FlickClean: Photo Cleaner

Most people have thousands of photos they will never look at again, and no patience for a file manager. FlickClean turns clearing your camera roll into something closer to a card game: one photo at a time, swipe left to bin it, swipe right to keep it. The bin only empties when you confirm it.

It is also the first thing I built that people paid for, which taught me more about software than any tutorial had up to that point.

The problem with existing cleaners#

I bought four competitors before writing a line of code. Every one of them had the same two failures:

• They ask for everything up front. Full library access on first launch, no explanation. Most people tap "Don't Allow" and never come back.
• Deletion feels dangerous. No undo, no staging, no confirmation. If a cleaner ever deletes one photo you cared about, you uninstall it that day.

So the whole design brief became: earn trust before asking for anything, and make every destructive action reversible until the last second.

How it works#

Photos are never deleted directly. A swipe moves an asset into a local staging set held in Core Data; nothing leaves the device library until the user taps Empty bin and clears the system confirmation. That single indirection is what makes the app feel safe enough to use quickly.

The heavy lifting is batched so the UI never blocks:

/// Requests deletion for every staged asset in a single system prompt.
/// PhotoKit shows one confirmation sheet per performChanges block, so
/// batching here is what keeps the flow from becoming unbearable.
func emptyBin(_ staged: [PHAsset]) async throws {
  guard !staged.isEmpty else { return }
 
  try await PHPhotoLibrary.shared().performChanges {
    PHAssetChangeRequest.deleteAssets(staged as NSArray)
  }
 
  await store.clearStaged()
  analytics.track(.binEmptied(count: staged.count))
}

An on-device Vision pass groups near-duplicate bursts so you review ten similar shots as one decision instead of ten. That alone cut the median time to clear 1,000 photos from roughly [[18 minutes]] to [[6 minutes]] in my own testing.

Monetization#

The free tier cleans [[100]] photos a month. Beyond that it is a subscription, handled through StoreKit 2 with RevenueCat in front of it so I did not have to build receipt validation myself.

Moving the paywall from launch to after the first completed clean roughly [[tripled]] conversion. The app did not change. Only the moment I asked did.

What I would do differently#

• I shipped analytics too late. For the first three weeks I was guessing at where people dropped off. Instrumenting the funnel should have been part of the first release, not the third.
• Core Data was heavier than this needed. The staging set is a list of identifiers. A plain file would have done the job with a fraction of the code.
• I underpriced at launch. Raising the price [[twice]] changed conversion far less than I feared.

Status#

Actively maintained. Current work is on a shared-album mode so families can clear a holiday album together, and on cutting cold-start time on libraries above [[50,000]] photos.