Skip to content
Zero2OneZero2One

Paid Ads (Google / Meta / Reddit)

Buy attention on Google, Meta, Reddit or LinkedIn — but only once you know what a customer is worth. Paid ads scale what already works; they rarely create demand that isn't there.

Cost
$$$
Time to results
Days
Difficulty
Medium
Best for
B2B + B2C
0/7

Proof first

8 companies already did this.

Half of these are warnings. Paid ads are easy to start and easy to waste money on — the filings and experiments show both sides.

Deep dives

Two case studies, start to finish.

Dropbox — when search can't find you

Dropbox · 2009

Paid search · $233–$388 per $99 customer

What they built: paid search and affiliate campaigns

What they didAfter launching in 2009, Dropbox hired an experienced search and affiliate marketer to buy traffic. Each new customer cost $233–$388, for a product that sold for $99.

Why it workedNobody was searching for a product category that barely existed yet, and the obvious keywords were already expensive. Drew Houston concluded that search harvests demand rather than creating it. Dropbox switched to a two-sided referral programme, which permanently lifted sign-ups by 60%.

Copy thisBefore buying search ads, check people already search for your solution — and that your price covers what a click costs.

Airbnb — test what ads really add

Airbnb · 2020

Marketing near zero · ~91% of traffic stayed

What they built: suspending almost all marketing spend

What they didWhen the pandemic hit, Airbnb suspended almost all discretionary marketing. Its IPO filing reports that about 91% of traffic still came through direct or unpaid channels. Its 2020 annual report shows performance marketing down $540.5M.

Why it workedMuch of what the ads had been "buying" was demand that would have arrived anyway. Airbnb then shifted its mix towards brand marketing. Brian Chesky said traffic held at 95% of the year before with marketing taken down to near zero.

Copy thisPause spend in one region or on one keyword set for a few weeks. If sales barely move, the ads weren't adding much.

The guide

7 steps, each anchored to a real example.

Tick each step off as you go. Your progress is saved in this browser.

Step 1 · ~1 day

Know your numbers before you spend

Work out your margin, how much a customer spends over a year, and therefore the most you can pay to win one. If you can't answer that, you're not ready for paid ads.

SpreadsheetStripe / ShopifyGoogle Analytics 4

How Warby Parker did this step

Illustration

Warby Parker's IPO filing says its customer acquisition cost averaged 15% of average revenue per customer, and that it is profitable on a customer's first order.

Source [2] ↓

Step 2 · ~2 hours

Start where people already search for it

Search ads catch people who already want your kind of product. If nobody searches for it yet, search ads will be expensive and slow — try social ads that show the problem instead.

Google Keyword PlannerGoogle AdsMicrosoft Ads

How Dropbox did this step

"Search is a way to harvest demand, not create it."

Paraphrased — see source

Dropbox paid $233–$388 per customer through search for a $99 product. Houston's lesson: search harvests demand, it doesn't create it.

Source [1] ↓

Step 3 · ~1 week

Test many ads, cheaply

Run several hooks, images and videos with a small daily budget each. Kill losers fast and move money to the winners. The ad itself matters more than the targeting.

Meta Ads ManagerMeta Ad LibraryReddit Ads

How Ridge did this step

"Our north star for an ad is the one-day return on ad spend from actual clicks."

Paraphrased — see source

Ridge's CEO says the company tests hundreds of new ads a week and judges each one on its one-day return from actual clicks.

Source [6] ↓

Step 4 · ~30 min / day

Scale the one channel that works

When one platform pays back, go deep there before adding another. Raise budgets gradually and watch cost per customer as you do.

Meta Ads ManagerTriple WhaleNorthbeam

How MVMT did this step

Illustration

MVMT advertised heavily on Facebook and reported about $71M in revenue for 2017. Movado bought it for about $100M up front in 2018.

Source [3] ↓

Step 5 · ~2 weeks

Prove the ads add something

Switch ads off in a few regions, or for some keywords, and compare sales with the rest. If nothing changes, the ads were taking credit for sales you'd have made anyway.

Geo holdout testGoogle Ads experiments

How eBay did this step

Illustration

In an experiment, eBay switched off paid search in some markets. Researchers found paid traffic was almost fully replaced by free channels, and the effect of brand ads on sales was indistinguishable from zero.

Source [9] ↓

Step 6 · ~1 hour

Protect your own name — cheaply

Check what appears when someone searches your brand. If competitors bid on it, a small brand campaign may be worth it. If they don't, you may not need one at all.

Google Ads brand campaignSearch console

How Basecamp did this step

"You basically have to pay protection money to Google to even have a chance."

Exact quote — see source

Basecamp found competitors' ads showing above its own first result, so it ran an ad on its own name saying it didn't want to run the ad.

Source [10] ↓

Step 7 · ~Monthly

Watch the whole picture, not just the dashboard

Every platform takes credit for the same sale. Each month, compare total marketing spend with total new revenue, and ask new customers where they heard about you.

Blended spend sheetPost-purchase survey

How Airbnb did this step

Illustration

When Airbnb cut almost all marketing in 2020, about 91% of its traffic still came through direct or unpaid channels.

Source [7] ↓

Before you start

Pros & cons.

Works in your favour

  • Fast — results in days, not months
  • Precise targeting by search, interest or job title
  • Every click and sale can be measured
  • Scales quickly once it works
  • A quick way to test which message lands

Go in with eyes open

  • Expensive, and it stops the day you stop paying
  • Ads often take credit for sales you'd have made anyway
  • New categories have no search demand to buy (see Dropbox)
  • Costs rise as competitors bid on the same buyers
  • Platforms change rules and tracking without warning

Make it yours

Three ways to run it.

The Warby Parker model

Pay back on the first order

Only buy customers who are profitable on their first purchase. Best for products with healthy margins and repeat buying.

The Ridge model

Creative testing at volume

Make lots of ads, judge each on short-term return, and scale the winners. Best for consumer brands with a visual product.

The Airbnb model

Brand first, performance second

Build demand through brand and word of mouth; use paid ads only where they clearly add sales. Best once people already know your name.

Copy and paste

Templates & resources.

Unit economics sheet

fill this in before spending anything

Average order value: ${x}
Gross margin: {y}%
Orders per customer per year: {n}
First-year profit per customer: ${x × y% × n}

Max you can pay per customer (cost to acquire): ${…}
Target (leave room for error): ${… × 0.6}

If your ads cost more than the target per customer → stop and fix the offer or the ad.

Ad test plan

one week, small budget

Goal: find one ad that wins customers under ${target}
Budget: ${x}/day per ad set, 7 days

Hooks to test:
1. The problem: "{pain}?"
2. The result: "{outcome} in {time}"
3. Proof: "{number} {customers} use it"

Rules: kill any ad above 2× target after ${spend}. Double budget on any ad under target.

Holdout test

check what the ads really add

Pick {2–3} regions (or keyword groups) as holdouts.
Turn ads OFF there for {2–4 weeks}; leave them ON everywhere else.

Compare new customers per week:
Holdout before: __  after: __
Ads-on before: __  after: __

If the holdouts barely drop → the ads were mostly taking credit.

From the showcase

Be the first founder featured here.

The famous examples above are proof it works. The next founder reading this wants proof it works for someone like them — that could be you.

Used “Paid Ads (Google / Meta / Reddit)”? Submit your product →

Discussion

Tried this? Tell us how it went.

Share a result, ask a question, or say what you'd do differently. The next founder to try this will thank you.

Loading the discussion…

NOW MAKE IT YOURS

You've seen how 8 companies did it. Your turn.

We're building an AI planner that turns this playbook into a step-by-step plan for your product. Leave your email for early access — one message when it opens, nothing else.

Already using this tactic? Show other founders what you built. Get featured →

Receipts

Sources & references.

Last reviewed

  1. 1.
    Dropbox: Startup Lessons Learned (cost per acquisition $233–$388 for a $99 product) (opens in a new tab)

    Drew Houston, 2010

  2. 2.
    Warby Parker registration statement (S-1): customer acquisition cost (opens in a new tab)

    US SEC, 2021

  3. 3.
    Movado Group acquires watch startup MVMT (opens in a new tab)

    TechCrunch, 2018

  4. 4.
    Movado Group announcement: MVMT revenue of ~$71 million in 2017 (opens in a new tab)

    US SEC (8-K), 2018

  5. 5.
    Hims & Hers annual report 2024: marketing expenses (opens in a new tab)

    US SEC (10-K), 2025

  6. 6.
    Ridge Wallet CEO on ad testing and marketing efficiency (opens in a new tab)

    Practical Ecommerce, 2022

  7. 7.
    Airbnb registration statement (S-1): ~91% of traffic direct or unpaid (opens in a new tab)

    US SEC, 2020

  8. 8.
    Airbnb annual report 2020: performance marketing down $540.5M (opens in a new tab)

    US SEC (10-K), 2021

  9. 9.
    Does paid search advertising work? (eBay experiments) (opens in a new tab)

    UC Berkeley Haas, 2014

  10. 10.
    Basecamp CEO calls Google search ads a shakedown (opens in a new tab)

    CNBC, 2019

  11. 11.
    Chesky: marketing near zero, 95% of traffic (opens in a new tab)

    CNBC, 2021

Company names and logos are trademarks of their owners, shown only to identify the companies discussed. Spot an error? Tell us below — we correct and date every change.

Corrections

No corrections so far.

Spot an error?

Tell us what's wrong — we fix it and log the change here.

Keep going

Related playbooks.

Referral & Word-of-Mouth Programs

$ · Months · Medium — what Dropbox did instead

Read playbook →

Content Marketing (SEO / YouTube)

Free · Months · Hard — earn the clicks instead

Read playbook →

Influencer Marketing (Post-Launch)

$$ · Weeks · Medium — paid reach through creators

Read playbook →