ReWords AI

Image editing is the highest-volume, lowest-visibility line item in most agency retainers, and it splits cleanly into two economic halves: automated optimization that should cost near zero per asset, and creative retouching that clients will pay premium rates for. Agencies that blend the two into a single hourly bucket subsidize the repetitive work with the high-margin work, and the gap widens as page speed, Core Web Vitals, and AI-mediated brand visibility become measurable client KPIs.

The Optimization Ceiling Test asks one question before any image pipeline work: how much of the remaining file weight is actually recoverable, and at what quality cost? Automated compression, resizing, and WebP/AVIF conversion deliver most of their benefit in the first pass. MageCDN reports reductions up to 90 percent on unoptimized uploads, but a second pass on already-compressed assets typically returns single-digit gains while risking visible artifacts on hero imagery. For agencies, the framework separates two budgets: the automated pipeline that runs on every client asset at near-zero marginal cost, and the manual retouching reserved for premium brand work where nuance carries the retainer. Apply the test per asset class. Product thumbnails and blog imagery belong in the automated lane; campaign hero shots and packaging renders do not. The trap is treating every image as a compression problem when the client is paying for a craft problem.

The Retouch Threshold Line is the point in a visual asset pipeline where automated optimization stops adding value and human retouching starts protecting brand equity. Below the line, compression, format conversion, and resizing are commodity operations that scale without judgment: MageCDN serves images through a dual Cloudflare and BunnyCDN stack with automatic AVIF/WebP compression, cutting file sizes up to 90 percent, and Imagify handles WebP/AVIF conversion inside WordPress without manual effort. Above the line, nuance matters: skin tones, product color accuracy, compositing, and campaign-specific crops where a wrong pixel costs a client relationship. Agencies that treat the whole pipeline as one bucket either overpay for commodity work or ship flat creative. The practical move is to draw the line per client tier: automate everything below it, and route premium brand work to a named retoucher or compositing service with a defined revision cap. Magic Layers can rebuild editable layers from flat PNGs when source files are gone, but it does not replace a retoucher's eye on a hero shot.

Asset Provenance Split is the practice of classifying every client image by whether the agency still holds the editable source file. Files with live source custody (PSD, layered Figma, original RAW) sit on a cheap retouch path: a color shift, a headline swap, or a crop for a new placement takes minutes. Flattened exports sit on an expensive path, because rebuilding layers is manual labor billed at creative rates. The framework matters because agencies quote retainers before knowing which path a client's library sits on, and the gap between the two paths is where scope creep hides. Magic Layers demonstrates the recovery route by separating flat PNG or JPEG files into named transparent layers, including text, subjects, decorations, and background, so a finished asset can be repurposed without the original file. Schist covers the other end, opening Affinity and Photoshop files directly. Audit custody before pricing the retainer, not after the first revision request lands.

Automate the optimization pipeline end to end, then keep or outsource the retouching step as a named, separately priced line item.

Separate the automated optimization pipeline from creative retouching in the contract, and price the retouching line per asset with a fixed revision ceiling.