. 5 · Traction Testing
   • Do not start a test until tracking exists. A spreadsheet is enough, but it must be there before money goes out.
   • Tuning ten social ads is optimization, not testing. Stay at the strategy level until a channel earns your focus.
5. Commit fully to the winner and cut the rest Put all traction effort into the one channel that tested best. Drop the weaker ones, even if they somewhat work. Ch. 4 · Bullseye
   • A startup tests SEM, trade shows and publicity, and SEM clearly wins. The tempting move is to keep the other two running. Finding more tactics inside SEM matters more.
6. Optimize the core channel every week Run at least one A/B test a week, which can make a channel two or three times more efficient. Use other channels only to feed the core one. Ch. 5 · Traction Testing
   • A company built on its blog writes guest posts elsewhere and buys Twitter and Facebook ads to push its best posts. Those feeders serve the core strategy. They are not separate focuses.
7. Rerun Bullseye when the channel stops moving the needle Every strategy decays, and growth stalls until a new channel unlocks. Check whether your current tactics can still reach your goal, and start the next round before they flatten out. Ch. 5 · Traction Testing
   • Tech news articles worked for DuckDuckGo in phase I. Later they could no longer convert enough people to change its search numbers.

10The book's words

These terms let a team talk about which channel to test, which one to commit to, and when to move on.

• Traction Hard evidence of customer demand, seen in a core metric like downloads, monthly revenue or weekly transactions.
  • The bar depends on the market: a few paying customers in enterprise software, possibly hundreds of thousands of users in consumer software.
• Traction channel One of nineteen broad routes to customers, from targeting blogs to community building.
  • A channel strategy is one specific approach inside a channel. Billboards, transit ads and magazine ads are strategies within offline ads.
• Bullseye A three-ring process: brainstorm every channel, test the few most promising, then focus on the one that works.
  • The framework grew out of talks with over forty founders. The idea is that Bullseye does for traction what Lean does for product development.
• Middle ring test A cheap, quick test of a candidate channel, under $1,000 and a month in phase I.
  • Spending just $250 on AdWords gives a rough read on whether search ads can work for a business.
• Core channel The single channel in the inner ring that gets all traction effort.
  • Other channels can still feed it, as publicity earns links for SEO, as long as one strategy stays dominant.
• Critical Path The fewest necessary milestones between you and your traction goal.
  • Critical Path is a way of working out which tasks you can skip. A hire or a feature only matters if you truly cannot do without it.
• Moving the needle Work that would have a measurable, significant effect on the traction goal if it succeeded.
  • The bar rises with scale. Adding 100,000 customers at 1 to 5 percent conversion means reaching 2 to 10 million targeted people.
• Law of Shitty Click-Throughs Every marketing strategy ends up with poor response rates as it gets crowded and ignored.
  • Andrew Chen coined it. Zynga dominated Facebook when competition was thin, and growing a game there the same way is now too crowded and expensive.
• Leaky bucket An early product that loses many of the customers you pay to bring in.
  • Cold customers, who have no tie to the founders, show where it leaks more honestly than friends do.
• Traction channel bias Ignoring promising channels because they are unfamiliar, unpleasant, or feel like schlep.
  • Business development and trade shows often get skipped as schlep. "Just because you hate talking on the phone doesn't mean your customers do."

11Chapter by chapter

Where each idea is in the book.

• PREFACE: Traction Trumps Everything Traction, meaning evidence that customers are buying or a free user base is growing, is the best way to raise a startup's odds of success. Most startups fail because they treat getting traction as an afterthought and only try the channels they already know.
• Ch. 2 · Traction Channels Traction is quantitative evidence of customer demand, and startups get it through nineteen distinct traction channels. Founders should consider all of them, because the best channel is hard to predict and is often one that competitors ignore.
• Ch. 3 · Traction Thinking Getting customers deserves as much attention as building the product, and both should run in parallel from day one. Measure traction against a hard-number goal, judge efforts by whether they move the needle, and look for engaged bright spots before deciding to pivot.
• Ch. 4 · Bullseye Most startups fail from poor distribution, and usually one traction channel works best at any stage. Bullseye finds it in three steps: brainstorm every channel, test the few most promising cheaply in parallel, then pour everything into the one that works.
• Ch. 5 · Traction Testing Traction comes from constant testing: cheap, quick tests across channels to find a promising channel strategy, then ongoing optimization and fresh strategy tests inside the one core channel you focus on. Keep everything numerical so you only pursue strategies that can move the needle.
• Ch. 6 · Critical Path Set one explicit traction goal that would materially change your company's outcome, then list only the milestones strictly needed to reach it. Use that Critical Path to decide what not to do, and fight your biases against unfamiliar or unpleasant traction channels.
• Ch. 7 · Targeting Blogs Getting featured on blogs and link-sharing communities that your prospects already read is one of the best ways to win a first wave of customers. It rarely scales into later phases, and that is acceptable, because unscalable tactics are often how companies get started.
• Ch. 8 · Publicity Publicity from traditional media now filters up from small blogs and forums to big outlets, so startups get press by pitching smaller sites first, building relationships with specific reporters, and packaging real milestones into emotional, shareable stories.
• Ch. 9 · Unconventional PR Unconventional PR, meaning publicity stunts and customer appreciation, is an uncrowded traction channel that can earn large press coverage and word of mouth for little money, though results are unpredictable.
• Ch. 10 · Search Engine Marketing (SEM) Paid search ads reach people who are already looking for a solution, so they are a cheap, fast way to test positioning and learn baseline metrics. If a campaign nears breakeven within a few weeks, SEM may be the channel to focus on.
• Ch. 11 · Social and Display Ads Display and social ads can reach very large or very narrow audiences at any budget. Social ads work best for indirect response: use them to amplify content people already share, build an audience, and convert that audience later.
• Ch. 12 · Offline Ads Offline ads (TV, radio, print, outdoor, direct mail) can be bought at almost any scale, so startups can test them cheaply in local markets. Because results are unpredictable and decay over time, the only reliable approach is to run many small tracked tests and scale what works.
• Ch. 13 · Search Engine Optimization (SEO) SEO brings in customers by ranking content in search results. It rests on two things, content matched to keywords and high-quality links, and a startup picks between a fat-head strategy and a long-tail strategy based on search volume and its realistic ability to rank on the first page.
• Ch. 14 · Content Marketing A company blog that publishes genuinely useful content can become a startup's largest source of customers, because good content keeps attracting search traffic, shares and press long after it is written. It needs at least six months, hand-driven promotion at first, and posts readers cannot find anywhere else.
• Ch. 15 · Email Marketing Email is a personal channel that works at every stage of the customer life cycle: finding prospects, activating them, retaining them, earning revenue and generating referrals. It works best when messages are personalized and triggered by what each customer actually does.
• Ch. 16 · Viral Marketing Viral marketing turns existing customers into a source of new ones through repeatable viral loops. Even when true exponential growth (K above 1) is out of reach, a loop built into the product and tuned through heavy testing can multiply the customers that other channels bring in, at very low cost.
• Ch. 17 · Engineering as Marketing Startups can turn engineering time into traction by building free tools, widgets and microsites that solve a problem for likely customers. These tools act as lasting marketing assets that keep producing leads long after the small up-front build.
• Ch. 18 · Business Development (BD) Business development gets traction by trading value with partners who can reach customers, not by selling directly. It works only when deals serve your traction goal, you understand the partner's incentives, and you work a large pipeline because most deals fail.
• Ch. 19 · Sales Sales turns leads into paying customers through personal contact, and it matters most for enterprise and expensive products. To scale it, build a repeatable funnel designed from the buyer's side: generate leads, qualify them, close them, and remove every point where prospects drop off.
• Ch. 20 · Affiliate Programs Affiliate programs let a startup pay only for real sales or qualified leads, which shifts the risk of bad traffic onto the affiliate. Start on an existing network with a simple payout set below what a customer is worth, then recruit customers and content creators as affiliates.
• Ch. 21 · Existing Platforms Large existing platforms like app stores, browser extension portals and social sites can deliver huge numbers of users fast, especially if you fill a gap the platform leaves open or arrive early on a new, uncrowded platform.
• Ch. 22 · Trade Shows Trade shows pack a year's worth of face time with customers, partners and press into a few days. They pay off only when a startup picks shows against clear goals and a budget, books meetings before arriving, and works the floor actively instead of waiting at the booth.
• Ch. 23 · Offline Events Running or sponsoring offline events, from small meetups to big conferences, can be a main source of traction. It works best when customers gather in person, ignore online ads, or sit behind long sales cycles, and few startups try it.
• Ch. 24 · Speaking Engagements Speaking at events puts a founder in front of a room of potential customers or partners who could move the business. Founders land these slots by pitching organizers, building credibility on small stages, and giving a few well-rehearsed story-driven talks that they extend through recording, social media and networking.
• Ch. 25 · Community Building Community building means investing in the ties among your customers so passionate users spread the word, keep quality high and help build the product. It works best when there is a clear mission, open feedback, connections among members and strict quality standards set early.
• APPENDIX: MIDDLE RING TESTS Each traction channel has a cheap, fast starter test that a phase I startup can run as a middle ring test in the Bullseye process. Each test should cost under a thousand dollars and take under a month, and better tests may come out of your own Bullseye work.