Reading the metrics that matter: CTR, CPC, CPA, ROAS, and Impression Share
Every part of this series has referenced one or more of these metrics individually. This part pulls them together — because reading any single one in isolation, without the others for context, routinely leads to the wrong conclusion.
The five core metrics, defined precisely
- CTR (Click-Through Rate) — clicks divided by impressions. High CTR means the ad is compelling relative to how often it's shown; it says nothing on its own about whether those clicks convert.
- CPC (Cost Per Click) — the actual average amount paid per click. Directly shaped by Quality Score (part 9) and competition for that keyword.
- CPA (Cost Per Acquisition) — total spend divided by number of conversions. The real cost to generate one subscription or lead.
- ROAS (Return On Ad Spend) — revenue generated divided by ad spend, usually shown as a percentage or ratio. Requires conversion value tracking (part 11) to calculate at all.
- Impression Share — covered fully in part 18: the percentage of eligible auctions an ad actually appeared in.
A real comparison: two ad groups, one misleading metric
Ad group "Coffee Subscription — General"
CTR: 8.2% CPC: $1.90 Conversions: 12 CPA: $31.67
Ad group "Coffee Subscription — Gift"
CTR: 8.4% CPC: $2.10 Conversions: 3 CPA: $147.00Both ad groups have nearly identical, genuinely strong CTR — reading CTR alone, they'd look equally healthy. CPA tells a completely different story: the Gift ad group is costing nearly five times as much per actual conversion. This is a real, common pattern: a compelling ad (high CTR) sending traffic to a landing page or targeting a keyword set that doesn't convert well is entirely possible, and CTR alone would never reveal it — this is exactly why part 8's landing page guidance and part 9's Quality Score components matter as much as the ad copy itself.
CPA vs. ROAS: which one actually reflects the business goal
Product A: $18 subscription, CPA $12 → profitable, but thin margin per sale
Product B: $180 subscription, CPA $35 → far more profitable in absolute termsA flat CPA target treats every conversion as equally valuable — clearly wrong here, since Product B is dramatically more valuable despite a higher acquisition cost. ROAS (or Target ROAS bidding from part 10) accounts for this directly, which is exactly why a business with meaningfully different-value conversions should optimize toward ROAS rather than a flat CPA target that implicitly treats a $18 and a $180 sale as equally good outcomes.
Impression Share as a growth diagnostic, not just a budget one
Beyond part 18's budget-constrained reading, a low Impression Share on a strong-converting campaign is a direct, quantified answer to "how much more could this campaign realistically grow" — a campaign at 40% Impression Share has real, measurable headroom before it's anywhere close to fully capturing its available auction volume, a genuinely different growth ceiling than a campaign already at 85%.
Reading metrics together, not separately
High CTR + high CPA → compelling ad, weak conversion (landing page or targeting issue)
Low CTR + low CPA → ad isn't earning clicks, but the clicks it gets convert well
High CPC + high Quality Score → still expensive, likely a genuinely competitive keyword
Low Impression Share + strong ROAS → real, quantified growth opportunityEach pairing above tells a different, specific story that neither metric alone would reveal — this is the actual skill of reading a Google Ads account: not memorizing what one number "should" be, but recognizing what a specific combination of numbers together is actually saying about where the real problem, or the real opportunity, sits.
Optimizing an account purely toward the lowest possible CPC, treating it as an end in itself. A cheap click that never converts is a worse outcome than a more expensive one that reliably does — CPC only matters in the context of what it eventually costs to generate an actual conversion, which is what CPA and ROAS actually measure.
Next: A/B testing — validating a specific change (an ad, a landing page) with real data instead of assuming it's an improvement.