Google Ads KPIs that actually matter (and the ones agencies hype)

The metrics an agency leads with reveal what they think their job is. If the first page of the monthly report is impressions and CTR, they think their job is "look busy." If it's revenue and CAC, they think their job is to make the business healthier. This piece is the metric stack you should hold every PPC vendor to, and the ones you should push back on. It is also a sanity check for in-house teams. For the broader picture see our complete buyer's guide.

The three tiers of metrics

Every PPC metric falls into one of three tiers. The mistake is treating them as equal.

Tier 1: outcome metrics. What the business actually cares about. Revenue (for e-commerce), pipeline-qualified leads and pipeline value (for B2B), customer acquisition cost (everywhere). These are the only metrics your CEO will ever care about.

Tier 2: efficiency metrics. How well the channel is converting spend into outcomes. ROAS, CPA, MER (marketing efficiency ratio), payback period. These are the operating metrics for the channel.

Tier 3: diagnostic metrics. Why the efficiency metrics moved. Conversion rate by campaign, impression share lost to budget, impression share lost to rank, click share, auction insights, search lost to ad rank. These help you debug.

The two tiers that don't make the list and shouldn't be primary KPIs: activity metrics (impressions, clicks, CTR) and vanity composites (engagement, attention scores, assisted conversion totals reported as if they were primary). They have diagnostic value; they have no business value.

The KPIs that actually matter (by business type)

E-commerce

1. Revenue from paid search. The number that shows up in your P&L. Track new-customer revenue separately from returning-customer revenue.
2. CAC (paid search). Total paid search spend divided by new customers acquired. Trend it monthly. Compare to LTV.
3. ROAS (with margin context). Revenue divided by ad spend. Useful only if you set a target against your gross margin. A 3× ROAS at 40% gross margin is unprofitable.
4. New customer ROAS. Revenue from new customers only divided by ad spend. The honest number; total ROAS over-counts returning buyers who would have come back anyway.
5. Repeat purchase rate from paid-acquired customers. Tells you whether paid is bringing in customers who stick or one-time buyers. Cohort by month.

B2B SaaS and lead generation

1. Pipeline-qualified leads (PQL or SQL) from paid search. Not MQL, not raw form fills. The stage in your CRM that genuinely predicts revenue. Push this back into Google Ads via offline conversion import.
2. Pipeline value (€). Sum of opportunity value created from paid search leads. Cohort by lead month.
3. CAC per closed deal. Paid spend divided by closed deals attributable to paid. Trail by sales cycle length.
4. CPL by stage. Cost per raw lead, cost per MQL, cost per SQL, cost per closed deal. The funnel from spend to revenue.
5. Lead-to-SQL conversion rate by source. Tells you whether paid leads are actually as good as they look at the top.

The diagnostic metrics worth tracking

When your efficiency metrics move and you need to know why, these are the ones to pull:

Impression share lost to budget. If high, you are leaving qualified impressions on the table. Either increase budget or tighten targeting.

Impression share lost to rank. If high, your ad rank is too low — bids, Quality Score, or ad relevance. Diagnose before throwing more budget.

Search lost top IS (top impression share). Tells you whether you are competitive in the top positions specifically. Below-the-fold positions have much lower CTR.

Conversion rate by campaign and ad group. Wide variance here is a sign of structural mismatch — landing pages, intent, audience.

Search term performance. The actual queries triggering your ads, with CPA, conversion rate, and search volume. The single most actionable diagnostic in Google Ads.

Auction insights. Who you compete with, when, and how often. Movement here is a leading indicator of competitive pressure.

Quality Score components. Expected CTR, ad relevance, landing page experience. Quality Score itself is too aggregated to act on; the components are diagnostic.

Asset performance (PMax). Which assets are performing in which channel slot. Sparse data but useful where you have it.

The vanity metrics agencies put on the first page

Impressions. A 40% increase in impressions tells you nothing about whether it was good impressions. If your CPA went up and ROAS went down, more impressions made things worse. Useful only as a diagnostic alongside conversion metrics.

CTR. A higher CTR is sometimes better (relevant audience), sometimes worse (broader, lower-intent audience). On its own, meaningless. In combination with conversion rate and CPA, useful.

Average position. Google removed average position as a metric for good reason. Some agencies still find a way to report on a proxy ("top impression share above 80%"). Position is not an outcome.

Engagement. Whatever the agency defines as engagement. Usually rolled-up CTR, time on site, multi-page sessions. Rarely correlates with revenue. Ask: "What is the definition, and how does it map to conversions?"

Assisted conversions. Useful as a diagnostic for understanding the conversion path. Misleading as a primary metric because agencies double-count: paid search "assisted" a conversion that Google organic actually closed, and the agency reports both as paid impact. Treat with suspicion.

Brand campaign ROAS. Almost always inflated. Customers searching your brand name were going to convert; ROAS on brand keywords is a measurement artifact, not channel performance. Strip out brand spend before reporting "paid search ROAS."

"View-through conversions" reported as primary. Some users who saw an ad and converted later would have converted anyway. Useful as a context metric, dangerous as a primary one.

How to set targets that aren't arbitrary

The most common KPI mistake we see: the agency sets a CPA or ROAS target that has no relationship to the unit economics of the business.

For e-commerce, your target ROAS should be set as:

Target ROAS = 1 / (Gross Margin × Allowable CAC ratio)

If your gross margin is 60% and you want CAC to be at most 30% of revenue (a reasonable benchmark for a growth-stage DTC), your target ROAS is 1 / (0.6 × 0.3) = 5.56×. A 4× ROAS is unprofitable for this business; an 8× ROAS is over-throttling.

For B2B, your target CAC should be:

Target CAC = LTV / 3 (rough rule of thumb)

If your average customer LTV is €18,000, target CAC is €6,000. Your CPL target then derives from CAC times the lead-to-close rate. If 5% of SQLs close, your target cost per SQL is €300. Working backwards from business reality, not setting CPA based on what the agency thinks looks good.

What a good monthly report looks like

The structure that tells you the agency is thinking about outcomes:

Page 1: outcomes vs target. Revenue or pipeline this month vs goal. New customer count vs goal. CAC vs target. Two sentences of narrative.

Page 2: efficiency. ROAS, CPA, MER, payback. Trend vs previous 6 months. Two paragraphs explaining what moved and why.

Page 3: where the budget went. Top campaigns by spend, conversion, ROAS. What was added, what was paused, what was tested.

Page 4: diagnostics. Impression share, search term shifts, auction insights. Charts.

Page 5: next month plan. Two or three experiments with hypothesis, budget, success criteria.

That's it. Five pages. Most reports we see are 12+ pages of charts and three sentences of opinion. The ratio should be inverted.

What to push back on in your next agency meeting

If your report leads with impressions, ask: "Can we move impressions to page 4 and lead with revenue and CAC?" If they push back, the conversation is about who the report is for.

If your ROAS target is the same across new-customer and returning-customer campaigns, ask: "Can we report new-customer ROAS separately, with a different target?" If they push back, ask why.

If the agency reports brand ROAS as a primary number, ask: "Can we strip brand from the headline ROAS and call it out separately?" If they push back, you have learned something.

If the agency uses "engagement" as a KPI, ask: "What is the definition and how does it map to a conversion in our funnel?" If the answer is vague, the metric is.

How AI-managed PPC changes the reporting shape

The structural difference an AI-managed service offers on reporting is that the dashboard is live and the audit trail of changes is searchable. The "monthly report" becomes a 30-minute strategic conversation about what the dashboard shows, rather than a deck-driven walkthrough. Both models can produce the same KPI hierarchy; the AI-managed shape removes the lag between something happening and you knowing about it. We cover this in AI-managed PPC vs Google Ads agency.

The honest summary

Most PPC underperformance is misdiagnosed because the metrics on the dashboard don't connect to the metrics in the business. Once you fix the metric hierarchy — outcomes first, efficiency second, diagnostics third, activity nowhere — most of the right decisions become obvious. The agency that resists this hierarchy is telling you something about how they want to be judged.

Where Logitelia fits

Logitelia's Growth Team reports on the structure above by default: outcomes first, efficiency second, diagnostics on demand. The live client portal shows the same hierarchy in real time. If you want to see what your account would look like reported this way, book a call and we'll walk through a sample dashboard.

What you measure shapes what you get. Measure the things that matter, push back on the things that don't, and the channel will tell you the truth about itself.