Google Ads Performance Analysis: What Metrics Actually Matter

Open your Google Ads dashboard and you see dozens of metrics: clicks, impressions, CTR, CPC, conversions, conversion rate, cost per conversion, impression share, ROAS, and more. Most account managers fixate on CTR and CPC. Neither metric tells you whether Google Ads is actually making you money.

This guide focuses on the metrics that predict business outcomes: the numbers that tell you whether to increase spend, decrease spend, or restructure your campaigns. If you manage Google Ads for a business (or your own business), these are the numbers to watch.

Worked Example: Which Search Term Meets a Spend Rule?

Question: Which search term meets this page's example spend rule for review?

Inputs: Made-up data, seed 23. The script sets spend by hand for five sample terms. It draws conversion counts with a Poisson model, a way to draw count data. The target cost per action (CPA), which is spend divided by conversions, is $50.00. The script gives "trail boots" an average of exactly 0 conversions, so it can only draw zero. The flag on it is by construction. "Sale shoes" also drew 0 conversions, at $121.00 of spend, which is below the line.

Method: Set the review line at three times the target CPA. Flag a term for pause review only if spend is over that line and it has no conversions. The three-times-target rule is an example rule set for this page. It has no source and is not a standard.

import numpy as np

seed = 23
rng = np.random.default_rng(seed)
keywords = ["north shoes", "running shoe", "trail boots", "shoe care", "sale shoes"]
spend = np.array([92.0, 168.0, 155.0, 41.0, 121.0])
conversions = rng.poisson([2.0, 3.0, 0.0, 1.0, 0.0])
target_cpa = 50.0
threshold = 3 * target_cpa

print(f"Simulated data, seed {seed}")
print(f"Target CPA: ${target_cpa:.2f}; pause review threshold: "
      f"spend > 3 x target = ${threshold:.2f} with zero conversions")
print("Keyword | Spend | Conversions | CPA")
for keyword, cost, conv in zip(keywords, spend, conversions):
    cpa = "n/a" if conv == 0 else f"${cost / conv:.2f}"
    print(f"{keyword} | ${cost:.2f} | {conv} | {cpa}")
review = [keyword for keyword, cost, conv in zip(keywords, spend, conversions)
          if cost > threshold and conv == 0]
print("Review for pause: " + (", ".join(review) if review else "none"))
print(f"Total spend: ${spend.sum():.2f}; "
      f"total conversions: {conversions.sum()}; "
      f"overall CPA: ${spend.sum()/conversions.sum():.2f}")

Printed output (copied from a run with python3 -I):

Simulated data, seed 23
Target CPA: $50.00; pause review threshold: spend > 3 x target = $150.00 with zero conversions
Keyword | Spend | Conversions | CPA
north shoes | $92.00 | 2 | $46.00
running shoe | $168.00 | 3 | $56.00
trail boots | $155.00 | 0 | n/a
shoe care | $41.00 | 0 | n/a
sale shoes | $121.00 | 0 | n/a
Review for pause: trail boots
Total spend: $577.00; total conversions: 5; overall CPA: $115.40

Result: Trail boots is the only keyword flagged for pause review. It spent $155.00, above the $150.00 line, and had zero conversions. Review it before you decide whether to pause it.

Limit: These made-up values show one case. They do not show how a real account will do.

The Metrics That Actually Matter

1. Return on Ad Spend (ROAS)

What it is: Revenue generated / ad spend. A ROAS of 4.0 means every $1 in Google Ads generates $4 in revenue.

Why it matters most: ROAS is the closest metric to answering "is this ad spend making money?" It directly connects cost to revenue. A campaign with a 2% CTR and $8 CPC that generates 5.0 ROAS is vastly more valuable than a campaign with a 12% CTR and $1.50 CPC that generates 1.2 ROAS.

What a good target looks like: It depends on your margins. Break-even ROAS is 1 divided by your gross margin: a business with a 40% margin breaks even at 2.5:1, one with an 80% margin at 1.25:1, and one with a 20% margin at 5:1. Set your target from your own costs.

2. Cost Per Acquisition (CPA)

What it is: Total ad spend / number of conversions. If you spent $3,000 and got 40 conversions, your CPA is $75.

Why it matters: CPA tells you the cost of each new customer or lead. Unlike ROAS (which requires revenue tracking), CPA works even when you track leads rather than direct revenue, which is the case for most B2B and service businesses.

How to use it: Set a target CPA based on your unit economics. If your average customer is worth $500 in lifetime value and you need a 3:1 return, your maximum CPA is $167. Any campaign consistently above that threshold needs review.

Track CPA by campaign, by ad group, and by keyword. Some accounts show that a small share of keywords brings many conversions. Check your own data to see if this is true for you. The insight is which keywords fall where.

3. Impression Share

What it is: The percentage of total available impressions that your ads actually received. If your impression share is 65%, your ads showed for 65% of the searches that matched your keywords.

Why it matters: Impression share tells you how much room you have to grow. If your best-performing campaign (high ROAS, low CPA) has 40% impression share, you are leaving 60% of that high-quality traffic on the table. That is a clear signal to increase budget or bids on that campaign.

Two types to monitor:

• Search impression share lost (budget): Your ads did not show because your daily budget ran out. Fix: increase daily budget on high-ROAS campaigns.
• Search impression share lost (rank): Your ads did not show because your ad rank was too low (bid or quality issue). Fix: improve ad quality or increase bids.

A profitable campaign with high budget-lost impression share is the most actionable finding in Google Ads analysis. It means you have a proven campaign that could scale. You just need to give it more budget.

4. Conversion Rate by Landing Page

What it is: Conversions / clicks for each landing page. Different from account-level conversion rate because it isolates the page performance from the keyword and ad performance.

Why it matters: You can have the best keywords and ads in the world, but if the landing page does not convert, you are wasting money. A landing page converting at 2% vs. 5% means 2.5x more conversions from the same traffic, without any additional ad spend.

What to look for: Export your campaign data with landing page URLs. Sort by conversion rate. Pages below your account average are dragging overall performance down. Pages above average might merit more traffic (higher bids on keywords pointing to those pages).

The Analysis Framework: Weekly, Monthly, Quarterly

Weekly Check

• Total spend vs. budget: are you pacing correctly?
• CPA by campaign: any campaigns above target?
• Review for pause any keyword with spend over 3x CPA target and zero conversions (this page's example rule, see the worked example)
• Check for new search terms to add as negatives (search terms report)

Monthly Analysis

• ROAS by campaign, which campaigns justify their spend?
• Impression share analysis, where are you losing profitable impressions?
• Landing page conversion rates: any underperformers to fix?
• CPA by keyword: identify your best and worst keywords

Quarterly Deep Dive

• Account structure review: do campaigns align with business goals?
• Keyword strategy: coverage gaps, new opportunities, wasteful broad match
• Competitive analysis: impression share trends vs. competitors
• Budget reallocation: shift money from low-ROAS to high-ROAS campaigns
• Statistical analysis of performance trends, are improvements significant?

Common Analysis Mistakes

• Optimizing for CTR instead of conversions: High-CTR ads attract clicks. High-conversion ads make money. These are often different ads. Write ad copy that pre-qualifies clicks (includes pricing, specifics) rather than maximizing click volume.
• Looking at account averages instead of distributions: An average CPA of $50 might mean every campaign is between $40-60, or it might mean one campaign is at $20 and another at $150. The distribution matters more than the average. Export the data and look at the spread.
• Ignoring conversion lag: Some conversions take days or weeks after the click. Analyzing yesterday's campaign performance using same-day conversion data will always undercount. Use a lookback window long enough to cover your usual conversion delay for any performance comparison.
• Not separating brand vs. non-brand: Brand campaigns (people searching your name) inflate overall ROAS. Always analyze brand and non-brand separately. Non-brand ROAS tells you how well you are acquiring new customers. Brand ROAS tells you how well you are capturing existing demand.
• Reacting to daily fluctuations: Daily conversion data is noisy, especially at lower volumes. A keyword with 2 conversions yesterday and 0 today did not "stop working". Compare longer periods before you act.

How to Export and Analyze Google Ads Data

Google Ads' built-in reporting is good for quick checks but limited for deeper analysis. For systematic performance analysis:

1. Export campaign report: Reports → Predefined reports → Campaigns. Include: campaign name, cost, conversions, conversion value, impressions, clicks, impression share, average CPC.
2. Export keyword report: Reports → Predefined reports → Keywords. Include: keyword, campaign, cost, conversions, conversion value, impression share.
3. Export search terms report: Keywords tab → Search terms. This shows actual queries that triggered your ads: critical for finding waste and new opportunities.
4. Analyze: Upload the CSVs to MCP Analytics for automated ROAS analysis, spend efficiency scoring, and optimization recommendations. Or build a spreadsheet with the formulas described above.

Frequently Asked Questions