One conversion doesn’t explain anything.
The phrase “campaign won’t convert well” is too general. There are several transitions between an ad showing and a confirmed FTD, and each answers its own question. As long as these steps are not separated, optimization turns into guesswork.
The minimum funnel looks like this:
Show → Click → Registration → FTD → Confirmed Action.
What metrics to count
| Phase | Formula | What diagnoses |
|---|---|---|
| CTR | clicks/shows | The power of the message and getting into the audience |
| CR click → registration | registration/clicks | Matching Landing to Expectation |
| CR registration → FTD | FTD/registration | Quality of the further user path |
| Approval rate | confirmed by FTD/FTD | Quality and conformity to conditions |
If the CTR is high but registrations are low, the problem is more often found after a click. If there are enough registrations, and there is almost no FTD, it is pointless to endlessly remake an advertising banner – you need to study the next stage.
Example of diagnostics
Campaign A received 5,000 clicks, 600 sign-ups and 120 FTDs. Campaign B – 3,500 clicks, 560 sign-ups and 140 FTDs. A has more traffic, but B has more efficiency.
A:
- click → registration = 12%;
- registration → FTD = 20%.
B:
- click → registration = 16%;
- registration → FTD = 25%.
At an equal price, a B click can be noticeably more profitable, although visually the volume of A looks more solid.
Where to look for a problem after a click
Check four things: page speed, matching the promise in the ad to the content of the landing page, understandability of the next step, and mobile display. Take a look at the GEO and the device separately. The average figure for all users can hide that Android is working normally, and a certain mobile browser is failing the funnel.
Where to find a problem after registration
Not only advertising promises are important in this area. Compare segments by source, GEO, and time. If a drop appears abruptly on a particular day, check for technical changes. If it is stable in only one source, the audience quality problem is more likely.
Why do we need to consider confirmation?
For a CPA, it is not necessary for every fixed FTD to become a paid action. Therefore, the cost of raw FTD and the cost of confirmed FTD are different indicators. Scaling the source by the first digit is dangerous.
Practical arrangements
- Don’t change your creative, landing and audience at the same time.
- Find the weakest funnel crossing.
- Break it down by GEO, device and SubID.
- Formulate one hypothesis.
- Change one big factor.
- Compare the results on a comparable volume.
If the tags in the campaign are not yet structured, start with the material about UTM and SubID. He'll be in this calculation base on schedule.
Look at the absolute losses.
Percentages are useful, but the business effect is best seen in absolute numbers. If the transition falls from 40% to 35% on 100 registrations, the loss is five FTDs. If it’s 15% to 14% on 10,000 registrations, the loss could be much larger. Prioritize the site not only by the lowest CR, but also by the potential number of returns.
Map of causes by stage
The problem before the click is more often related to the audience and the message. Between clicking and registration - with the match of landing, speed and clarity. Between registration and FTD - with the quality of expectations and further user path. Between FTD and confirmation - with the terms of the offer and the quality of the source.
This card doesn't prove the cause, but it helps not to test the wrong layer.
How to make a weekly review
Choose one of the most expensive failures each week. Write down: Segment, current CR, absolute number of lost actions, estimated cause, and one next test. In a month, you'll have four proven hypotheses instead of a dozen chaotic changes.
If the problem appears simultaneously in multiple unrelated sources at the same time, be sure to check for a common technical or product cause before changing advertising campaigns.
Example of prioritization
Let’s say you have 10,000 clicks. A loss of one percentage point on click → registration can mean 100 registrations. If 25% of registrations turn into FTDs, the potential cost of this loss is about 25 FTDs. This arithmetic helps to understand which part of the funnel deserves attention first.
Watch out for Conversion Delay
Not all users make the journey in one day. If the source has a longer time from click to FTD, the “for today” report will understate its quality. To compare campaigns, use the same window after clicking or signing up.
Funnel as a language of communication
When a manager, a media buyer, and a developer look at the same stages, the discussion becomes more accurate. Instead of “something broke” we can say: “clicks reach, registrations have not changed, and FTD after registration fell only on iOS.” This is almost a complete diagnostic task.
