traction（traction 搜索）

Note: This skill is independent analysis and commentary, not a reproduction of the original text. It synthesizes the book's core ideas with modern startup practice, surfaces where frameworks are outdated or incomplete, and integrates perspectives from adjacent disciplines. For the full argument and context, read the original book.

Traction

"Almost every failed startup has a product. What failed startups don't have is enough customers." - Gabriel Weinberg

Should You Use This Skill?

Do you have a product (or MVP) but not enough customers?
|-- YES --> Do you know which acquisition channel works?
|   |-- NO  --> THIS SKILL. Run the Bullseye Framework.
|   +-- YES --> Is that channel still working?
|               |-- YES --> Keep going. Re-run Bullseye when it saturates.
|               +-- NO  --> THIS SKILL. All channels saturate eventually.
+-- NO  --> Do you have a product at all?
            |-- NO  --> Use Lean Startup first. But read the 50% Rule below.
            +-- YES, but nobody wants it --> Use Lean Startup / Four Steps first.

The Core Insight

Poor distribution - not product - is the number one cause of startup failure. Most founders are product people who neglect traction. They assume that if they build something great, customers will come. They won't.

The 50% Rule

"Spend 50% of your time on product and 50% on traction."

This is not "build first, then market." It is: pursue product development and traction in parallel from day one. Traction work informs product decisions. Product decisions constrain traction options.

Marketo built SEO and a blog before the product existed. By launch, they had a 14,000-buyer pipeline waiting.

Why Founders Get This Wrong

FOUNDER BIAS:                          REALITY:
"We just need a better product"        You need customers, not features
"Marketing is for later"               Traction takes as long as product
"We'll figure it out after launch"     Post-launch is too late to start
"Our product will sell itself"          No product sells itself at zero users
"We tried ads, they didn't work"       You tried ONE of 19 channels

The 19 Traction Channels

Every startup's growth comes from one or more of these channels. Most founders only consider 3-4 of them. The rest are invisible - and that's where the opportunity is.

"Founders tend to use traction channels they're already familiar with. This is a mistake. The channel that will work for your startup is probably one you haven't considered."

The Bullseye Framework

Five steps to find the ONE channel that moves the needle right now.

Step 1: Brainstorm

Generate at least one concrete idea for every one of the 19 channels. Not "we could try SEO" but "we could write comparison pages for every competitor and rank for '[competitor] alternative'."

Audit your biases:

• Which channels are you ignoring because they seem "unsexy"?
• Which channels would your competitors never try?
• Which channels feel like too much work? (That's where the opportunity is.)

Step 2: Rank

Sort all 19 into three columns:

Step 3: Prioritize

Narrow Column A to exactly 3 channels. Not 5. Not 1. Three.

For each, answer four questions:

1. Cost to acquire a customer through this channel?
2. How many customers are available through this channel?
3. Are the customers the RIGHT customers (quality)?
4. What's the time to acquire (cycle time)?

Step 4: Test

Run cheap, parallel tests on all three inner-circle channels.

Testing rules:

• Spend ~$250 per channel to get a signal (not $25,000)
• Run 2-4 ads, not 40
• Test whether the channel CAN work, not whether it's optimized
• Time-box: 1-2 weeks per test
• Measure: cost per acquisition, volume, quality, cycle time

"The goal of testing is not to get traction. It is to determine whether a channel could work."

Step 5: Focus

At any given time, one channel dominates your customer acquisition. Once you find it through testing, pour all traction effort into it. Optimize aggressively. Wring it dry.

When the channel saturates (rising costs, diminishing returns), re-run the Bullseye Framework from Step 1.

BULLSEYE FLOW:
  Brainstorm (all 19) --> Rank (A/B/C) --> Prioritize (top 3)
       --> Test (cheap, parallel) --> Focus (one channel)
                                         |
                                         v
                                    Channel saturates?
                                    |-- NO --> Keep optimizing
                                    +-- YES --> Re-run Bullseye

Three Product Phases

Different things "move the needle" at each phase:

Phase-Channel Map

PHASE I (finding product/market fit):
  Best: Targeting Blogs, Speaking, Community Building,
        Unconventional PR, Offline Events, Sales (direct)
  Why: unscalable, high-touch, feedback-rich

PHASE II (growth with product/market fit):
  Best: SEO, SEM, Content Marketing, Email Marketing,
        Viral Marketing, Engineering as Marketing
  Why: scalable, measurable, compounding

PHASE III (scaling):
  Best: Offline Ads, Affiliate Programs, Business Development,
        Existing Platforms, Trade Shows
  Why: volume, partnerships, broad reach

Critical Path

A single Traction Goal (e.g., "1% search market share" or "$100K MRR"), then an ordered list of absolutely necessary milestones with dependencies.

Every activity is judged against the critical path: "If it is not on the path, don't do it."

TRACTION GOAL: _________________________

CRITICAL PATH:
  Milestone 1: _____________ (dependency: none)
  Milestone 2: _____________ (dependency: Milestone 1)
  Milestone 3: _____________ (dependency: Milestone 1)
  Milestone 4: _____________ (dependency: Milestones 2+3)
  ...
  TRACTION GOAL

Everything not on this path is a distraction.

Channel Summaries

1. Viral Marketing

Six loop types: word of mouth, inherent (product usage exposes non-users), collaboration (multi-user features), embedded (signatures, badges), incentivized (referral rewards), social broadcasts (auto-sharing).

Math: Viral coefficient K = invites sent per user x conversion rate. K > 1 = exponential growth. Even K > 0.5 significantly amplifies other channels.

Critical insight: Viral cycle TIME matters more than K. YouTube's cycle time is minutes (watch, share, watch). If your cycle time is weeks, even K > 1 is slow.

2. Public Relations

The filter-up media chain: Small forums (HN, reddit) --> mid-tier blogs (TechCrunch, Lifehacker) --> major outlets (NYT, CNN). Don't pitch the top. Pitch the small site that the top reads.

Use HARO (Help A Reporter Out) to respond to journalist queries. Build reporter relationships before you need them.

3. Unconventional PR

Two types: publicity stunts (engineered spectacles) and customer appreciation (small, scalable acts of kindness).

WePay dropped a 600-pound block of ice with frozen money at PayPal's dev conference. Dollar Shave Club's $5K video got 12K customers in 2 days - the same reach as Gillette's $60M annual budget.

4. Search Engine Marketing (SEM)

Only works when existing search demand exists. Long-tail keywords + negative keywords. Quality Score is primarily driven by CTR.

Benchmark: ~$250 spend gives a rough channel read. Average AdWords CTR ~2%.

5. Social & Display Ads

Demand generation (creating awareness) vs. demand harvesting (capturing existing intent). Social ads work best when you amplify what's already getting organic traction.

6. Offline Ads

Remnant inventory (unsold ad space) can be negotiated up to 90% off rate card. DuckDuckGo's $7K/month San Francisco privacy billboard doubled their userbase that month.

7. Search Engine Optimization (SEO)

Two strategies: Fat-head (top 18.5% of search volume, direct category terms) and Long-tail (70% of volume, specific multi-word queries). Validate fat-head terms with AdWords first. Content + links. Stay white-hat.

8. Content Marketing

Takes ~6 months minimum to gain traction. Feeds 8 other channels. HubSpot gets 70-80% of leads from content vs. 10% industry average. Inbound leads cost half as much and close 100% better.

9. Email Marketing

For acquisition, engagement, retention, revenue, AND referrals. Lifecycle emails keyed to drop-off points. Patrick McKenzie: email subscribers were 70x more likely to buy than other leads.

10. Engineering as Marketing

Build free tools as permanent marketing assets. HubSpot's Marketing Grader: 3M+ sites used. Codecademy's Code Year: 450K signups, nearly doubled userbase. Host on its own SEO-friendly domain.

11. Targeting Blogs

Best Phase I channel. Find niche blogs your audience reads. Mint acquired 20K pre-launch signups through blog outreach, 1M users in 6 months.

12. Business Development

BD trades value through partnerships; sales trades dollars for product. Five types: standard, joint ventures, licensing, distribution, supply. Pipeline of 50-100 active prospects.

13. Sales

For high-priced products. SPIN Selling framework: Situation, Problem, Implication, Need-payoff. Funnel: Generate --> Qualify --> Close. Every drop-off is a blockage to remove.

14. Affiliate Programs

$2B+ retail affiliate market. Best for high-CAC verticals: financial services, insurance, dating, hosting. Advertiser bears no click risk - commission only on clean conversions.

15. Existing Platforms

App stores, browser extensions, social platforms, marketplaces. Be early on emerging platforms. AirBnB's Craigslist integration. Evernote's multi-platform day-one strategy.

16. Trade Shows

Best for B2B in compressed time. Outbound (walking the floor) is 3-4x more valuable than inbound (waiting at your booth). Visit as a guest the year before exhibiting.

17. Offline Events

Conferences, meetups, parties. Twitter's SXSW 2007: $11K on flat-panel screens, tweets jumped 20K/day to 60K/day. "Just about the only money Twitter's ever spent on marketing."

18. Speaking Engagements

Best for B2B/enterprise. Two-talk doctrine: build two solid talks, reuse with light tweaks. Each slide should be a complete story so a 60-min talk compresses to 20 by removing slides.

19. Community Building

Five pillars: establish mission, foster cross-connections (member-to-member, not just member-to-company), communicate personally, be transparent, ensure quality. DuckDuckGo: every employee was a community member first.

Testing Principles

Test vs. Optimize

Testing answers: "Could this channel work at all?" (2 ads + 2 landing pages, ~$250)

Optimizing answers: "How do we maximize this channel?" (A/B tests, landing page iterations, copy refinement)

Don't optimize a channel before you've proven it can work. Don't test 40 variations when 4 will give you a signal.

The Law of Shitty Click-Throughs

All marketing strategies decay over time. - Andrew Chen

Every channel saturates. Banner ads once had 78% CTR. Now: 0.1%. This means:

• No channel works forever
• Early movers in a channel have an advantage
• You must constantly re-evaluate (re-run Bullseye)

Measurement

Four things to measure per channel test:

1. Cost per acquisition (CPA) - what does one customer cost?
2. Volume - how many customers are available?
3. Quality - are these the RIGHT customers?
4. Time - how long from first touch to customer?

Tools: Mixpanel, Clicky, Optimizely, Unbounce

Decision Trees

"Which channel should we try first?"

What phase are you in?
|-- Phase I (pre-product/market fit)
|   --> Do you have a niche audience?
|       |-- YES --> Targeting Blogs + Community Building + Speaking
|       +-- NO  --> Unconventional PR + Offline Events + Direct Sales
|
|-- Phase II (product works, need growth)
|   --> Is there existing search demand for your category?
|       |-- YES --> SEO + SEM + Content Marketing
|       +-- NO  --> Is your product inherently shareable?
|                   |-- YES --> Viral Marketing + Social Ads
|                   +-- NO  --> Email Marketing + Engineering as Marketing
|
+-- Phase III (scaling what works)
    --> What's your price point?
        |-- High (enterprise) --> BD + Trade Shows + Sales
        |-- Medium (SaaS) --> Affiliate + Existing Platforms + SEO
        +-- Low (consumer) --> Offline Ads + Viral + Social Ads

"Our current channel stopped working"

Has CPA risen above LTV?
|-- YES --> Is this channel-wide (competitors too)?
|   |-- YES --> Channel is saturating. Re-run Bullseye.
|   +-- NO  --> Your creative/targeting is stale. Refresh before abandoning.
+-- NO  --> Has volume plateaued?
            |-- YES --> You've captured available demand. Layer a second channel
            |           or re-run Bullseye for new primary.
            +-- NO  --> Keep optimizing. You haven't maxed it yet.

Critical Numbers & Rules of Thumb

Common Failure Patterns

Modern Relevance (2014 --> 2026)

Where Traction Still Applies

• The 19-channel taxonomy is comprehensive and still valid
• Bullseye Framework is timeless - systematic channel testing works in any era
• The 50% rule is still the most common founder mistake to violate
• "Do things that don't scale" remains the best Phase I advice

Where It Shows Its Age

• AI-native distribution - LLM recommendations, AI agents, and chatbot integrations are a new channel not covered
• Creator economy - influencer marketing has matured into its own channel beyond "targeting blogs"
• Product-led growth (PLG) - free tiers and self-serve onboarding blur the line between product and traction
• Short-form video - TikTok/Reels/Shorts didn't exist; social ads chapter is pre-TikTok
• Privacy changes - iOS 14.5 ATT, cookie deprecation, and GDPR have materially changed paid acquisition
• Tool recommendations - many specific tools mentioned are defunct or superseded

What Weinberg Got Permanently Right

• Systematic channel testing beats intuition
• Most founders only consider 3-4 of 19 available channels
• One channel dominates at any given time
• All channels saturate - constant re-evaluation required
• Traction and product development should run in parallel
• The overlooked, "unsexy" channel is often the winner

Supporting Files

• frameworks.md - Bullseye Framework deep dive, critical path methodology, channel testing protocol, each of the 19 channels with detailed tactics and mechanics, phase-channel matching, Law of Shitty Click-Throughs
• cases.md - Mint ($170M acquisition via blog targeting), Half.com (unconventional PR), DuckDuckGo (billboard + community), Dollar Shave Club ($5K video), JBoss ($350M acquisition via sales funnel), HubSpot (content + engineering as marketing), Dropbox (viral), Twitter SXSW (offline events), Evernote (existing platforms), WePay (publicity stunt)
• examples.md - Bullseye brainstorm worksheet, channel test template, critical path planner, CPA/LTV calculator, channel scoring matrix, viral coefficient calculator, end-to-end example startup
• integration.md - Relationship to Lean Startup (traction testing IS Build-Measure-Learn for growth), relationship to Four Steps (Customer Creation = Traction), relationship to $100M Leads (tactical overlap), relationship to Crossing the Chasm (beachhead channel selection), master sequence

Honest Scope of the Book

• Published: 2014 (Second Edition: 2015)
• Examples: Mostly 2008-2014 startups. Some are now household names (Dropbox, Twitter, Evernote); others are defunct or acquired.
• Empirical base: Author interviews with 40+ founders and growth experts. Practical advice, not academic research.
• Where it shines: Channel selection and testing methodology. The Bullseye Framework is the most systematic approach to growth channel selection available.
• Where it's weak: Light on retention and engagement (focuses on acquisition). Doesn't address pricing or positioning. Channel descriptions are tactical snapshots that age quickly. No framework for when to STOP a channel beyond "it saturates."
• Author's context: Gabriel Weinberg is the founder/CEO of DuckDuckGo. Justin Mares is a serial entrepreneur. Both have direct experience with many of the channels described.