Every ad term you'll hit, in plain English.
No jargon walls, no assumed knowledge. 48 terms an owner or marketer actually runs into — what each one means, and why it matters for your money.
The numbers you'll see
CPC — Cost Per Click
What you pay each time someone clicks your ad.
Why it matters · A low CPC is nice, but cheap clicks that never convert cost you more than expensive clicks that do. Judge it alongside conversion rate, not on its own.
CPM — Cost Per Mille
The cost to show your ad 1,000 times (mille = thousand), whether or not anyone clicks.
Why it matters · Mostly a Meta/awareness metric. Rising CPM can mean more competition for your audience or fatigued creative pushing costs up.
CTR — Click-Through Rate
The percentage of people who clicked after seeing your ad (clicks ÷ impressions).
Why it matters · A quick read on whether your ad and audience match. Falling CTR over time is often the first sign of creative fatigue.
CPA — Cost Per Acquisition
What it costs, on average, to get one conversion — a sale, a lead, a signup.
Why it matters · The number that usually matters most. If CPA jumps, something upstream broke — Zephra flags that same-day rather than at month's end.
CPL — Cost Per Lead
What you pay for one lead (a form fill, a call, an enquiry).
Why it matters · Low CPL is only good if those leads become customers. A cheap lead that never closes is worse than an expensive one that does.
ROAS — Return On Ad Spend
Revenue earned for every dollar spent on ads. A 3x ROAS means $3 back for every $1 in.
Why it matters · The headline efficiency number for ecommerce. But it's only as honest as your conversion tracking — undercounted sales make ROAS look worse than reality.
Conversion Rate
The percentage of clicks (or visitors) that complete the action you want.
Why it matters · A healthy click-through with a poor conversion rate usually points at the landing page, not the ad — a common, fixable misdiagnosis.
AOV — Average Order Value
The average amount a customer spends per order.
Why it matters · Higher AOV means you can afford a higher CPA and still profit — it changes what 'good' performance looks like for your account.
LTV — Lifetime Value
The total revenue a customer brings over their whole relationship with you, not just the first purchase.
Why it matters · If customers come back, you can spend more to acquire them than a single-sale view suggests. LTV is what separates aggressive-but-smart from reckless spending.
Impressions
The number of times your ad was shown.
Why it matters · Raw reach. On its own it means little — impressions without clicks or conversions are just spend.
Google Ads
Quality Score
Google's 1–10 rating of your keyword, based on expected click-through rate, ad relevance, and landing page experience.
Why it matters · A higher score means cheaper clicks and better positions. The single number hides which of the three parts is the problem — Zephra breaks it down so you fix the right thing.
Ad Rank
How Google decides whose ad shows where — roughly your bid multiplied by your Quality Score (plus other factors).
Why it matters · It's why a competitor with a smaller budget can outrank you: better relevance can beat a bigger bid.
Impression Share
The percentage of available impressions your ads actually got, versus how many they were eligible for.
Why it matters · 'Lost impression share' is an abstract percentage until you turn it into money — Zephra converts it into the budget you'd need to stop losing traffic to competitors.
Negative Keywords
Search terms you tell Google to never show your ads for.
Why it matters · The main lever for cutting wasted spend — blocking 'free', 'jobs', or 'cheap' searches that click but never buy. The catch is not cutting real customers by mistake, which is why it's review-first.
Search Terms
The actual phrases people typed before your ad appeared (different from the keywords you targeted).
Why it matters · Where the waste hides — broad-match keywords trigger on searches you'd never choose. Reviewing them is how negative keywords get found.
Match Types (Broad / Phrase / Exact)
How loosely Google matches your keyword to a search. Broad casts widest; exact is tightest.
Why it matters · Broad match reaches more people but wastes more budget on irrelevant searches; exact is safer but narrower. The right mix depends on how much data you have.
Performance Max (PMax)
Google's fully-automated campaign type that runs across Search, Display, YouTube, Gmail and Maps from one setup.
Why it matters · Powerful but a black box — you hand over targeting and get little visibility. Zephra watches inside it for weak assets, stalled conversions and falling CTR that Google won't surface.
Smart Bidding
Google's automated bidding strategies (like Target CPA or Target ROAS) that set bids per auction using its own signals.
Why it matters · Great once you have enough conversion data — premature if you don't. And it quietly ignores manual bid adjustments, a common silent waste Zephra flags.
tCPA / tROAS
Target CPA and Target ROAS — Smart Bidding strategies where you set the cost-per-acquisition or return you want and Google bids toward it.
Why it matters · Only reliable when your conversion tracking is accurate. Set a target on bad data and Google optimises confidently toward the wrong thing.
Ad Strength
Google's Poor–to–Excellent rating of a responsive search ad, based on how many varied headlines and descriptions you've given it.
Why it matters · A 'Poor' rating quietly limits how often your ad shows. Zephra flags weak ads and why, instead of you checking one at a time.
Meta Ads
CBO / ABO
Campaign Budget Optimization (Meta sets budget across ad sets) versus Ad set Budget Optimization (you set it per ad set).
Why it matters · The wrong choice can starve your best ad sets of the data they need to learn. Getting the structure right is often a bigger win than any single ad tweak.
Advantage+
Meta's suite of AI-automated campaign and audience features (the Meta counterpart to Google's Performance Max).
Why it matters · Convenient and increasingly default, but opaque about why it's spending where it is — the same 'trust the black box' tradeoff, which is why oversight matters.
Delivery / Optimization Event
The specific action you tell Meta to optimise delivery toward — a click, a landing-page view, a purchase.
Why it matters · Meta only chases the goal you set. Ad sets left optimising for clicks long after there's purchase data is a quiet, common source of wasted spend.
Placements
Where your ads can appear — Feed, Reels, Stories, Marketplace, Audience Network, Messenger and more.
Why it matters · Some placements (Audience Network, Messenger inbox) can absorb real budget with zero conversions. Excluding the dead ones is easy money back.
Audience Network
Meta showing your ads on third-party apps and websites outside Facebook and Instagram.
Why it matters · Often the first placement to check when spend looks high and conversions don't — it's a frequent budget drain for lead-gen and considered purchases.
Lookalike Audience
A new audience Meta builds by finding people similar to a source list — your customers, leads, or site visitors.
Why it matters · Only as good as the source you feed it. A lookalike built from your best-converting customers beats one built from all traffic.
Custom Audience
An audience you define from your own data — a customer list, site visitors, or people who engaged with your content.
Why it matters · The foundation of retargeting and lookalikes. Accurate tracking is what keeps these audiences populated and useful.
Creative Fatigue
When an audience has seen an ad so many times that it stops working — click-through falls and cost rises.
Why it matters · The tax on winning creative. Catching it early (rising frequency, falling CTR) and refreshing before performance drops is far cheaper than reacting after.
Frequency
The average number of times each person has seen your ad.
Why it matters · Climbing frequency with falling CTR is the classic fatigue signal. It's the early-warning number Zephra watches so you refresh in time.
Event Match Quality (EMQ)
Meta's score for how well your Conversions API data matches real people in its system.
Why it matters · Low EMQ means Meta under-attributes your results even when your ads are working — so it optimises on an undercount. A quiet but serious drag on performance.
Tracking & attribution
Conversion
A tracked action that matters to your business — a purchase, a lead, a call, a signup.
Why it matters · Everything the algorithms optimise toward depends on conversions being counted accurately. Miss them and the whole system optimises on bad data.
Pixel
A snippet of code on your site that reports visitor actions back to Google or Meta from the browser.
Why it matters · The old standard — and increasingly blocked by ad-blockers and iOS privacy settings, which is why pixel-only tracking now undercounts real results.
Server-Side Tracking
Reporting conversions to the ad platforms from your server instead of only the visitor's browser.
Why it matters · Catches the conversions pixels miss — often 20–40% more. It's the difference between optimising on the truth and optimising on an undercount.
CAPI — Conversions API
Meta's server-side method for sending conversion events directly, bypassing browser blocking.
Why it matters · The Meta half of accurate tracking. Its data quality is measured by Event Match Quality — set it up poorly and you gain little.
Enhanced Conversions
Google's method of matching conversions using hashed (privacy-safe) customer data like email or phone.
Why it matters · Recovers conversions that a click ID alone would miss — common on longer paths to purchase across devices or delayed visits.
gclid
Google Click ID — a unique tag Google adds to a click so a later conversion can be traced back to the exact ad and keyword.
Why it matters · The thread that connects a sale to the ad that earned it. Lose it (or the click's privacy-stripped on iOS) and attribution breaks.
Attribution
Deciding which ad, click, or channel gets credit for a conversion.
Why it matters · Both Google and Meta tend to over-claim credit for the same sale. Comparing them honestly — on the same underlying data — is the only way to trust the numbers.
Attribution Window
How long after a click or view a conversion still counts — e.g. a 7-day click window.
Why it matters · Longer windows credit more conversions to ads; shorter ones are stricter. Comparing two platforms on different windows is comparing apples to oranges.
iOS ATT (App Tracking Transparency)
Apple's rule that apps must ask permission before tracking you across other apps and sites.
Why it matters · The reason a large share of iPhone conversions go uncounted by default — and the reason server-side tracking went from nice-to-have to necessary.
First-Party Data
Data you collect directly from your own customers and site — as opposed to data bought or borrowed from third parties.
Why it matters · As third-party tracking gets blocked, your own first-party data (customer lists, site behaviour) becomes the most reliable fuel for targeting and matching.
AI & automation
AI / Machine Learning
Software that learns patterns from data to make predictions or decisions, instead of following a fixed set of hand-written rules.
Why it matters · The difference from old 'automation': a rule fires the same way forever; a model adapts as your account's data changes. 'AI-powered' alone means little — ask what it actually does.
LLM — Large Language Model
The kind of AI behind chat assistants — trained on huge amounts of text to understand and generate language.
Why it matters · It's what lets you ask your ads a question in plain English. Its weakness is inventing facts, which is why grounding answers in your real data matters.
Agentic AI
AI that doesn't just answer — it can take actions toward a goal, like adjusting a budget or pausing an ad.
Why it matters · The emerging 2026 category term. The important question isn't whether it's 'agentic' but whether it acts with your approval or on its own.
MCP — Model Context Protocol
An open standard for safely connecting an AI system to real tools and data through a fixed set of permitted actions.
Why it matters · It's what lets Zephra AI actually execute a change, not just describe one — through specific, reviewed actions rather than open-ended access. See our Trust & Safety page for how it's kept safe.
Human-in-the-Loop
A design where a person approves or can override the AI's decisions, rather than the AI acting fully on its own.
Why it matters · Now a crowded claim — most tools say it. What actually differs is the specifics: what you approve, whether you see the reasoning, and whether you can undo it.
Black Box
A system that produces decisions without showing how or why it reached them.
Why it matters · The core risk of 'fully autonomous' ad tools: when performance drops, there's no reasoning to inspect. Explainability is the opposite — and what you should look for.
Guardrails
Hard limits and safety controls that constrain what an automated system is allowed to do.
Why it matters · Spend caps, circuit breakers, approval gates — the things that let you give AI autonomy on the safe stuff while keeping the risky stuff on a leash you hold.
Read vs. Write Access
The distinction between an AI being able to look at your account (read) versus change it (write).
Why it matters · A safe design keeps these separate: the AI can always read, but can only write through a small set of specific, reviewed actions — never arbitrary commands.
Now put the terms to work — on your actual ads.
Sign up free in 2 minutes — no card. Already running ads? Get an audit of where money's leaking. New to ads? Watch Zephra build your first campaign. You only pay when you go live.
No credit card · Free until you launch · Cancel anytime
Prefer to talk it through? Book a free 30-minute consultation →