Metric glossary

Total Spend

Total amount spent on ads in the selected period.

What to do: Spend on its own is never "bad" — read it next to CPA and ROAS. If spend grew and conversions did not, open Reports sorted by spend and find the campaign that absorbed the difference.

Impr.

Number of times your ads were shown.

What to do: Few impressions on an active budget means narrow targeting, low bids, or a budget cap — check Search Impression Share and the two "lost share" metrics to see which of the three it is.

Reach

Unique users who saw your ad at least once (Meta only).

What to do: Reach that stops growing while budget grows means the audience is exhausted — extra budget now buys repeat impressions to the same people, not new ones.

Frequency

Average number of times each person saw your ad. Formula: Impressions ÷ Reach.

What to do: Frequency above roughly 3–4 in a short window is creative burnout: refresh the creative or widen the audience. Ignoring it drives CPM and CPA up on its own.

Clicks

Total number of clicks on your ads.

What to do: Clicks falling while impressions hold steady means CTR dropped: the ad no longer matches what people search for. Refresh headlines and creative before touching bids.

CTR

Click-Through Rate — percentage of impressions that resulted in a click. Formula: Clicks ÷ Impressions.

What to do: Low CTR is a relevance problem. Rewrite headlines around the actual search query, add the missing assets/extensions, and move irrelevant queries to negatives.

CPC

Cost Per Click — average spend per click. Formula: Spend ÷ Clicks.

What to do: Rising CPC is either competition or falling quality. Check Quality Score first — improving relevance lowers CPC without giving up position, while cutting bids gives up volume.

CPM

Cost Per 1,000 Impressions — how much you pay for 1000 ad views. Formula: Spend ÷ Impressions × 1000.

What to do: High CPM usually means a narrow audience or expensive placements. Widen targeting, check Frequency for creative burnout, and exclude placements that deliver impressions but no clicks.

Conv.

Total conversions — purchases, leads, sign-ups, or other goal completions tracked by the platform.

What to do: A sudden drop in conversions is a tracking problem until proven otherwise. Open Tracking Health before you change bids — a broken tag looks exactly like a collapse in performance.

All conv.

All conversions — every conversion action Google tracks for the account, both the primary ones counted in “Conv.” and the secondary ones that are not. A separate Google Ads metric (metrics.all_conversions), shown alongside Conv., never added to it. Only Google Ads reports it, so a row that also covers another platform shows “—” rather than a Google-only figure next to an all-platform Conv. — select the Google Ads channel to see it.

What to do: A large gap between All conversions and Conv. means most of your tracked actions are secondary goals. That is fine — but check which actions Google counts as primary before you judge CPA, because CPA is calculated from Conv. only.

CVR

Conversion Rate — percentage of clicks that resulted in a conversion. Formula: Conversions ÷ Clicks.

What to do: Low CVR with a healthy CTR points at the site, not the ads: the landing page is slow, or it does not deliver what the ad promised. Fix the page before raising budgets.

CPA

Cost Per Action — average spend per conversion. Formula: Spend ÷ Conversions. Also called CPL for lead gen.

What to do: Above target CPA: in Ads Reports find the search terms, placements and audiences with spend and zero conversions and cut those first. Lowering bids everywhere costs you the volume that was working.

Revenue

Total revenue attributed to ad-driven conversions by the ad platform.

What to do: Revenue lagging while conversions hold means the average order value fell — or part of the purchases stopped reaching tracking. Check AOV first, then Tracking Health.

AOV

Average Order Value — average revenue per conversion. Formula: Revenue ÷ Conversions.

What to do: A falling AOV usually means traffic drifted to cheap SKUs or a promo. Check which products the spend goes to and exclude low-margin categories from the campaigns that must stay profitable.

ROAS

Return on Ad Spend — revenue earned per dollar spent. Formula: Revenue ÷ Spend. E.g. 3.5 means $3.50 earned per $1 spent.

What to do: Below break-even ROAS: move budget from the worst campaigns to the best (Opportunities → budget planner) instead of cutting everywhere. Know your break-even point — it is margin-dependent, not 1.0.

SIS

Search Impression Share — percentage of eligible search impressions your ads actually received (Google Search & Shopping only).

What to do: Search Impression Share under ~60% means you are absent from most relevant auctions. The next two metrics say why: budget or ad rank — the fix is different for each.

Lost IS (budget)

Share of eligible impressions you did not get because the campaign ran out of budget.

What to do: On a campaign that already hits its target CPA or ROAS this is the cheapest growth available — raise the daily budget. On an unprofitable one, fix efficiency first.

Lost IS (rank)

Share of eligible impressions lost because your ad rank (bid × quality) was too low.

What to do: Rank losses are cured by quality as much as by bids: raise ad and landing-page relevance, then the bid. Bidding alone buys the same position at a worse CPC.

Add to Cart

Number of Add-to-Cart events attributed to your Meta ads.

What to do: Carts filling but purchases not following is a checkout problem, not an ads problem: shipping cost shown late, forced registration, or a payment step that fails on mobile.

CVR ATC

Add-to-Cart Conversion Rate — percentage of clicks that resulted in an add-to-cart. Formula: Add to Cart ÷ Clicks.

What to do: A low add-to-cart rate on paid traffic means the product page loses people: price, photos, or availability. Compare it with your organic traffic rate before blaming the campaign.

Budget

Daily budget set in the ad platform — shown at the campaign level for campaign budgets (CBO) and at the ad-set level for ad-set budgets (ABO).

What to do: A campaign flagged "limited by budget" at target CPA should get more budget. Above target CPA, clean up the wasted spend first — a bigger budget only scales the leak.

Quality Score

Google's 1–10 rating of a keyword: expected CTR, ad relevance, and landing page experience.

What to do: A QS of 4 or below with real spend behind it: move the keyword into its own ad group and write the ad around that exact query. That is cheaper than paying the rank penalty with bids.

Sessions LC

GA4 sessions attributed via last-click model. Shows actual site visits from your ads according to Google Analytics.

What to do: A gap versus platform clicks is normal (different attribution windows). Alarming is near-zero GA4 sessions against real clicks — that means broken UTMs or auto-tagging, so check Tracking Health.

Conv. LC

GA4 conversions attributed via last-click model: purchases for e-commerce accounts, lead events (generate_lead, form_submit, sign_up, …) for leadgen/SaaS, all key events otherwise. Often differs from platform-reported conversions due to different attribution windows.

What to do: GA4 will almost always report fewer conversions than the ad platform — it credits the last click, the platform credits its own view/click windows. Use one of them per decision, not both at once.

Revenue LC

GA4 revenue attributed via last-click model. A more conservative revenue estimate compared to the platform's own attribution.

What to do: If GA4 revenue is far below the platform figure, verify the purchase event actually sends a value and currency. Zero-value purchases silently destroy every ROAS number downstream.

ROAS LC

ROAS calculated from GA4 last-click revenue. More conservative than platform ROAS. Formula: Revenue LC ÷ Spend.

What to do: Treat this as the conservative floor of your ROAS and the platform number as the ceiling. If a decision flips depending on which one you use, the campaign is too close to break-even to scale.

Sessions

Visits to the site: one session covers everything a person does until 30 minutes of inactivity.

What to do: Sessions dropping across every channel at once is usually a measurement break (tag removed on deploy, consent banner change), not a market change. Verify before reacting.

Users

Distinct people who visited, as counted by GA4 (browser/device based, so cross-device visitors can count twice).

What to do: Users far below sessions means people come back often — good for retention, but do not read that repeat traffic as new demand when planning budgets.

Key events

GA4 events you marked as important (form submit, sign-up, purchase). The GA4 replacement for the old "goals".

What to do: If everything is a key event, nothing is: mark only the events with business value, otherwise every conversion-based report and bidding strategy inherits the noise.

Purchases

Completed e-commerce purchase events recorded by GA4.

What to do: Purchases in GA4 below what the CRM or store admin shows means the event misses on some paths (guest checkout, an alternate payment flow). Reconcile the two before trusting any revenue report.

Bounce rate

Share of sessions with no engagement — under 10 seconds, no key event, no second page.

What to do: A high bounce rate on paid traffic points at the ad-to-page mismatch or load speed. Compare paid against organic on the same page: if only paid bounces, the promise in the ad is wrong.

Pages / session

Average number of pages viewed per visit.

What to do: Read this together with conversions: many pages and few conversions is usually people searching for something they cannot find, not deep engagement.

Engagement time

Average time the site was actually in the foreground and being used per session.

What to do: Under ~15 seconds on a landing page means the page is not answering the ad's promise in the first screen. Fix the first screen before rewriting the whole page.

Bot sessions

Sessions identified as automated traffic rather than people.

What to do: A visible share of bot traffic inflates sessions and deflates conversion rate. Filter it out before you judge a campaign by CVR, or you will optimise against noise.

Rage clicks

Repeated fast clicks on the same spot — the signature of a visitor expecting something to respond and getting nothing.

What to do: Open the page with the most rage clicks and find the element people are hitting: a broken button, a non-clickable image that looks clickable, or a form that fails silently. This is the fastest CVR win on any page.

Dead clicks

Clicks on elements that do nothing at all.

What to do: Usually styling that mimics a link or button. Either make the element work or stop it looking interactive — every dead click is a visitor who thought they were moving forward.

Quick back

Visitors who opened the page and returned to the previous one within seconds.

What to do: On a paid landing page this is the clearest sign of a mismatched promise: the ad said one thing and the page shows another. Align the headline with the ad, not with the brand pitch.

Script errors

JavaScript errors thrown in the visitor's browser on this page.

What to do: Errors on a checkout or form page can block conversions for an entire browser or device class while every dashboard still looks normal. Treat them as the first suspect when CVR falls with no other change.

Excessive scroll

Sessions with unusually long scrolling — people hunting for something instead of reading.

What to do: Move whatever they are looking for (price, contact, specification) higher up the page rather than adding more content.

Scroll depth

How far down the page visitors get on average.

What to do: If most visitors never reach your call to action, the page is not too long — the CTA is too low. Repeat it above the fold.

AI visibility

Share of the tracked prompts where an AI engine mentioned your brand in its answer.

What to do: Low visibility is an answer-content problem, not a bidding one: AI engines quote pages that state facts plainly — pricing, comparisons, specifications. Start with the prompts where competitors are cited and you are not.

Position in AI answer

Where your brand appears inside the AI answer — first mention, later, or only in the sources.

What to do: Being cited late is closer to being invisible than to ranking first. Chase the prompts where you are already mentioned but low — those move with much less work than prompts where you are absent.

Citations

Links the AI engine listed as sources for its answer.

What to do: Look at which of your pages get cited and which competitor pages beat them. Citation goes to the page that answers the question directly, not to the homepage.

AI sessions

GA4 sessions that arrived from AI platforms (ChatGPT, Perplexity, Gemini and others).

What to do: This traffic is earned, not bought — so judge it by conversion quality, not volume. If it converts better than paid, that is the argument for investing in AI visibility.

Organic position

Your page's position in Google's organic results for a tracked keyword.

What to do: Ranking in the top 10 while AI answers ignore you is a specific, fixable gap: the page has authority but does not state the answer in a quotable form. Add a direct, factual answer near the top.