DM-006·// CHANNELS··22 min read

Paid advertising: the ad auction, campaign structure, and what actually moves CAC

Paid is the most measurable channel and the most punishing. The auction does not care about your strategy deck. Here is the operator-level view that actually moves CAC.

Paid advertising is the most measurable channel and the most punishing. The auction does not care about your strategy deck. This guide is the operator-level view of how the auction works, why campaign structure is destiny, what actually moves CPI and cost per trial, and how the seam between an ad and the App Store listing or onboarding screen the user lands on makes or breaks the spend.

Two worlds share the “paid” name. Paid search (Google Ads) captures existing demand through auctions on keywords. Paid social (Meta, LinkedIn, TikTok) creates demand through interruption, targeting based on who someone is rather than what they are searching for. The mechanics, creative requirements, and economics are fundamentally different.

// 01The paid advertising landscape

Two paid paradigms

Intent-based (pull) platforms show ads in response to a query. Google Search and to a lesser extent Reddit fall here. Conversion rates tend to be high because you are meeting people who are already problem-aware or solution-aware. Interest-based (push) platforms show ads to people based on demographic, behavioral, or interest signals, regardless of whether they are actively looking. Meta, LinkedIn, TikTok, YouTube. Conversion rates are lower per impression, but the addressable audience is vastly larger and creative is the dominant variable.

  • Google Universal App Campaigns: Workhorse for mobile install volume. Pushes one creative pool across Search, Play, YouTube, and Display. Typical CPI for a consumer subscription app: $0.80 on Android, $1.40 on iOS.
  • Apple Search Ads: Pure intent: bid on App Store search terms. Highest-converting iOS channel; install-to-trial-start often runs above 18% on branded and category queries.
  • Meta Ads (especially IG Reels): Creative-driven; vertical video is the dominant format. Powerful for cold prospecting, retargeting, and lookalikes built from trial-start events.
  • TikTok Ads: Native UGC-style creative wins. Cheap reach for younger sleep-curious audiences; creative fatigue is fast, so plan for weekly refreshes.
  • YouTube creator integrations: Mid-roll spots inside wellness, productivity, or sleep-niche channels. Underrated for warming audiences before a Google or Meta retarget pass.
// FIGURE 01 · INTERACTIVE
Each ad platform has a distinct fingerprint

Click a platform to compare across intent, reach, targeting precision, creative demand, and cost efficiency.

IntentReachTargetingCreative NeedCost Eff.
Intent5/5
Reach4/5
Targeting3/5
Creative Need2/5
Cost Eff.4/5

// 02How the ad auction works

Ad Rank = Bid × Quality Score

The highest bid does not always win. An advertiser with a $2 bid and a Quality Score of 10 outranks an advertiser with a $5 bid and a Quality Score of 3.

What is Quality Score?

  • Expected click-through rate: Will people click this ad given the query? Historical ad performance informs this.
  • Ad relevance: How closely does your ad copy match the search query or targeted interest?
  • Landing page experience: Is the page useful, relevant, fast? Does it deliver on what the ad promised?
Actual CPC = (next advertiser’s Ad Rank ÷ your Quality Score) + $0.01

A higher Quality Score not only wins better positions but actively reduces the cost of every click. This is why “just bid higher” is almost always worse than “improve your relevance and landing page.” Improving the post-click experience (store listing copy, app preview video, onboarding flow) directly improves the third Quality Score input. Better post-click experiences mean higher Quality Scores, which mean better positions and lower CPCs simultaneously.

// FIGURE 02 · INTERACTIVE
Why the highest bid doesn't always win

Ad Rank = Bid × Quality Score. Drag the sliders to see how a higher Quality Score beats a higher bid, and lowers your actual CPC.

$3.00
8/10
Your result
Position #1
Actual CPC: $2.51 (not your full bid)
Auction result · ordered by Ad Rank
#1 · YouBid $3.00 × QS 8 = 24.0
#2 · Competitor ABid $5.00 × QS 4 = 20.0
#3 · Competitor BBid $2.50 × QS 7 = 17.5
The CRO connectionQuality Score is heavily influenced by landing page experience. Improving the page lifts QS, which simultaneously wins better positions and lowers cost per click: a compounding win.

// 03Campaign structure and hierarchy

Every ad platform uses a four-level hierarchy: Account → Campaign → Ad Group → Ad. Settings cascade down. The practitioner rule is SKAGs (Single Keyword Ad Groups) or tightly themed ad groups. Group only keywords that share intent, write ads that speak directly to that intent, and send them to a landing page that matches.

// FIGURE 03 · INTERACTIVE
Account → Campaign → Ad Group → Ad

Click a campaign to expand. Each level inherits settings from its parent but can override budget, targeting, and creative.

Account
A service business · $135/day total
Ad group · 2 ads
Exact brand match
"[your brand]" · "[your brand] reviews"
Ad-to-page congruence.If the ad says “Wake up rested in 7 days” and the App Store screenshot leads with “Track your steps,” you’ve broken the promise. The user bounces back to the feed and the install never happens. Good campaign structure forces good post-click congruence.

// 04Keyword match types and intent

Tighter targeting means lower volume but higher relevance and conversion. Most experienced practitioners begin new campaigns with phrase and exact match only, paired with a robust list of negative keywords. Once you have data, selectively test broad match for expansion, but always with aggressive negatives.

// FIGURE 04 · INTERACTIVE
The reach vs. relevance trade-off

Hover any match type to see how it widens the funnel, and what it lets through. Tight match types convert better; broad match types reach more queries.

Exact
[landing page audit]
Triggers only for queries with the same meaning as your keyword.
Reach15%
Relevance95%
Phrase
"landing page audit"
Triggers for queries that include the meaning of your phrase.
Reach40%
Relevance80%
Broad
landing page audit
Triggers for anything Google's AI thinks is related. High volume, high waste.
Reach90%
Relevance45%
Negative
-free, -jobs
Blocks your ads from showing on queries you want to avoid.
Reach0%
Relevance100%
Hover a card to see a worked example.
Awareness stage   Example query
─────────────────  ─────────────────────────────────────────
Unaware           No keyword targeting possible (use paid social)
Problem-aware     "why am i always tired"
Solution-aware    "best sleep tracking app"
Product-aware     "[app name] vs Sleep Cycle"
Most aware        "[your brand name] vs [competitor]"

// 05Creative testing and ad variants

On paid search, ad copy matters but the post-click experience usually matters more. On paid social, creative is the single biggest lever. Two apps targeting the same audience with the same budget will see wildly different CPI based entirely on their creative: the Reel hook in the first 1.5 seconds often does more work than the targeting.

  • Headline: Emotional vs factual. Question vs statement. Number vs description.
  • Image / video: For paid social, usually the biggest lever. Product vs lifestyle vs UGC vs abstract.
  • Body copy: Short vs long. Feature-focused vs benefit-focused. First person vs second person.
  • Call to action: "Start 7-day free trial" vs "See your sleep score" vs "Try it tonight."
  • Offer framing: Discount vs bonus vs guarantee vs social proof.
  • Audience: Same creative against different audience definitions.
// FIGURE 05 · INTERACTIVE
Plot every ad variant on two axes

Each circle is one ad: size scaled to spend. The quadrant tells you the action: scale, fix the landing page, retest, or kill.

CTR (%) →CONVERSION RATE (%) →WINNERSCLICKBAITLOW REACHKILLABCDEFGH
Ad B · Winner · Scale
Stop Losing 70% of Your Paid Traffic
CTR: 5.1%
Conv. rate: 6.2%
Spend: $1,400
Circle size represents spend. Bigger circles carry higher statistical confidence.
The orange quadrant is diagnostic gold. Ads with high CTR but low install-to-trial-start are strong creative with weak post-click experience. The buyer was hooked enough to install, but the onboarding lost them, exactly the seam a paid ops team should own end-to-end.

// 06Retargeting: the highest-ROI channel

Most people who tap an install ad never finish the install, and most who install never start a trial. With install-to-trial-start hovering around 18%, the bulk of paid-spend impressions evaporate on first contact. Retargeting recaptures that audience at a fraction of the cost of acquiring new users, because the audience is already warm.

// FIGURE 06
Why retargeting triples your conversion pool

A simulated funnel showing how retargeting recaptures lost visitors. Watch the inflection at step 5.

Ad Impressions
100,000
Clicked
2,500
−97.5%
Engaged
1,250
−50.0%
Left without converting
1,225
−2.0%
Retargeted (new pool)
1,225
−0.0%
Returned via retargeting
245
+20%
Converted
74
−69.8%
Recovery mathWithout retargeting, only 25 of 2,500 clickers convert (1% end-to-end). With retargeting you recapture 1,225 warm visitors at a fraction of the cost per click, converting another 49, a roughly 3× lift on paid campaign efficiency.
  • In-app event retargeting: SDK fires events for install, onboarding step, trial-start, paywall-view. Segmentation matters: a paywall-abandoner should see a different ad than someone who never opened the app a second time.
  • Customer-list retargeting: Upload hashed emails from trial signups; the platform matches to advertising IDs. Use this to suppress current subscribers and to retarget churned trialists with a win-back offer.
  • Video retargeting: Target people who watched 75% of your Reels or TikTok creative. Dramatically warmer than cold prospecting and often the cheapest install source you can find.
  • Lookalike audiences: Seed Meta or TikTok with your paying-subscriber list; the platform finds new users who behaviorally resemble them. A 1% lookalike of $48-LTV subscribers is often the best-performing cold audience.

// 07Budget allocation and bidding

How you split a budget across campaign types matters as much as the total.

  • Brand search (incl. Apple Search Ads brand): Cheapest paid channel. CPCs often <$1 and CPI often <$0.50. Allocate a small but meaningful amount. Sleep Cycle and other competitors will bid on your app name otherwise.
  • Non-brand search & UAC prospecting: The growth engine. People searching "best sleep tracking app" or scrolling Reels who don't yet know about you. Higher CPI; this is where incremental subscribers come from.
  • Retargeting: The efficiency layer. Low cost per trial-start because the audience already installed or watched the creative. Limited by audience size: you can only retarget those you've reached.
// FIGURE 07 · INTERACTIVE
$3,000/mo across three campaign types

Adjust the sliders. Lead volume and blended CPL update against realistic B2B service benchmarks.

LEADS/MO153
20%
55%
25%
Blended CPL
$19.57
Brand Search
Cheapest. Capped by brand search volume.
$600 · CPL $8 · 75 leads
Non-Brand Search
Scalable but expensive. The growth engine.
$1,650 · CPL $45 · 37 leads
Retargeting
Warm audience. Requires upstream traffic to fuel.
$750 · CPL $18 · 42 leads
The automation paradox. Platform-automated bidding (Target CPA, Target ROAS, Maximize Conversions) works well when you have enough conversion data to train it (~30+ conversions per ad group per month). Below that threshold, automated bidding often performs worse than manual bidding because algorithms lack signal.

// 08Six things to carry forward

  • 01: Paid search captures existing demand; paid social creates new demand. The same ad won't work in both places.
  • 02: The highest bid does not always win. Quality Score lowers your CPC and raises your position simultaneously.
  • 03: Campaign structure is destiny. Tightly themed ad groups → relevant ads → relevant pages → higher QS → cleaner data.
  • 04: Match type is a knob on the reach-relevance trade-off. Start tight, expand with data, build robust negative lists.
  • 05: Retargeting is almost always the highest-ROI paid channel because the audience is warm.
  • 06: Automation requires data. Below 30 conversions/month, manual bidding often beats Target CPA.
// PUT IT TO WORK

You can run an experiment from this article in under five minutes.

Pick the strongest claim above. Pre-fill it as a real experiment in Xi — hypothesis, metric, success and kill thresholds — and you’ll have evidence by next month, not opinion.

Run an experiment