Open any ad dashboard. Twelve numbers. Most of them don't matter. And the ones sitting up top in big type? Usually the least important. Which makes sense, because those are the ones the platform can make look good.
That isn't an accident of interface design. Cheap, immediate measurement is one of the four things that won digital advertising its budget in the first place, and it arrived with a bill. What's easy to count starts crowding out what's worth counting. Knowing which of these numbers deserves your attention is the defence.
So here's what each metric actually measures. How they connect. And which single number tells you whether you're making money. Worked examples throughout, using one campaign so you can watch a change in one metric shove the others around.
Our example campaign: ₹50,000 spent, 500,000 impressions, 5,000 clicks, 100 sales at ₹2,000 each.
The five that matter
CPM, Cost Per Mille (cost per 1,000 impressions)
What it measures: what you pay to be seen a thousand times.
CPM = (Spend ÷ Impressions) × 1,000
₹50,000 ÷ 500,000 × 1,000 = ₹100 CPM
You paid ₹100 every thousand times your ad showed up.
What it actually tells you: how expensive your audience is. Rising CPM usually means more competition for the same people. Or a narrower audience. Or a seasonal spike, retail CPMs climb steeply near festivals and holidays.
The trap: CPM measures appearing, not being seen. An ad can be served and never scroll into view. Sitting in a slot below the fold that the user never reaches. If your platform reports viewability separately, that's the honest number, and it's usually smaller than you'd guess.
CTR, Click-Through Rate
What it measures: the percentage of people who saw your ad and clicked.
CTR = (Clicks ÷ Impressions) × 100
5,000 ÷ 500,000 × 100 = 1% CTR
What it actually tells you: whether your ad is relevant to the people seeing it. It's the fastest read you get on creative and targeting quality. Low CTR? One of those two is wrong, almost always.
The trap, and it's a big one: a high CTR isn't automatically good. A clickbait headline grabs clicks from people who bounce two seconds later, so you pay for traffic that was never going to buy. CTR is a diagnostic, not a goal. Useful when you're comparing two ads shown to the same audience. Close to meaningless on its own.
CPC, Cost Per Click
What it measures: what you pay for one visit.
CPC = Spend ÷ Clicks
₹50,000 ÷ 5,000 = ₹10 CPC
How it connects: CPC isn't its own thing. It falls out of CPM and CTR. Same impressions, better CTR, lower CPC. So the way to reduce CPC is usually to improve the ad, not to lower your bid.
CPC = CPM ÷ (CTR × 10)
₹100 ÷ (1 × 10) = ₹10 ✓
The trap: cheap clicks are the easiest thing in advertising to buy. Broaden your targeting, lower your bids, and CPC drops immediately. Sales drop with it. Optimising for CPC is the single most common way to spend money confidently and get nothing back.
CPA, Cost Per Acquisition
What it measures: what you pay for one actual customer.
CPA = Spend ÷ Conversions
₹50,000 ÷ 100 = ₹500 CPA
This is the first metric that means something on its own. Everything above measures activity. This measures outcome.
How it connects: CPA is CPC divided by conversion rate. Our 5,000 clicks produced 100 sales, a 2% conversion rate.
CPA = CPC ÷ Conversion rate
₹10 ÷ 0.02 = ₹500 ✓
Worth internalising that relationship. It shows you there are two levers, and most people only pull one. Halving your CPC and doubling your conversion rate do the same thing to CPA. And the conversion rate is usually the cheaper thing to fix, because it lives on your landing page rather than in an auction you don't control.
ROAS, Return On Ad Spend
What it measures: revenue generated per rupee spent.
ROAS = Revenue ÷ Spend
₹2,00,000 ÷ ₹50,000 = 4× (or 400%)
Four rupees back for every one spent. Sounds decisive. It isn't yet. See the next section, which is the part most articles skip.
The metric that actually decides it: break-even ROAS
ROAS measures revenue, not profit. A 4× ROAS is excellent on a 60% margin and loss-making on a 20% margin. Quoting ROAS without margin is how marketing teams most often report a success that lost money.
What you need is the ROAS at which you break even.
Break-even ROAS = 1 ÷ Gross margin
| Gross margin | Break-even ROAS |
|---|---|
| 20% | 5.0× |
| 30% | 3.3× |
| 40% | 2.5× |
| 50% | 2.0× |
| 60% | 1.7× |
| 70% | 1.4× |
Our campaign returned 4×. On a 60% margin (break-even 1.7×) it's genuinely profitable. On a 20% margin (break-even 5.0×)? It lost money while every dashboard showed green.
This one calculation is the difference between running marketing and guessing at it. There's a full walkthrough here.
How they chain together
Each metric feeds the next. Which is why fixing the right one matters.
Impressions → CTR → Clicks → Conversion rate → Sales → Revenue → ROAS
Cost attaches at each stage. CPM buys impressions. CPC buys clicks. CPA buys customers.
The practical use is diagnostic. When results are poor, walk the chain and find the first broken link.
| Symptom | Likely cause | Where to fix it |
|---|---|---|
| Low impressions | Budget or bid too low, audience too narrow | Campaign settings |
| Low CTR | Wrong audience, or weak creative | Targeting and ad |
| Good CTR, low conversion | Landing page or offer | Your site, not the platform |
| Good conversion, poor ROAS | Margin, or price | Your business, not the marketing |
Look where the last two point. Most "our ads don't work" problems are not advertising problems. They're landing page problems. Offer problems. Margin problems. No amount of campaign optimisation fixes any of that. This is the most useful thing on this page.
The numbers most people over-weight
Impressions and reach. Almost meaningless alone. Being seen by a million uninterested people is worse than being seen by a thousand relevant ones.
Engagement. Likes, comments, shares. They feel like progress and correlate weakly with revenue. A viral post that sells nothing is a hobby.
Clicks. A click means someone was curious. It doesn't mean they wanted your product. Half the clicks on any given day are people who tapped by accident, closed the tab, or wandered off to something else on their phone.
Platform-reported conversions. Both Google and Meta report conversions inside their own attribution windows, and both count generously. Add two dashboards together and you'll "achieve" more sales than you actually made. Reconcile against your own order count, always.
Two more worth knowing
AOV, Average Order Value. Revenue ÷ orders. Push AOV up and ROAS improves without you touching the campaign, which makes it one of the cheapest levers in the box. Bundles, upsells, free-shipping thresholds. They all move it.
LTV, Lifetime Value. Total margin from a customer across their whole relationship with you, not just the first order.
LTV is what lets you outbid competitors legitimately. If a customer is worth ₹2,000 on the first purchase but ₹8,000 over two years, you can afford a first-purchase CPA that looks reckless to someone measuring only the first sale. Businesses with repeat purchases and no LTV model are almost always underspending.
The caution: LTV assumes people come back. That's a forecast, not a fact. Use a conservative figure. And make sure your cash flow can survive the gap between paying for acquisition today and earning it back over two years. Cash flow vs profit covers exactly that failure.
A worked before-and-after
Same campaign. Two changes. The landing page lifts conversion from 2% to 3%, and a bundle pushes AOV from ₹2,000 to ₹2,400. Nothing about the advertising changes.
| Before | After | |
|---|---|---|
| Spend | ₹50,000 | ₹50,000 |
| Clicks | 5,000 | 5,000 |
| CPC | ₹10 | ₹10 |
| Conversion rate | 2% | 3% |
| Sales | 100 | 150 |
| CPA | ₹500 | ₹333 |
| AOV | ₹2,000 | ₹2,400 |
| Revenue | ₹2,00,000 | ₹3,60,000 |
| ROAS | 4× | 7.2× |
ROAS nearly doubled and not one campaign setting was touched. That's the point. The biggest gains usually sit outside the ad account, in the page, the offer and the pricing. Which is where almost nobody looks, because the dashboard doesn't show them.
Attribution windows, briefly, because they distort everything above
Every conversion number on this page depends on a setting most people never look at. The attribution window. That's the period after an ad interaction during which a sale still gets credited back to the ad.
A platform set to a 7-day click window credits any purchase within seven days of a click. Set it to 1 day and the same campaign looks like it produced far fewer sales. Nothing about the campaign has changed.
Three consequences worth holding onto.
Comparing campaigns on different windows is meaningless. Before you conclude one outperformed another, check they were measured the same way. This catches people out constantly when they compare across platforms, because the defaults differ.
Longer windows flatter upper-funnel activity. Awareness campaigns look far better on a 28-day window than a 1-day one, because they influence purchases that happen later. Neither number is wrong. They answer different questions.
View-through conversions are the loosest of all. These credit a sale to someone who saw an ad without clicking it. Sometimes real influence. Sometimes coincidence, someone who was buying anyway happened to be served an impression. Treat click-based conversions as your real number and view-through as a directional extra. Never as the headline.
The practical rule. Pick one window, keep it fixed, and judge everything against your own order count. Changing the window mid-quarter makes your entire history incomparable, which is a surprisingly common self-inflicted wound.
What to actually watch
Weekly, in this order.
- CPA against your target. The operational number.
- ROAS against break-even ROAS. The profitability number.
- Conversion rate. The cheapest thing to improve.
- CTR. Only as an early warning that creative is fatiguing.
Ignore impressions, reach and engagement unless you're diagnosing something specific. And check everything against your own sales records rather than the platform's, because the platform isn't a neutral reporter of its own contribution.
FAQs
What is a good CTR?
There's no universal figure. It varies enormously by platform, industry and placement, and a high CTR from clickbait is worse than a moderate one from a relevant ad. Use CTR to compare two ads shown to the same audience. Not as a target in itself.
What's the difference between CPA and CPC?
CPC is what you pay for a click. CPA is what you pay for an actual customer. CPA equals CPC divided by your conversion rate, so a ₹10 CPC with a 2% conversion rate gives you a ₹500 CPA. CPC measures traffic. CPA measures outcome.
What is a good ROAS?
Whatever exceeds your break-even ROAS, which is 1 divided by your gross margin. On a 50% margin you break even at 2×. On a 20% margin you need 5×. A 4× ROAS is strongly profitable in one business and loss-making in another, which is why the number means nothing without margin.
Why are my clicks cheap but I'm not making sales?
Cheap clicks are the easiest thing in advertising to buy. Broaden targeting, lower bids, and CPC drops immediately, along with sales. If CTR looks fine but conversions aren't there, the problem is on your landing page or in your offer. Not in the campaign.
Should I optimise for CPC or CPA?
CPA, always. CPC measures activity and can be improved by attracting worse traffic. CPA measures outcome. And of the two levers behind CPA, conversion rate is usually the cheaper one to fix, because it lives on your own page rather than in an auction.
Can I trust the conversion numbers in Google Ads and Meta?
Not as absolute figures. Each reports conversions inside its own attribution window and counts generously, so adding two dashboards together gives you more sales than you actually made. Use them to compare campaigns within one platform, and reconcile totals against your own order records.
Key takeaways
- CPM, CTR and CPC measure activity. CPA and ROAS measure outcome.
- CPC isn't independent. It falls out of CPM and CTR, so better ads lower it more reliably than lower bids do.
- CPA = CPC ÷ conversion rate. Two levers, and conversion rate is usually the cheaper one.
- ROAS means nothing without margin. Break-even ROAS = 1 ÷ gross margin.
- Walk the chain to diagnose. Most "ads don't work" problems are landing page, offer or margin problems.
- Reconcile every platform number against your own sales records.
Related reading: your marketing either makes money or it doesn't, Google Ads vs Meta Ads, and cash flow vs profit.
Top comments (0)