+55 XP

Paid digital marketing: foundations & core concepts

The unit you buy in paid digital is smaller than most marketing plans admit: one impression, sold at auction, settled in the time a feed takes to load. Everything above it (channel mix, agency fees, the board conversation about customer acquisition cost) is an aggregation of hundreds of millions of those individual auctions. A CMO who cannot describe what happens inside one of them is delegating the mechanics and inheriting the invoice. This lesson defines that object and the vocabulary the rest of this module uses without stopping to explain it: bid, ROAS, CAC, attribution window.

What paid digital marketing actually is

Paid digital marketing means paying an owner of attention to place your message in front of an audience you specify, at a price set by a live auction rather than a rate card. Unlike SEO or organic content, which accumulate slowly and keep working after you stop, paid delivers volume the hour you fund it and stops the hour you don't. That trade makes it a capital allocation decision before it is a creative one.

The object being bought is the impression. When a search results page or a feed loads, the platform gathers every advertiser eligible for that specific person in that specific moment, ranks them, and renders one ad. That event is the impression: one person, one slot, one auction, one winner. You never buy a campaign. You buy several hundred thousand of these one at a time, and what you call a campaign is the set of instructions the machine follows about which ones to enter and how hard to compete.

Your bid is the maximum you are willing to pay for the outcome you selected. On Google, Meta, TikTok and LinkedIn the raw number is usually hidden behind an objective (target CPA, target ROAS, maximise conversions, cost cap), but a bid is still being submitted on your behalf in every auction, and the objective you chose is what sets it. Billing then happens in one of three units: CPC, where you pay per click (standard on Google Search and LinkedIn); CPM, where you pay per thousand impressions regardless of clicks (standard on Meta, TikTok and programmatic display); and CPA, where you pay only when a defined action happens. Note the mismatch that confuses most reporting: you are billed per click or per thousand impressions, but the machine is buying single impressions, and it decides which ones based on what it predicts you will get from them.

All four of these platforms sell the inventory described here. Their published guidance documents their own auction; it is not neutral advice about how much of your budget belongs in it.

Sub-concept 1: the intent spectrum

Paid channels do not reach the same buyer at the same moment, and the price differences follow directly from that. Google Search captures demand that already exists. When an operations director types "warehouse management software pricing", they have declared themselves. Bidding on that query is demand capture. Meta and TikTok interrupt people who were not thinking about you at all, which is demand creation, and TikTok sits furthest from intent of the four: no query, sound-on, creative doing the work of targeting. LinkedIn sits in between, reaching people in a work mindset with firmographic precision but usually with no active purchase intent behind the impression.

The practical consequence is that a CPA from Google Search and a CPA from TikTok are not comparable numbers. One harvests demand somebody else created, often your own brand and content. The other creates it. Ranking channels on a single cost-per-conversion league table will always kill demand creation first, and the effect shows up in search volume two or three quarters later.

Sub-concept 2: how the auction actually decides

Winning is not simply bidding the most. Google ranks ads on Ad Rank, which combines your bid with quality signals: expected click-through rate, ad relevance to the query, and landing page experience. Quality Score, the 1 to 10 diagnostic Google reports back to you, is the visible summary of those signals. A competitor bidding $8 with a Quality Score of 9 can outrank you at $12 with a Quality Score of 4 and pay less per click for the privilege. Meta runs the same logic under different names, ranking on bid multiplied by estimated action rates plus an ad quality term.

Read that as a pricing rule rather than a piece of platform trivia: relevance is a discount on media. Two advertisers in the same auction with the same budget will pay different prices for the same customer, and the gap is set by how well the creative, the audience and the landing page agree with each other.

Sub-concept 3: CAC, ROAS and the attribution window

Three numbers carry every paid conversation, and two of them are routinely misquoted.

CAC (customer acquisition cost) is everything you spent to acquire customers in a period divided by the number of new customers acquired. Paid CAC counts media, creative production, tooling and agency fees. Blended CAC divides total sales and marketing cost by all new customers, including organic ones. Executives quote blended CAC and then use it to judge a paid channel, which flatters the channel every time organic is strong.

ROAS (return on ad spend) is revenue attributed to ads divided by ad spend. It is measured on revenue, not margin, which is why a 4:1 ROAS can be a losing trade: at 25% gross margin, four dollars of revenue per dollar spent breaks even before you have paid anyone a salary.

The attribution window is the period after a click or a view during which a conversion still gets credited to that ad. Meta's default is 7-day click and 1-day view. Google Ads commonly defaults to a 30-day click window. These are settings, not facts about buyer behaviour. Widen the window and your ROAS improves without a single additional sale, which is the most common way a paid programme appears to be improving while the business does not.

Last-click attribution, which hands 100% of the credit to the final touch before purchase, makes branded search look like genius and every upper-funnel channel look useless. B2B buyers touch a vendor many times across months before anyone fills in a form. If your model counts only the last of those touches, you will defund the channels that create demand, overfund the ones that harvest it, and then wonder in month six why the harvest is thinner.

Google Ads Tutorial 2024 - Google Ads for Beginners

Watch on YouTube

Sub-concept 4: targeting precision, and what happened to it

Audience specificity is what separates this from buying a billboard. LinkedIn targets by job title, seniority, company size, industry and named account lists. Meta targets by interest, behaviour, life events and lookalikes built from a customer file you upload. Google targets by keyword intent, in-market segments and customer match. TikTok offers demographic and interest targeting, but its delivery leans hardest on what the creative itself attracts.

The precision has eroded since Apple's App Tracking Transparency arrived with iOS 14.5 in April 2021 and cut the identifier that fed cross-app measurement. Every platform responded by pushing advertisers toward broader audiences and letting the model find the buyer, with modelled conversions filling the gaps in reporting. Manual audience slicing matters less than it did in 2019. Creative volume and the quality of the conversion signal you send back matter more.

Real-world cases

Case 1: Meta after ATT. In February 2022 Meta told investors it expected roughly $10 billion of revenue headwind that year from Apple's changes, and its stock fell sharply on the news. Meta also shortened its default attribution windows. Advertisers who had never changed the setting saw reported ROAS drop while actual sales held steady. The lesson is not about Apple. It is that the number on your dashboard is produced by a measurement configuration owned by the seller.

Case 2: Google narrowing the attribution menu. In 2023 Google Ads retired first-click, linear, time-decay and position-based attribution, leaving data-driven and last-click. Google has also repeatedly delayed, then stepped back from, removing third-party cookies in Chrome. Both moves changed what advertisers could measure without any advertiser being consulted. Build your CAC reporting so it survives the platform changing its mind.

Case 3: TikTok as a third auction. TikTok for Business opened its self-serve auction to advertisers in 2020 and pulled meaningful budget out of Meta and YouTube within two years, mostly because cheap CPMs met creative that did not look like advertising. Then in April 2024 the US passed a divest-or-ban law aimed at the app. A channel can be your best CAC line and a single-point-of-failure at the same time.

Meta Ads Tutorial for Beginners 2024

Watch on YouTube

CMO action items

  • Write down the attribution window in force on every paid platform you run, then rerun the last six months under one consistent window before your next budget decision. If nobody on the team knows what the settings are, your channel comparison is arithmetic on incompatible units.
  • Label every dollar of paid budget as demand creation, demand capture or retargeting. If your team cannot produce that split in an afternoon, the problem is not reporting, it is that nobody has decided what each channel is for.
  • Set a minimum evidence bar per channel before anyone is allowed to declare a verdict: at least three or four weeks and enough conversions to be reading a result rather than noise. Channels killed after two weeks and $500 were never tested.
  • Report paid CAC and blended CAC side by side to the board, with gross margin next to any ROAS figure. Present one without the other and you will be asked to scale something that loses money per unit.

Common mistakes that kill results

Mistake 1: sending paid traffic to the homepage. A homepage serves several audiences and several objectives. A paid click is one person with one problem, arriving with an expectation the ad just created. Dedicated landing pages that repeat the ad's promise, with one offer and one call to action, convert materially better than homepage traffic, and they also raise landing page experience scores, so you pay less per click for the same position.

Mistake 2: confusing activity with performance. Impressions, clicks and CTR tell you the ad is running. Cost per qualified opportunity, paid CAC and payback period tell you whether it is working. If the monthly deck opens with impressions and buries cost per opportunity on slide eleven, the team is optimising for the first number.

Mistake 3: treating platform-reported conversions as company truth. Every platform counts conversions it believes it caused, including modelled ones, and the sum of the platforms will exceed the orders in your own system. Reconcile against your CRM or finance data monthly and manage to that gap rather than pretending it is not there.

Mistake 4: setting the budget once a year. Auction prices move with competitor entry, seasonality and platform changes. A plan built on January CPCs is structurally wrong by Q3. Tie reallocation to a quarterly review of actual CAC and payback, not to spend against plan.

Resources

  • 🔗
    Google Ads Help: About Quality Score

    Google's official documentation explaining exactly how Quality Score is calculated and how each component affects your auction performance and actual CPC.

  • 🔗
    Unbounce Conversion Benchmark Report

    Industry-wide conversion rate data by sector and channel type, useful for benchmarking whether your paid landing page performance is competitive or lagging.

What to do, from this lesson

These actions are compiled in the role's Playbook.

  • Send every paid campaign to a dedicated single-offer landing page
See the full action playbook →