The TikTok ad manager shows dozens of numbers, and all of them look "important". In practice budgets burn not because of missing data, but because advertisers watch the wrong metrics or read them in isolation. Here is the minimal set of indicators a business really needs - and the logic for connecting them.
The basic dictionary: five metrics you can’t skip
| Metric | What it means | What it tells you |
|---|---|---|
| CPM | Cost per 1,000 impressions | How expensive your auction and audience are |
| CTR | Clicks as a share of impressions | Whether the creative hooks: the opening, the offer, the visual |
| CPC | Cost per click | A derivative of CPM and CTR - what one visit costs |
| CPL / CPA | Cost per lead / target action | What a request, registration or purchase costs |
| ROAS | Revenue ÷ ad spend | The main payback metric: how much money every invested dollar returns |
How the metrics connect: a four-link funnel
Any campaign can be laid out as a chain: impression → click → lead → sale. Each metric describes one transition in that chain, so a problem should always be located at a specific link - not in "the ads in general":
- High CPM - you’re competing for an expensive audience, or the algorithm hasn’t learned yet. Check targeting, geo and frequency.
- Low CTR - the creative doesn’t stop the scroll. The problem is in the first 3 seconds of the video or a weak offer.
- Clicks but no leads - a "broken" handover: slow landing page, awkward form, the ad promises something the page doesn’t deliver.
- Leads but no sales - a question of lead quality and how fast your sales team responds, not of the traffic.
Optimize one link at a time. If you change the creative, the audience and the landing page at once - you’ll never know what actually worked.
Three mistakes that eat budgets
- Optimizing for CTR instead of CPA. A clickable creative is not yet a profitable creative. "Viral" videos often bring cheap clicks and expensive leads. Compare creatives by cost per target action, not by clicks.
- Panic in the first 2-3 days. The algorithm needs roughly 50 conversions per ad group to leave the learning phase. Killing a campaign on day two because of an "expensive lead" resets the learning - and you pay for it again.
- "Average temperature across the hospital". Account-wide ROAS masks reality: inside there are always combos pulling the result up, and dead weight. Watch metrics at the level of each "audience + creative" combo.
Which numbers count as normal
Universal benchmarks don’t exist - CPL in real estate and in e-commerce differ by an order of magnitude. But the correct calculation logic does: start from your economics, not from the market. Work out how much one customer earns you (including LTV) and your lead-to-sale conversion - and you get your maximum allowable CPL, above which ads run at a loss. That personal threshold matters more than any "niche average".
The minimal dashboard: daily vs weekly
- Daily: spend, lead count, CPL per active combo. The goal is to catch an anomaly within 24 hours, not a week.
- Weekly: ROAS, the CPM trend, frequency, creative fatigue (CTR over time), lead → sale conversion.
- Monthly: the channel’s full unit economics: CAC, LTV, TikTok’s share of your total customer flow.
This is exactly the dashboard (Looker Studio, auto-refreshed) we build for clients on the Scale and Custom plans - so decisions are made on numbers, not gut feeling.
Bottom line
TikTok Ads metrics are not a report "for the record" - they’re an early-warning system. CPM and CTR show where the problem is, CPL shows what it costs, ROAS shows whether you earn at all. Read them as a chain, optimize one link at a time - and your budget stops vanishing into the void.