Quick answer. Read your metrics as a chain. CTR tells you if the creative grabs attention. CVR tells you if the page and offer convert. CPI tells you what one install costs. ROAS tells you if the money comes back. Each number points to a different fix.

When I open a fresh campaign, I do not read every column. I read four numbers in order. They tell me where the problem is and where it is not.

This guide walks you through CPI, CTR, CVR, and ROAS the way I actually use them on the job. No theory you will never touch. Just what each number means and what you do when it looks bad.

What do CPI, CTR, CVR, and ROAS actually mean?

Let me define them in plain words first. No jargon.

  • CTR (click-through rate): out of the people who saw your ad, how many clicked. If 100 saw it and 2 clicked, that is 2 percent.
  • CVR (conversion rate): out of the people who clicked, how many did the thing you wanted (install, signup, purchase).
  • CPI (cost per install): how much you paid for one install. Total spend divided by installs.
  • ROAS (return on ad spend): revenue divided by ad spend. Spend 100 dollars, get 150 back, that is a ROAS of 1.5.

Memorize the order: view, click, convert, pay back. That is the journey one user takes. Your metrics just measure each step of it.

Which metric do I look at first?

I start with CTR. It is the cheapest signal you can get. If people are not clicking, nothing downstream matters.

Here is how I think about it. A low CTR usually means the creative or the audience is off. The ad is not stopping the scroll. I do not touch the landing page yet, because the page is not the problem when nobody is clicking.

If CTR is fine but CVR is low, the problem moved downstream. People click and then leave. That points at the landing page, the offer, or a mismatch between the ad and what the user lands on. A classic mistake: the ad promises one thing, the page shows another.

Only after CTR and CVR look healthy do I judge CPI and ROAS. Those two are the result of everything above plus your bids and competition.

What is a good number for each metric?

This is the question everyone asks, and the honest answer is: it depends on the platform, the country, and the vertical. Anyone who gives you one magic number is guessing.

So do this instead. Run your campaign for a few days, write down your numbers, and treat those as your baseline. From then on you compare to yourself, not to a blog post.

Rough starting reference points I keep in my head for mobile UA, just so you are not flying blind:

  • CTR on a cold paid social audience: often somewhere around 1-2 percent. Below that, look at the creative.
  • CVR from click to install: varies a lot, but a sudden drop versus your own baseline is the alarm.
  • ROAS: you want it trending toward 1.0 and above over the payback window you care about.

Use these as a sanity check, not a target. Your baseline beats any benchmark because it reflects your real product and your real traffic.

How do I avoid fooling myself with the numbers?

Small numbers lie. If 30 people clicked and 1 converted, your CVR is not really 3.3 percent. You just do not have enough data yet. Wait for volume before you trust a rate.

Watch the time window too. ROAS on day 1 looks terrible for almost every app, because users pay later. Always say which day you are measuring: day 1, day 7, day 30. Compare like with like.

And do not average across things that are different. One country can carry the whole campaign while another bleeds money. If you only look at the blended number, you miss it. Break the report down by country, by placement, by creative. The story is in the breakdown.

Key takeaways

  • Read metrics in order: CTR, then CVR, then CPI and ROAS. Each one points to a different fix.
  • Your own baseline matters more than any benchmark you read online.
  • Small samples and short time windows lie. Wait for volume and pick a clear payback window.

Frequently asked questions

Is a high CTR always good?
Not always. A high CTR with a low conversion rate means the ad attracts clicks but the wrong people, or it overpromises. You paid for clicks that did not turn into installs or sales. Always read CTR next to CVR, never on its own.
What is the difference between CPI and ROAS?
CPI is what one install costs you. ROAS is whether the money comes back as revenue. A low CPI looks great, but if those installs never spend, ROAS stays bad. CPI measures cost, ROAS measures the return. You need both.
How much data do I need before trusting a metric?
Enough that the rate stops jumping around. As a rough rule, wait for a few hundred clicks before you judge CTR, and a meaningful number of conversions before you judge CVR. With tiny samples, one extra conversion swings the whole percentage.
Why is my day 1 ROAS so low?
Because most users pay later, not on the first day. Day 1 ROAS being under 1.0 is normal for many apps. Track ROAS by window: day 1, day 7, day 30. Judge the campaign against the payback period your business actually uses.