Change Logs and Edit Windows in Apple Search Ads: Why Stacking Edits Ruins Your Read
Most indie developers who run Apple Search Ads have the same habit. They open the dashboard on a Monday, see a keyword that looks expensive, lower its bid, raise a different bid that seems underpriced, add three new keywords, and bump the daily budget. By Thursday the numbers look different, and nobody can say which change caused it. The problem is not the account or the audience. It is that you changed too many things at once to learn anything.
This post is about edit windows: how long you should let a change run before judging it, and how to keep a simple record so your future self can tell what happened.
Why stacked edits destroy your read
Every change you make to a campaign affects the auction in some way. A higher max CPT on one keyword can pull in more impressions and more taps, which shifts your blended TTR. Adding keywords to an ad group changes which queries you are competing for. A budget increase can let a campaign reach queries it previously never spent on. When you make all of these in one session, the results you see next week are the sum of several effects, and you cannot separate them.
The practical consequence is that you end up explaining outcomes with stories rather than evidence. "The Tuesday bid cut worked" might be true, or the weekend search volume might have gone up, or the new keywords might be pulling the average. Without isolation, every explanation is equally plausible, which means you learn nothing and keep tweaking.
Set one hypothesis per edit
Before you change anything, write down what you expect it to do. Keep it short and specific:
- Lowering the max CPT on a keyword with high taps and low installs should reduce spend on that keyword without dropping installs much.
- Adding a new exact keyword to a clean ad group should produce taps from that exact query, with the search term report confirming it.
- Raising the budget on a campaign that is hitting its cap should increase impressions in the same keywords, not open new query types.
A hypothesis tells you what to measure. If you cannot write one, the edit is probably a guess, and guesses are fine for exploration but should be labeled that way.
Pick a window that fits the metric
Apple does not publish a fixed learning period for Search Results bids, so treat any specific number as a working rule you set for yourself, not a platform guarantee. The right window depends on what you are measuring.
Spend and impressions
These move quickly. Within a day or two you can see whether a bid change moved delivery at all. Judge these on a few full days, and remember that weekday patterns can distort a short sample.
Taps and TTR
Taps need more volume than impressions to be meaningful. A keyword with fifteen impressions a day will swing wildly. Use a window long enough to collect a handful of taps, and compare against the same weekdays from the previous week where possible.
Installs and conversion rate
These are the slowest and noisiest. For a low-volume keyword, you may need weeks to see a clear pattern. Do not declare a conversion rate change from a single weekend of installs.
Revenue
Revenue lags installs. Purchases and trials mature over days or weeks after the install, and attribution data from your revenue tool arrives on its own schedule. Judge revenue on cohorts that have had time to mature, not on the last few days.
Build a minimal change log
You do not need a project management tool. A spreadsheet or a plain text file with one row per edit is enough. For each change, record:
- The date and time you made it
- The campaign, ad group, and keyword affected
- What you changed, with old and new values (bid, budget, match type, status)
- The hypothesis in one sentence
- The date you plan to evaluate it
- The result, written after the window closes
The last column matters most. When you write the result down, you commit to a conclusion that you can later check against the next change. Over a few months, this log becomes more useful than any single dashboard view, because it shows which kinds of edits actually helped your account.
Rules for running edits in parallel
Sometimes you genuinely need to change several things. A competitor may have raised their bids, a new version may have shipped, or a country may need a budget adjustment. Some practical guardrails:
- Change one variable per ad group when you can. Bid changes on one ad group and keyword additions on another are easier to separate than both in the same place.
- Never edit the same keyword twice inside its window. If you need to correct a mistake, log the second edit as its own row and restart the clock.
- Keep budget changes separate from bid changes. A budget increase alters which queries get reached, which contaminates any bid test running alongside it.
- If you must make a broad sweep, such as a seasonal budget change, log it as a single event and evaluate the account as a whole rather than keyword by keyword.
A worked example
Suppose you have an exact keyword with a max CPT you think is too high. Your hypothesis is that lowering it will cut spend without losing much install volume. You log the change on a Monday, plan to evaluate it the following Monday, and leave everything else in that ad group alone. On the Monday after, you compare spend, taps, and installs against the prior week's same weekdays. If spend fell but installs fell proportionally, your hypothesis was wrong, and you log that. You do not then also raise the budget to compensate. You make the next decision with the evidence you have.
The numbers in this example are not benchmarks. The point is the discipline: one change, one hypothesis, one window, one written result.
Takeaway
Edit windows are not about patience for its own sake. They are about making sure each change teaches you something specific. Write a hypothesis before every edit, pick a window that matches the metric, keep a plain change log, and avoid stacking unrelated adjustments in the same few days. If you want a second set of eyes on which changes are worth making today, AdsBuddy reads your Apple Search Ads data alongside revenue and suggests a short, prioritized list of changes, but you approve and apply each one yourself.