The big idea: Paid media buys reach and speed. Spend it on a clear objective, test small, and scale what works for the budget your client actually has.
See it in action
Run a Campaign Lab
You are running paid media for a client. Pick a goal, split the budget across channels, then choose who sees the ads. Watch the numbers change and read why.
Client: Trailhead Outfitters (invented outdoor gear brand). Monthly budget: $10,000. Product: the Ridge hiking boot, average order $90.
Show the math by channel
| Channel | Spend | Impr. | Clicks | Conv. | CPC | CPA | ROAS |
|---|
Illustrative assumptions (not benchmarks)
These rates are invented for teaching, shaped like real patterns. Real numbers vary by industry, season, creative and auction competition. Rates stay flat here so the math is easy to follow. Real costs usually rise as you scale a channel.
| Channel | CPM | CTR | CVR | Avg order |
|---|
One social ad set gets $3,000 of the budget. Who should see it, and with what ad?
Key concepts
- Paid, owned and earned: Paid is media you buy, owned is channels you control (site, email), and earned is attention others give you. They work best together.
- Objective first: Choose awareness, consideration or conversion before picking platforms, audiences or creative.
- Targeting: Reach people by demographics, interests, behaviors, keywords or retargeting people who already visited.
- Key metrics: CPM (cost per thousand impressions), CTR (click-through rate), CPC (cost per click), CPA (cost per acquisition) and ROAS (return on ad spend).
- Test, then scale: Start with a small budget, compare ads, and move money to what performs.
Learn it in depth
The deep dive
Paid media in plain language first, then the detail. Open the “Go deeper” boxes when you want more.
1. The basics in one minute
Paid media means you pay a platform to show your message to people. You are buying attention (someone sees the ad), action (someone clicks), or outcomes (someone buys or signs up). Almost every ad platform runs an auction each time an ad can be shown.
2. The paid media landscape
Pull channels catch demand that already exists: people are searching or shopping. Push channels create demand: you interrupt people who were not looking yet. Healthy plans use both.
Go deeper: how media gets bought
- Self-serve auction: you set budget, bid and targeting in the platform (Google Ads, Meta Ads Manager, TikTok Ads Manager). Most student and small-business campaigns live here.
- Programmatic: software buys impressions one at a time across many publishers. A DSP is the buyer’s tool. An SSP is the publisher’s tool. An exchange is the marketplace in between.
- Direct or guaranteed deals: you agree on price and volume with a publisher ahead of time. Common for CTV and big sponsorships.
- Pricing models: CPM (per thousand impressions), CPC (per click), CPV (per view), CPA (per action, offered on some platforms). You choose a pricing model and a bid strategy separately.
3. How ad auctions work
The highest bid does not always win. Platforms want ads people find useful, because annoyed users leave. So the winner is the ad with the best mix of bid and quality.
Ad rank ≈ bid × quality
Worked example: B bids the least and still wins, because its ad and landing page are far more relevant. In the classic version of the auction, B pays only enough to beat C: 15 ÷ 8 = $1.88, plus a cent. Better quality means you win more and pay less.
Go deeper: what “quality” means on each platform
- Search (Google, Microsoft): expected click-through rate, ad relevance to the keyword, and landing page experience. Google shows these as a 1 to 10 Quality Score per keyword. Real ad rank also considers ad assets, the search context and minimum thresholds, and exact pricing is not public. Treat the formula as a model, not a calculator.
- Social (Meta and others): Meta describes its auction as bid × estimated action rate (how likely this person is to do what you optimize for) plus ad quality (feedback such as hides, and low-quality signals like clickbait). The ad that creates the most value for both the advertiser and the user wins.
- Programmatic display and CTV: mostly first-price auctions now: you pay what you bid. Buyers use bid shading tools to avoid overpaying.
- Why it matters for your career: raising the bid is the expensive lever. Improving relevance (tighter keyword groups, matched creative, faster landing pages) is the cheap one.
4. The metrics glossary
Every example uses one Trailhead Outfitters campaign: $2,000 spend, 200,000 impressions, 80,000 people reached, 2,400 clicks, 60 sales, $5,400 revenue.
Go deeper: profit metrics most beginners miss
- Break-even ROAS = 1 ÷ gross margin. Trailhead keeps 40% of each sale after product costs, so break-even ROAS = 1 ÷ 0.40 = 2.5. Our ROAS of 2.7 is barely profitable before shipping and returns.
- Incremental ROAS (iROAS) counts only sales the ads caused. If a holdout test shows 20 of the 60 buyers would have bought anyway, incremental revenue is 40 × $90 = $3,600, and iROAS = $3,600 ÷ $2,000 = 1.8. Below break-even. The dashboard said 2.7.
- MER (marketing efficiency ratio) = total revenue ÷ total marketing spend, across all channels. It ignores who gets credit, so it is a good reality check against platform reports.
- CAC and LTV: customer acquisition cost versus customer lifetime value. A $60 CPA can be fine if the average customer buys three times.
5. Pixels, conversion tracking and privacy
Platforms can only optimize for conversions they can see. A pixel or tag is a snippet of code on your site that tells the ad platform “this person just bought.” Without it, a sales campaign is flying blind.
Go deeper: UTMs, modeled conversions and clean rooms
- UTM parameters are tags on your link (utm_source, utm_medium, utm_campaign) so your own analytics knows where a visit came from. Use one naming convention, in lowercase, for every campaign.
- Modeled conversions: when platforms cannot observe every conversion, they estimate the missing ones from patterns. Your reports now include educated guesses. That is normal; just know it.
- Data clean rooms let two companies (for example a brand and a retailer) match their customer data for measurement without either side seeing the other’s raw records. Retail media leans on them heavily.
- Test before launch: make a real test conversion and confirm it shows up in the platform. Broken tracking is the most common and most expensive launch mistake.
6. Attribution basics
Attribution is how you split credit for a sale across the ads someone saw. Different models tell very different stories about the same buyer. Pick a model and watch the credit move.
Credit for one $90 boot sale
Go deeper: why platform numbers never add up
- Every platform grades its own homework. Meta, Google and TikTok each count the sale if their ad was seen or clicked within their window. Add up their reports and you often get more sales than you actually had.
- Click-through vs view-through: a click-through conversion followed a click. A view-through conversion only followed seeing the ad. View-through is real for video and CTV, but easy to overcount.
- Incrementality tests answer the real question: what would have happened without the ads? Hold out a random group, or turn ads off in some regions (geo test), and compare.
- Marketing mix modeling (MMM) uses statistics on weekly spend and sales to estimate each channel’s contribution, without tracking individuals. Open-source MMM tools from Google and Meta made it common for mid-size brands.
- Google Ads and GA4 now default to data-driven attribution and dropped most fixed rule models. The rule models still matter for understanding how credit can be split.
7. AI and automated campaigns
Platforms now offer campaign types where the machine makes most decisions. Google’s Performance Max runs one campaign across Search, YouTube, Display, Gmail, Discover and Maps. Meta’s Advantage+ campaigns automate audiences, placements and creative combinations. You set the goal and the guardrails. The algorithm decides who, where and what.
Your job shifts from pulling levers to feeding the machine well: accurate conversion data, real profit values, lots of creative variety, and clear exclusions. Garbage in, garbage out, at scale.
Go deeper: risks and guardrails
- It optimizes what you measure. Optimize for leads and you may get cheap junk leads. Send lead quality or purchase value back so it learns what good looks like.
- It takes the easy wins. Automated campaigns can drift toward retargeting and your own brand searches, people who were going to buy anyway. Use new-customer goals, brand exclusions and incrementality tests.
- Learning periods. New or heavily edited campaigns need time and conversion volume to stabilize. Big daily changes reset learning.
- Less transparency. You see fewer placement and search-term details. Check the reports you do get and keep brand-safety exclusions on.
- Generative AI creative: platforms can now write headlines and generate or resize images and video. Review everything for accuracy, brand voice and required disclosures before it runs.
8. Launch checklist
Check items off as you go. Your progress stays on this page until you reload.
Frameworks from class
Consensual advertising
Advertising still works, but lazy ads annoy and drive people to block them. Paid media should be a fair value exchange: the viewer gets something useful or entertaining for their attention. Watch for retargeting that follows people around with something they already bought. It feels more intrusive than helpful.
Paid works best on top of earned
Paid reach amplifies a story people already want to share. REI’s #OptOutside spread through news coverage, social sharing and member email with zero paid ads. Build something worth talking about first, then use paid media to put it in front of more of the right people.
Try it
Use Meta’s Ad Library to look up a competitor of your client and study their active ads. What objective do they seem to be chasing? Then draft a paid plan for a small monthly budget: objective, platform, audience, two ad ideas to test, and the metric you’d judge them by.
Tips and best practices
- Write the objective and one KPI before you open an ad platform. The objective you pick tells the algorithm what to chase. Pick sales and judge it by ROAS or CPA; pick awareness and judge it by reach, frequency and CPM, not clicks.
- Test your tracking before you spend a dollar. Make a real test purchase or form fill and confirm it shows up in the ad platform and in your analytics. Broken conversion tracking quietly wastes more budget than bad creative.
- Know your break-even ROAS. Break-even ROAS is 1 divided by your gross margin, so a 40% margin needs a ROAS of 2.5 just to break even. A ROAS that looks good on the dashboard can still lose money.
- Improve relevance before you raise bids. Auctions reward bid times quality. Tighter keyword groups, ads that match the audience, and a fast landing page that keeps the ad’s promise often lower your cost more than a bigger bid.
- Start small, test one thing at a time, then scale the winner. Run two or three creative versions with enough budget to produce real results, and decide on a date you set in advance. Raise budgets in steps so the campaign’s learning does not reset.
- Do not trust any one platform’s report alone. Each platform counts sales it touched, so their totals usually add up to more than you sold. Compare against your own sales data, total revenue divided by total spend, and holdout tests when you can.
- Feed automated campaigns good inputs. Campaigns like Performance Max and Advantage+ decide placements and audiences for you. Your job is clean conversion data, real purchase values, lots of creative variety, and exclusions such as existing customers.
- Build first-party data with every campaign. Email sign-ups, accounts and purchase lists belong to the brand and survive privacy changes. Collect them with clear consent, and they become your best audiences for lookalikes and retargeting.
Quick check
Think through your answer, then select a question to reveal one.
What’s the difference between CPC and CPA?
CPC is what you pay per click. CPA is what you pay for each completed action, like a purchase or sign-up.
Why choose an objective before a platform?
The objective decides which platforms, audiences, creative and metrics make sense.
Go deeper
- Meta Ad Library
- Google Ads Transparency Center