ANSWER

Quick answer

Inspect the rendered head and network response instead of assuming the browser chose your newest PNG. A stale ICO frame, Apple file, or manifest icon can display while the SVG you edited remains unused.

Last verified
September 2, 2026
Evidence
Primary documentation plus scoped technical analysis
Primary sources
2
01

Identify the selected file

Inspect the rendered head and network response instead of assuming the browser chose your newest PNG. A stale ICO frame, Apple file, or manifest icon can display while the SVG you edited remains unused.

Open the exact response over a checkerboard, white, charcoal, and browser-gray background. This distinguishes empty alpha from white pixels that only disappear on light chrome.

Connected next stepCase Study: A Transparent Mark Inside a Maskable Icon for the closest prerequisite or comparison.

02

Separate transparency from adaptive backgrounds

Ordinary PNG and SVG favicons can preserve transparency. Apple and maskable application surfaces may crop or place artwork differently, so a transparent ordinary icon should not be relabeled as maskable without a full-bleed composition.

An all-white symbol needs a deliberate contrasting plate or theme-aware SVG. Relying on a browser to invent a background produces inconsistent results.

Connected next stepFavicon Formats Compared: SVG, PNG and ICO for the next connected implementation decision.

03

Re-export and invalidate one layer

Confirm the source canvas contains visible RGB and alpha data, export without unintended matte flattening, and inspect each ICO frame independently.

Deploy corrected bytes, purge the relevant CDN object, and refresh a clean profile. Change the URL only when your caching policy requires versioning; do not rotate every candidate at once.

Connected next stepOpen the most relevant production tool and verify the decision with a working output.

04

Keep a diagnostic capture

Save the selected URL, response headers, decoded dimensions, alpha-channel preview, ICO frame list, background samples, and browser version. That evidence separates an export defect from a cache or selection problem and prevents the next maintainer from repeating the same blind re-exports.

05

Trace alpha through every export layer

Inspect the master in an editor that can display RGB channels separately from alpha. Fully transparent pixels can contain arbitrary color values, and semi-transparent edge pixels may have been matted against white during an earlier export. When those pixels are later composited or converted into ICO, the hidden matte can appear as a pale fringe or block. Export directly from the clean master, preserve straight alpha where the encoder expects it, and compare decoded results rather than the editor’s checkerboard alone.

ICO deserves frame-by-frame inspection because its images can use different encodings and masks. One large frame may preserve alpha while a 16-pixel BMP-style frame uses an incorrect AND mask and appears blank. Extract every frame to PNG, view it over contrasting backgrounds, and confirm dimensions and bit depth. If a generator repeats one faulty reduction across the container, rebuilding only the outer ICO directory will not repair the underlying pixels.

06

Distinguish browser chrome from image background

A white square may be the selected image canvas, a system-added treatment, or simply the page’s white mark disappearing into light browser chrome. Compare the raw image with screenshots from active, inactive, pinned, private, light, and dark contexts. Also check high-contrast and forced-color modes where available. A self-contained contrasting plate is often more reliable than depending on every browser to choose a helpful background.

After correction, deploy one controlled change and capture its hash. Verify the CDN and origin return that hash, then test a new filename in staging or a clean profile before changing the canonical production URL. If the clean context still shows a square, the pixels or selection remain wrong. If it succeeds, work outward through profile, service-worker, bookmark, shortcut, and search caches. This sequence keeps a visual symptom from turning into repeated unmeasured re-exports.

07

Prevent the symptom during design review

Add a transparency contact sheet to the release process. Render every browser, ICO, Apple, ordinary PWA, and maskable asset over white, black, mid-gray, the brand background, and a checkerboard. Show native 16-, 32-, and 48-pixel browser previews without zoom. Include a maskable crop preview and an extracted image for every ICO frame. Fail the review if the silhouette disappears, an unintended rectangular matte becomes visible, semi-transparent fringes change color, or the selected background differs from the package specification. Store the contact sheet beside the output inventory. This catches blank-square problems before caching obscures them and gives designers a shared artifact for discussing whether a white plate is intentional. It also proves that transparency was tested as rendered pixels across purposes, not inferred from an editor layer name or a PNG extension.

TOOLS

Apply this guide with the relevant tools

Use the smallest tool that verifies the specific decision described above. Each destination keeps the source format, deployment evidence, or platform purpose explicit.

DOWNLOADABLE EVIDENCE

Review the headings, sources, implementation artifact, and update record for this guide as structured JSON.

Download guide evidence ↓
Evidence scope
Guide
Why Is My Favicon Showing a Blank White Square?
Coverage
Behavior can vary by browser, operating system, cache state, deployment configuration, and later software releases. Claims are limited to the cited specifications and documented tests.
Q&A

Questions, answered

Does transparency itself create a white square?+

No. The square can come from white source pixels, a platform-added background, mask composition, or a different selected file.

Why is only the ICO blank?+

One or more embedded ICO frames may have lost alpha or been generated from the wrong source; inspect each frame.