A few years into running my own post-production consultancy, I took on a project that nearly broke me: 500 product shots for an e-commerce client, all needing the same sequence of adjustments. Resize to 2000px on the long edge, sharpen, convert to sRGB, export as JPEG at quality 9. Straightforward stuff. But at the volume they needed, and on a Friday afternoon deadline, “straightforward” stops feeling that way fast. I finished the job, but I also spent the following weekend building a system that meant I’d never sit through that particular grind again. Droplets were the centerpiece of that system, and they’ve quietly become one of the most valuable tools in how I work.

What a Droplet Actually Is (And Why It’s Not Just a Shortcut)

Most people who use Photoshop actions eventually discover the Batch command under File > Automate. It works, but it requires Photoshop to be open, you to navigate menus, and your full attention at the start. A droplet is different. It’s a small executable file, typically sitting on your desktop or in a project folder, that you can drag images or entire folders onto. Photoshop launches automatically if it isn’t already open, runs the action you’ve baked into the droplet, and processes every file in the folder without any further input from you.

Under the hood, a droplet is essentially a self-contained batch processing instruction. Photoshop compiles your action and your output settings into a tiny application file, roughly 50 to 200KB depending on complexity. On Mac, it runs as a standard application. On Windows, it’s an .exe. The key distinction from a regular batch operation is portability and independence: once built, the droplet doesn’t care what project you have open, what your current color settings are, or whether you remembered to navigate to the right source folder. You drag, it runs.

Building a Droplet That Holds Up in Production

Go to File > Automate > Create Droplet. The dialog box looks almost identical to the Batch dialog, which means most of the logic transfers directly.

Start by choosing your action. If the action doesn’t already exist, build it first and make sure it runs cleanly on a representative sample of files. For product photography, I typically build actions that include: Camera Raw Filter adjustments baked in with specific settings (not just “open in Camera Raw”), an Unsharp Mask at 85/1.0/3, a Convert to Profile step targeting sRGB IEC61966-2.1, and a Flatten Image before export. That sequence exports consistently across hundreds of files with no surprises.

In the Save and Close section, point your output to a dedicated folder and choose your file naming convention. I use a prefix plus a four-digit sequence number, so client-facing files arrive labeled something like “SKU_0001.jpg” rather than “IMG_3847_final_v2_ACTUAL_FINAL.jpg.” Under file type, JPEG at quality 9 hits the right balance for most e-commerce specs: files stay under 500KB while retaining enough detail for zoom functionality.

One setting most guides skip: check “Suppress File Open Options Dialogs” and “Suppress Color Profile Warnings.” Without these, Photoshop will halt on every raw file or profile mismatch and wait for you to click OK. That defeats the purpose entirely.

The Override Problem and How Actions Break

Here’s where people run into trouble. If your action includes a specific “Save As” step, the droplet’s output settings won’t override it. You’ll end up saving files to whatever location was hardcoded into the action when you recorded it, which is almost always wrong on a different project or machine.

The fix is simple but not obvious: when you record the action, don’t include a Save As step. Let the droplet handle output entirely through its Destination settings. Use “Save and Close” or set an explicit output folder in the droplet dialog. This keeps the action modular and the droplet in control of the final output. I learned this the hard way during an early test run where I processed 80 images and then couldn’t find any of them because they’d all been saved to a folder path on an external drive I no longer had mounted.

Droplets Across Multiple Machines and Client Handoffs

If you’re working with a team, droplets are genuinely useful for standardizing output without requiring everyone to understand actions. A designer or assistant can drag a folder of files onto a droplet and walk away. They don’t need to know what an Unsharp Mask radius does. You’ve already made that decision.

One important limitation: droplets are platform-specific. A droplet built on a Mac won’t run on Windows. If you’re handing a workflow to a Windows-based team, you’ll need to rebuild the droplet on a Windows machine using the same action. The action itself, saved as an .atn file, transfers fine between platforms. It’s only the compiled droplet application that doesn’t cross over.

For client handoffs where they’re providing ongoing file batches, I’ll sometimes build a droplet, nest it inside a shared Dropbox folder with a brief PDF of instructions, and they can run it themselves. It reduces back-and-forth and keeps output consistent without me being in the loop for every batch.

When Droplets Are the Wrong Tool

Droplets don’t handle conditional logic. If some files need one treatment and others need a different one based on content, aspect ratio, or filename, a droplet will apply the same action to everything without hesitation. For that kind of variable processing, you need scripting, either JavaScript in Photoshop or something external like Python with the Photoshop API.

I keep a spreadsheet that tracks hours saved by automation tools I’ve built over the years. Droplets account for a significant portion of a number that’s now over 2,400 hours. But that number only holds because I was honest about what droplets are good at: high-volume, consistent, repeatable sequences. When the work gets complex enough to need branching decisions, I upgrade the tool, not the workaround.

The single most valuable thing you can do with droplets is build one before you need it. Once you’ve run a sequence of adjustments more than 20 times, that’s your signal: this task is now a droplet.