What every performance channel has in common
Strip away the interface and every performance channel is the same machine. A budget buys entries into an auction or a placement. Entries become impressions. A share of impressions become clicks, priced per click (CPCAverage cost of each click.) or folded into a price per thousand impressions. A share of clicks become the result the campaign was built for, and that conversion rate (CVRShare of visits or clicks that convert.) is mostly a property of the landing page and the offer rather than the channel. Each result is worth something to the business, and the gap between what it is worth and what it cost is the only reason to spend.
Because the machine is the same, the discipline is the same. Cost per acquisition (CPACost of each conversion.) and return on ad spend (ROASRevenue generated per unit of ad spend.) summarise the chain on every platform, and the business's blended acquisition cost (CACCost to acquire one new customer.) is what all the channel figures must add up to. The KPI tree framework is the general version of this chain; this hub applies it to paid acquisition. What differs between platforms is where the levers sit: which stage of the chain a platform lets you influence directly and which it decides for you with automation. Google's and Meta's help centres document their own hierarchies, bidding options and reporting, and the platform roots on this site cover them; the fundamentals here are what you need to know before reading any of them.
Economics, structure, tracking
The three foundations answer three different questions, and a programme that skips any of them fails in a predictable way.
| Foundation | The question | What goes wrong without it |
|---|---|---|
| Economics | What may a result cost, and how is that derived from margin, lifetime value and payback? | Targets copied from platform suggestions or last year's average; profitable growth cut because it looked expensive |
| Structure | How are campaigns split so that each budget, target and report maps to a real decision? | Reports that cannot say which product, market or stage is profitable; automated bidding starved of data by over-splitting |
| Tracking | Is every result recorded once, correctly, with the source it came from? | Campaigns judged on clicks; bidding optimising to the wrong event; discrepancies nobody can explain |
Economics and structure have their own core guides below, and the conversion events and values each platform optimises towards have a further guide. Tracking setup is shared with every other discipline on the site and is covered in the analytics root; the section on tracking foundations later on this page explains what a performance marketer must confirm before spend starts.
The core guides
Performance economics: CPA, ROAS and payback walks the chain from spend to margin, shows how to derive CPA and ROAS targets from unit economics rather than from what a platform proposes, separates the blended view from the channel and campaign views that platforms report, explains why average returns hide the marginal return that decides the next pound, and covers the payback and cash constraints that stop a correct target from being affordable. Its worked figures sit inside a labelled example.
Campaign structure across channels sets out the principle that structure follows decisions, not platforms: split by objective, stage, product or market only where a budget or target has to differ, and consolidate everywhere else so that automated bidding has enough results to learn from. It covers naming conventions and tracking parameters, budgets and targets per structural unit, and ends with a table of what genuinely differs by platform, routing to the Google Ads, Meta Ads and programmatic structure guides for the rest.
Read economics first. The granularity of the targets decides how finely campaigns need to be split, so the structure guide assumes the targets exist.
Further guides
Further guides cover what the bidding is told to value and what sets the price it pays:
- Conversion events and values across ad platforms — choosing the events and values each platform optimises towards: one definition of a result, proxies, deduplication, CRM outcomes and consent gaps.
- How ad auctions set the price of performance media — how search, social and programmatic auctions decide what you pay: bids, predicted response, quality, competition, first-price buying and what you control.
Tracking foundations before spend
Automated bidding on every major platform optimises towards the conversion events it is given. If the event is wrong, late, duplicated or missing, the platform optimises towards the wrong thing with complete confidence, and the reports will look healthy while the business does not. The tracking plan guide in the analytics root covers how to decide what to measure and how to document it; the performance marketer's job is to confirm a shorter list before the first campaign goes live. The conversion events guide sets out the cross-platform method behind that list: one written definition of each result, values that mean the same everywhere, deduplication, outcomes fed back from the CRM and the gaps consent leaves.
- One event per real result. A purchase, a qualified lead, an install. Not a page view of the thank-you page that fires on refresh, and not a form submission that the sales team later discards.
- The same event on every platform. Each platform's conversion should mean the same thing so that CPA figures can be compared. Where a platform receives a different event, the report must say so.
- A value where the result has one. ROAS bidding and reporting need the order value passed through; lead generation needs at least a proxy value by lead type.
- Deduplication where web and server-side events both fire. Double-counting flatters every campaign.
- A test conversion recorded and found in the platform before launch, not after the first week's spend has been judged on nothing.
The core metrics guide explains what each tracked metric means; this hub's concern is that the metrics exist and can be trusted.
A language school discovers its tracking before its targets
Brightwater Languages, an online tutoring service, launches search and paid social campaigns with a target cost per trial lesson of £40. After a month both platforms report trials at roughly £30 and the team plans to double budget. Before doing so the operations lead reconciles the platforms with the booking system and finds 410 platform-reported trials against 260 actual bookings: the booking confirmation page was firing on every reload, and both platforms had been given the same page view as their conversion. Corrected, the real cost per trial is £47 on search and £58 on social. Budget is held flat, the event is rebuilt on the server side with a booking identifier for deduplication, and the targets are re-derived once a clean month exists.
The performance vocabulary
Each of the terms below has its own glossary page with the definition, formula and direction. This hub uses them without redefining them.
| Term | What it summarises | Glossary |
|---|---|---|
| Cost per click | The price of one visit, set by the auction | CPC |
| Conversion rate | The share of visits that produce the result | CVR |
| Cost per acquisition | What one result cost, the lead-generation target | CPA |
| Return on ad spend | Revenue per unit of spend, the e-commerce target | ROAS |
| Customer acquisition cost | The blended cost of one new customer across all spend | CAC |
The CPA calculator and ROAS calculator turn a campaign's rates into these figures, and the marketing funnel framework shows where each sits in the journey. The performance marketing hub places this vocabulary in the wider system.
Common performance fundamentals problems
| Problem | Usual cause | Guide |
|---|---|---|
| Target CPA set from a platform's suggested bid | No chain from margin to target | Performance economics |
| Profitable campaign cut because its ROAS looked low | Average return read where the marginal return mattered; payback ignored | Performance economics |
| Channel CPAs add up to less than the blended CAC | Platform attribution counted as truth; blended view never built | Performance economics |
| Reports cannot separate markets or products | Structure split by platform habit, not by decision | Campaign structure across channels |
| Automated bidding never leaves learning | Over-split structure starving each campaign of results | Campaign structure across channels |
| Platform and analytics disagree by a wide margin | Different events, duplicates or missing parameters | Tracking plan, Conversion events across platforms |
| Bidding buys a result the business does not value | The tag set-up chose the event; a proxy optimised for and judged on itself | Conversion events across platforms |
| Click or impression prices rise and nobody knows why | A dearer mix of queries, people or placements; seasonal competition; bids raised to fix a creative problem | How ad auctions price media |
