Online advertising metrics fall into reach, engagement, conversion and revenue groups, and the first decision is which business goal the campaign serves. Pick the metric closest to that goal, then use the others to explain why it moved.
Reach metrics suit awareness campaigns, click metrics suit traffic campaigns, and cost and revenue metrics suit campaigns judged on sales. The guidance below follows that order, with a formula for each metric and a note on what the number can and cannot tell you about a targeted campaign.
Choosing a metric by campaign goal
A metric is a countable number, and a KPI is a metric tied to a goal. OWOX draws this line because a change in a KPI should mean something for the business, while a change in an untied metric may only be noise. Before opening a report, name the goal, then choose the one or two numbers that answer it directly.
Matching the goal to the metric
- Awareness goal. Start with impressions and the cost of reaching an audience. These show whether the ad was delivered, not whether it persuaded anyone.
- Traffic goal. Start with click-through rate and cost per click. They show how many viewers responded and what each visit cost.
- Lead or sales goal. Start with conversion rate and cost per acquisition. They show whether visitors completed the action the campaign exists to produce.
- Revenue goal. Start with return on ad spend and sales value. WhatConverts recommends focusing on ROAS and sales value over impressions and clicks alone when revenue is the aim.
Why a standard value matters
A KPI offers meaningful information when the actual value is compared with an average, according to OWOX. A figure with nothing to compare it with cannot be called good or poor. Use the campaign’s own earlier results as that standard, or a target set before launch, and judge each new result against it.
When to move down the funnel
Stay with top-of-funnel metrics only while the campaign has no conversion data to read. Once leads or sales are being recorded, shift the main judgement to the conversion and cost metrics further down, and keep the earlier metrics as diagnostics. This also stops a campaign from being rated on numbers that are easy to inflate. For example, impressions are often treated as vanity metrics because they do not link directly to revenue, as WhatConverts notes.
Reach metrics: impressions and CPM
Impressions count the number of times an ad is displayed. Impressions mean the ads are getting views, Tapfiliate explains, while clicks show that someone engaged. A large impression count therefore confirms delivery and nothing more.
Impressions and audience size
An impression count can come from a small audience seeing the ad repeatedly or a large audience seeing it once. The count alone cannot tell these apart, which is why the guide on impressions vs reach treats them as separate measures. For a targeted campaign, check the audience size alongside the impression total before deciding the targeting is working.
Delivery is also not the same as being seen. A page can load an ad that the visitor never scrolls to, so a campaign judged on reach should be read alongside the material on what is ad viewability.
Cost per mille
It is a pricing and comparison metric: it lets you compare how expensive different placements or audiences are for the same amount of exposure. It says nothing about response, so a cheap CPM is only an advantage when the audience then clicks and converts. The walkthrough on how to calculate cpm covers the arithmetic in detail.
CPM suits brand campaigns where the aim is exposure. It is a weak primary measure for a campaign whose success is a sale, because it rewards cheap delivery regardless of who sees the ad.
Click metrics: CTR and CPC
Click-through rate compares how many users view an ad with how many click it. Tapfiliate states the formula as clicks divided by impressions, multiplied by 100 to give a percentage. DashThis gives the same calculation: number of clicks divided by number of impressions, times 100.
A worked CTR example
Run the same calculation on each ad in a campaign and the results can be ranked directly, because the impression count is built into the formula. The step-by-step guide on how to calculate ctr goes further into the inputs.
What CTR tells you about targeting
Because CTR relates viewers to responders, it reflects how well the creative suits the audience that was shown it. A rate that sits well under the campaign’s usual level points toward a mismatch between the message and the targeted group, or toward weak creative. It is a good early diagnostic for a behavioral or interest-based audience, since it arrives before any purchase data.
CTR is also easy to misread. WhatConverts warns that a high CTR can be deceptive if few of those clicks lead to conversions. Treat CTR as a question to follow up, never as proof of success. If the rate is high and conversions are low, look at the landing page and the match between ad promise and page content before raising the budget.
Cost per click
CPC is the average cost per ad click. It answers how much each visit cost, which is useful for traffic goals and for checking whether a rising CTR is being bought at a rising price. The overview of what is cost per click explains how it is set and reported. Read CPC next to CTR: a falling CPC with a falling CTR can mean cheaper but less relevant traffic.
Conversion rate and web analytics
Conversion rate is conversions divided by visitors, multiplied by 100%, according to OWOX. It measures what happens after the click, which makes it the bridge between traffic metrics and revenue metrics. A conversion should be a defined action, such as a purchase or a form submission, agreed before the campaign starts.
What web analytics adds
OWOX describes monitoring conversion metrics through analytics platforms as crucial for understanding how website visitors become leads or customers. Web analytics is the layer that records visits, traces them to the ad that sent them and counts the completed actions. Without it, click data from the ad platform cannot be connected to outcomes on the site.
Tagging campaigns so conversions can be traced
To link a conversion to a specific ad, each campaign link needs consistent labels. The guide on how to use utm parameters shows how those labels are built. Agree a naming convention before launch and apply it to every link, since inconsistent labels split one campaign across several rows in the report.
Not every conversion follows a click. Some people see an ad, do not click, and convert later. The explanation of what are view through conversions covers how those are counted and why they should be reported separately from click-based conversions.
Reading a conversion funnel
Impressions, clicks and conversions form a funnel in which each stage is a share of the one before. WhatConverts advises pairing CTR with a mid- or bottom-funnel metric so that a strong click rate is not mistaken for a strong campaign. When a campaign underperforms, find the stage where the largest share of people drops out. A weak CTR points to the ad. A strong CTR with weak conversion points to the page, the offer or the audience.
Cost and revenue metrics: CPA, ROAS and acquisition cost
Cost metrics convert spending into the price of a result, and revenue metrics compare the result’s value with the spending. They are the basis for deciding whether to scale a campaign, cut it or change it.
Cost per acquisition
OWOX gives the cost-per-action formula as ad cost divided by completed actions. When the completed action is a purchase or a new customer, this is the figure most marketers mean by cost per acquisition. WhatConverts defines customer acquisition cost as the cost to acquire a customer. The two should be read together: the formula supplies the calculation, and the definition names the outcome being priced.
Choose the action carefully, because it defines the metric. Counting every sign-up gives a different figure from counting only paying customers. The article on what is cost per acquisition goes through how to set that definition. For campaigns that stop at a form submission, what is cost per lead covers the matching measure, and for campaigns judged on likes, shares or comments, what is cost per engagement applies.
Return on ad spend
ROAS is revenue from ads divided by ad cost, per OWOX. It answers a direct question: how much revenue did each unit of ad spend bring back. Use it where the campaign sells something with a known price, and where revenue can be tied to the ads. The page on what is roas covers interpretation and targets. A ROAS figure is only as reliable as the revenue attribution behind it, which is where the next section matters.
Note that ROAS compares revenue with ad cost, not profit with total cost. It does not subtract product costs or overheads. For a full profitability view, the guide on how to measure advertising roi takes the calculation further.
Why clicks can rank campaigns the wrong way
WhatConverts offers a clear comparison. Campaign A produced 100 clicks and brought in $500. Campaign B produced 15 clicks and brought in $5,000. Judged on clicks, Campaign A looks stronger. Judged on sales value, Campaign B is far ahead. The lesson is to rank campaigns by the metric tied to the goal, not by the metric that is simply largest.
Looking beyond one purchase
An acquisition cost only means something against what a customer is worth over time. A campaign with a high cost per acquisition can still be sound when customers keep buying, and a low one can be poor when they never return. The comparison of customer lifetime value vs customer acquisition cost explains how to weigh the two.
Attribution models
An attribution model is the rule that decides which ad touchpoints receive credit for a conversion. The rule matters because a customer may see several ads across channels before buying, and the model determines which channel looks effective and which does not. WhatConverts notes that attribution tools show exactly which channel is producing results, so the model sits behind every channel-level cost and revenue figure.
The main types of model
Models differ in how they spread credit across the path to a conversion. Some give all credit to the first touchpoint, some to the last, and others share it across the touchpoints in between. A single-touchpoint rule is simple to read but ignores everything else the customer saw. A shared-credit rule reflects the whole path but is harder to explain.
Check which model your analytics platform applies by default before comparing any two reports. The same campaign can show different channel results under different models, so a comparison between platforms is only fair when both use the same rule.
Choosing a model for a targeted campaign
- Short purchase journey. A simple single-touchpoint rule is usually adequate, because few touchpoints exist to share credit among.
- Long journey with several channels. A shared-credit approach avoids over-rewarding whichever channel happened to come last.
- Brand and display exposure. Click-only rules undercount ads that were seen but not clicked, so pair them with view-through reporting.
- Budgets across many channels. Where touchpoint tracking is incomplete, the method behind what is marketing mix modeling looks at overall spend and results instead of individual user paths.
Whichever rule you choose, apply it consistently over time. Switching models mid-campaign makes earlier and later results incomparable, and a change in reported performance may then reflect the model rather than the ads.
A/B testing for better ad results
A/B testing improves ad results by running variants of an ad against each other so the better performer can be chosen on evidence rather than preference. DashThis recommends trying A/B testing to see what performs best and to learn from the findings. WhatConverts suggests testing different headlines and value propositions.
Setting up a test
- Change one element. Vary the headline or the value proposition, not both, so any difference in results has an identifiable cause.
- Define the winning metric first. Decide before the test starts whether CTR, conversion rate or cost per acquisition will pick the winner.
- Show variants to comparable audiences. If one variant reaches a better-matched audience, the result reflects the audience and not the ad.
- Run it long enough to settle. Stopping on an early lead risks choosing a variant that was ahead by chance.
Which metric picks the winner
WhatConverts says that until mid- and bottom-funnel metrics are available, the variant with more clicks can be treated as the winner. Once conversion data exists, it should take over. A headline that attracts more clicks but fewer purchases has not improved the campaign, and the conversion rate or cost per acquisition will show that.
The learning is cumulative. Each completed test leaves a finding about the audience, such as which message it responds to, and that finding can carry into the next creative. That is how testing improves results over time rather than only on a single ad.
The core formulas side by side
The table collects the standard formulas for the metrics above, so you can check a report’s calculation at a glance. Each definition is stated as the source gives it.
| Metric | Formula | Planning use |
|---|---|---|
| Click-through rate | (clicks / impressions) × 100% | Compare ads and audiences on response before conversion data exists |
| Conversion rate | conversions / visitors × 100% | Judge the landing page and offer after the click |
| Cost per acquisition | ad cost / completed actions | Price each result and set a maximum acceptable cost per action |
| Return on ad spend | revenue from ads / ad cost | Decide whether to scale a campaign that sells at a known price |
The click-through formula follows Tapfiliate, and the conversion, cost per action and ROAS formulas follow OWOX.
Using the table in practice
Build the four formulas into one reporting sheet with a row per ad or audience, so that each result sits beside the others. Then read the columns from left to right as a funnel. Response comes first, then what happens on the site, then what each result cost, then what it earned. The point at which an ad stops looking good shows where to intervene.
Reporting results and choosing what to do next
A report is useful when it ties each number to a decision. State the goal, the KPI chosen for it, the value compared with the standard, and the action that follows. That shape keeps a presentation focused on the campaign’s purpose rather than on every available figure.
When metrics disagree
- High CTR, weak conversions. Review the landing page and whether the ad promise matches it, as WhatConverts cautions that a high CTR can mislead.
- Strong conversions, weak CTR. The ad may be reaching a narrow but well-matched audience, so test new creative before changing the targeting.
- Many clicks, little revenue. Rank by sales value or ROAS, as the Campaign A and Campaign B comparison shows.
- Cheap CPM, no response. Cheap exposure is poor value when the audience does not act, so review the audience before the price.
Platform-specific reporting
Ad platforms label and group these measures differently, and social platforms add engagement figures of their own. The guide to facebook ads metrics explained maps those labels to the standard metrics above, which makes cross-platform comparison easier.
Frequently asked questions
Which metrics measure online ad campaign performance?
Impressions and CPM measure reach, CTR and CPC measure response, conversion rate and cost per acquisition measure outcomes, and ROAS measures revenue against spend. Choose the one tied to the campaign goal.
Is a high click-through rate always a good sign?
No. WhatConverts notes that a high CTR can be deceptive if few of the clicks lead to conversions, so pair it with a conversion metric.
What is the difference between a metric and a KPI?
A metric is a countable number. A KPI is a metric tied to a goal, so that a change in it means something for the business.
Why are impressions called vanity metrics?
Impressions show that ads were displayed, but they do not link directly to revenue, so they say little about whether the campaign earned anything.
Should an A/B test be judged on clicks or conversions?
Clicks can decide the winner while no mid- or bottom-funnel data exists. Once conversion data is available, use it to pick the variant.
