Apple Ads for a New App: Zero Impressions at $4 a Tap, and One $29 Install
Claude (Fable 5.2)
Written by Claude (Fable 5.2) from the campaign plan, its spend ledger and the session transcripts. Earlier Claude sessions built and tuned the campaigns; dested set the budget, said "enable it" and edited. The first person below is dested's voice.
I gave Apple Ads a $1,000 budget to find players for Paint Paws, a puzzle game I shipped this month. Through November 8 it had spent $29.09 and found one.
The spend isn't the surprise. The surprise is that I couldn't make it spend. My best keywords showed zero ads at $1.65 a tap, zero at $2.50 and zero at $4.00.
The setup
Eight Search Results campaigns, four themes in two regions (US, and Canada, UK and Australia together):
| Theme | Match type | Keywords |
|---|---|---|
| Puzzle | exact | 8, such as "color sort", "pixel puzzle", "cat puzzle" |
| Competitor | exact | 6 names of bigger games in the genre |
| Brand | exact | the app's own name |
| Discovery | broad, Search Match on | the same 14 terms as broad, with every exact term as a negative |
That's the structure Apple's own guides recommend. Exact groups are where you pay for intent. Discovery is where you find new terms.
New users only, iPhone and iPad, manual bids, caps adding up to $78.33 a day.
Exact match never served, at any price
Apple suggested $1.65 a tap. On day one, every exact keyword had 0 impressions. Only broad Discovery served.
The usual advice for no impressions is to raise the bid. I did, twice.
At $4.00 a tap, "color sort" in the US showed my ad zero times in two days. Even an exact match on my own app's name got nothing.
So price wasn't the limit. My working explanation is relevance: Apple decides whether an app belongs on a query before the auction runs, and a new app with no ratings and no install history doesn't qualify for much. I can't see that score, so treat it as a theory that fits the numbers.
Broad match spent, on the wrong searches
Discovery was the only thing running, at about $5 a day of a $78 cap.
| Campaign | Spend | Impressions | Taps | Installs |
|---|---|---|---|---|
| Discovery, CA GB AU | $17.95 | 1,300 | 13 | 1 |
| Discovery, US | $11.14 | 906 | 8 | 0 |
| Everything exact | $0.00 | 36 | 0 | 0 |
| Total | $29.09 | 2,242 | 21 | 1 |
Most broad search terms are hidden behind "(Low volume terms)". The two Apple did name were "clash royale" (58 impressions) and "cat collector" (11). Neither is a puzzle game. Both became negatives.
The broad version of one big competitor's name took $4.39 for 3 taps and no installs. People searching for a specific game want that game.
The console hid day one from me
On the first morning the dashboard showed $0.00 spent. The campaigns had been live for hours.
The default range is "Last 7 days", and it excludes today. Reporting is in UTC, and my campaigns started at 00:40 UTC. Everything that had happened was "today".
Set the range to Today before deciding something is broken.
Two budget facts the UI won't tell you
There is no campaign total budget any more. The current UI only has a daily budget. If you want a hard stop at a total, set an end date or build the stop yourself.
The daily budget is soft. Apple can overspend on a single day. With an end date, the ceiling is days times daily budget. Without one, it's the daily budget times 30.4 for the month.
I set an end date on every campaign. My $1,000 limit lives in a ledger I check, not in Apple's settings.
What I changed
- Stopped raising bids. Three prices, same zero. I cut the dead exact groups' caps instead.
- Bought different placements. Four new campaigns: the Search tab, and "while browsing" on other apps' product pages. Neither needs Apple to match me to a query.
- Built a custom product page for ads. My default page opens with three video previews whose poster frames are uncaptioned gameplay. The ad version has no previews and leads with captioned screenshots. It runs beside the default ad, so Apple splits the taps and I can compare install rates.
Total daily cap after the changes: $78.23, down a dime.
If you're about to do this with a new app
- Expect exact match to be silent at first, and don't read that as a bidding problem. Check impressions per ad group on day one. If exact shows zero and broad shows hundreds, more money won't fix it.
- Put every exact keyword into Discovery as a negative, or you can't tell which group a tap came from.
- Read the search terms report daily and add negatives. Mine turned up a strategy game in the first 24 hours.
- Twenty-one taps is not a conversion rate. One install in 21 is 4.8%, and it would be 9.5% with one more. I'm not drawing a conclusion from it yet.
The campaigns end on November 11. I'll post the full week, including whether the Search tab and the custom page did any better, once the numbers are in.