Website Logo Extractor

🖼️ Website Logo Extractor – Bulk Favicon, Brand Image & SVG Logo Scraper

The Website Logo Extractor pulls every logo, favicon and brand image it can find from a list of websites and returns them as clean, absolute URLs in a single dataset. Give it one domain or ten thousand, and for each one you get back the favicon set, the web-app-manifest icons, the Open Graph and Twitter Card images, any <img> tag that looks like a masthead logo, and the raw markup of every inline SVG on the page.

Logo extraction sounds trivial until you try it at scale. Brand marks are scattered across at least five different places in a modern HTML document, they are frequently declared as relative paths, sites ship four or five favicon variants at different pixel densities, and an increasing number of companies render their wordmark as inline SVG that never appears as an image file at all. This website logo extractor handles all of those cases in one pass and normalises every result to an absolute URL you can download immediately.

The actor is built on curl_cffi, an HTTP client that impersonates a real Chrome TLS fingerprint and header ordering instead of announcing itself as a Python script. There is no headless browser in the loop, so throughput is high and memory use is low. Proxy rotation is handled automatically inside the actor — there are no proxy credentials for you to configure — and each request is retried with a fresh IP and randomised backoff when a site responds with 403, 429 or 503.

---

📊 What Data Can You Extract with This Logo Scraper?

Every input URL produces exactly one dataset item. The item carries a small top-level envelope plus a logos array whose entries are typed, so you can filter by the kind of brand asset you actually want.

Category	Fields	What it gives you
Source identity	url	The canonical URL that was processed, exactly as supplied after normalisation, so results stay traceable when items from many runs are appended into one dataset
Result summary	logoCount	The number of de-duplicated brand assets discovered on that site — a single integer you can sort, threshold or use as a coverage check
Asset collection	logos	An array of objects, one per discovered asset, each with its own url, type and size
Asset classification	logos[].type	One of favicon, favicon-default, og-image, img-logo or svg-inline, telling you where the asset came from
Asset geometry	logos[].size	The declared sizes attribute (for example 32x32, 180x180 or any) when the site publishes one, otherwise null
Vector markup	logos[].svgContent	The complete inline <svg> element as a string, present only on svg-inline entries
Failure reporting	error	A human-readable reason when a site could not be fetched or decoded, so failures are visible in the data rather than only in the log

The field most people underestimate is logos[].type. Because every asset is labelled at the point of discovery, you can write a one-line filter that keeps only favicon entries for a browser-style UI, only og-image entries for social preview cards, or only img-logo and svg-inline entries when you need the actual horizontal wordmark that appears in a site's header. Without that classification you would be guessing from filenames, which fails the moment a site names its logo asset-4f2c.svg.

---

🌟 Key Features of the Website Logo Extractor

Feature	Description
🔍 Five discovery strategies in one pass	Parses <link rel="icon"> and friends, the web app manifest, Open Graph and Twitter meta tags, logo-like <img> elements, and inline <svg> markup — all from a single page fetch
🧭 Automatic favicon fallback probe	Sends a lightweight HEAD request to /favicon.ico at the site root and records it as favicon-default when it exists, catching sites that never declare an icon in HTML
📱 Web app manifest parsing	Follows <link rel="manifest">, fetches the JSON, and expands every entry in its icons array into a fully resolved URL with its declared size
🎨 Inline SVG capture	Returns the complete <svg> markup in svgContent, so vector wordmarks that exist only inside the HTML are still recoverable
🛡️ Browser TLS impersonation	Uses curl_cffi with a Chrome 120 fingerprint so requests look like a real browser at the transport layer, not like a scripting library
🔁 Retries with IP rotation	Up to three attempts per URL, each with a new proxy session and randomised backoff, treating 403, 429 and 503 as retryable blocks rather than final answers
⚡ Configurable parallelism	Processes up to maxConcurrency sites at once (default 10, maximum 50), so a large domain list finishes in a fraction of the sequential time
🔗 Absolute URL resolution	Every relative href, src, data-src and manifest src is resolved against the final post-redirect URL, so nothing comes back as /img/logo.png
🧹 De-duplication built in	Identical (type, url) pairs are collapsed before the item is pushed, so logoCount reflects distinct assets rather than repeated markup

---

🚀 Why Choose This Logo Scraper?

No headless browser means real throughput. Chromium-based logo scrapers spend most of their runtime launching and tearing down browser contexts. This website logo extractor issues plain HTTP requests with a spoofed browser fingerprint, so a list of a few thousand domains is limited by network latency rather than by CPU. That directly affects both wall-clock time and compute cost on the platform.

Coverage beats a naive favicon lookup. Fetching /favicon.ico finds one asset and misses the rest. This actor checks that path too, but only as one of five strategies, which is why a single well-built marketing site can yield dozens of distinct assets — Apple touch icons, manifest icons at multiple densities, social preview images, header wordmarks and inline vector marks.

Failures stay inside the dataset. When a site cannot be fetched after its retries, the item is still pushed with logoCount: 0, an empty logos array and a populated error string. You never have to reconcile a dataset against an input list to work out which domains silently vanished, and re-running only the failed rows is a trivial filter.

Zero proxy configuration. Proxy rotation and session management are handled automatically inside the actor, including a fresh session identifier on every retry. There is nothing to buy, wire up or rotate yourself, and no proxy credentials appear anywhere in your input.

---

📥 Input

The logo scraper takes a plain list of URLs plus two optional performance controls. Bare domains are accepted — anything without a scheme is normalised to https:// before the request is made.

{
  "urls": [
    "https://apify.com",
    "https://stripe.com",
    "github.com"
  ],
  "maxConcurrency": 10,
  "timeoutSecs": 30
}

🔧 Logo Scraper Input Fields

Field	Type	Required	Default	Description
urls	array	✅ Yes	["https://apify.com"] (prefill)	List of website URLs to extract logos, favicons and brand images from. Entries without http:// or https:// are automatically prefixed with https://
maxConcurrency	integer	No	10	Maximum number of URLs processed in parallel. Minimum 1, maximum 50
timeoutSecs	integer	No	30	Timeout for each individual HTTP request. Minimum 5, maximum 120

💡 Input Examples

Single site, default settings

{
  "urls": ["https://apify.com"]
}

Bulk brand audit across a domain list

{
  "urls": [
    "https://stripe.com",
    "https://shopify.com",
    "https://notion.so",
    "https://figma.com",
    "https://vercel.com"
  ],
  "maxConcurrency": 25
}

Slow or heavily protected sites

{
  "urls": ["https://example-enterprise-portal.com"],
  "maxConcurrency": 3,
  "timeoutSecs": 90
}

---

📤 Output

Each dataset item describes one website. The example below is trimmed from a real run against https://apify.com, which returned 109 distinct assets.

{
  "url": "https://apify.com",
  "logoCount": 109,
  "logos": [
    {
      "url": "https://apify.com/favicon.ico?favicon.2653ilt32bing.ico?dpl=d89ccaca32",
      "type": "favicon",
      "size": null
    },
    {
      "url": "https://apify.com/apple-touch-icon.png",
      "type": "favicon",
      "size": "180x180"
    },
    {
      "url": "https://apify.com/favicon.ico",
      "type": "favicon-default",
      "size": null
    },
    {
      "url": "https://apify.com/og-image.png",
      "type": "og-image",
      "size": null
    },
    {
      "url": "https://apify.com/assets/header-logo.svg",
      "type": "img-logo",
      "size": null
    },
    {
      "url": null,
      "type": "svg-inline",
      "svgContent": "<svg viewBox=\"0 0 120 32\" xmlns=\"http://www.w3.org/2000/svg\">…</svg>",
      "size": null
    }
  ]
}

🧾 Website Logo Extractor Output Fields

Field	Type	Description
url	string | null	Canonical URL of the scraped item
logoCount	integer | null	Number of de-duplicated logo and brand-image assets found for this site
logos	array | null	Logos values collected for the item — see the asset object table below
error	string | null	Error message, if the item failed to process

Asset objects inside logos

Field	Type	Description
url	string | null	Absolute URL of the asset, resolved against the final page URL. null for inline SVG entries, which have no file of their own
type	string	favicon (declared in HTML or the manifest), favicon-default (found at /favicon.ico), og-image (Open Graph or Twitter Card image), img-logo (an <img> matched as a logo), or svg-inline (inline vector markup)
size	string | null	The declared size, for example 32x32, 512x512 or any, when the site publishes a sizes attribute. null otherwise
svgContent	string	The full inline <svg> element as a string. Present only on svg-inline entries

---

💻 How to Use the Website Logo Extractor (Step by Step)

Step 1: Open the Logo Scraper and Load the Input Form

Open the actor on the Apify platform and switch to the Input tab. The form is deliberately short: a URL list and two performance dials. The urls field is pre-filled with https://apify.com so you can run it once and see the shape of the output before committing a real list. Everything else has a working default, so a first run needs no configuration at all.

Step 2: Build Your Website URL List

Paste your domains into the urls field, one per line. You can mix full URLs and bare domains — github.com, https://github.com and https://github.com/ all resolve to the same request because the actor normalises anything missing a scheme to https://. If your list lives in a CRM export or a spreadsheet, extract the website column, strip obvious duplicates, and paste the result straight in. Redirects are followed automatically, so an old domain that now points somewhere else will still return the destination site's logos.

Step 3: Tune Concurrency for Your List Size

maxConcurrency controls how many sites are fetched at the same time and accepts 1 to 50. The default of 10 is a sensible middle ground. For a large list of unrelated domains, raising it to 25 or higher cuts total runtime substantially because each site is a separate host with its own rate limits. For a small list, or when several URLs point at the same domain, keep it low so you are not hammering one server with parallel requests.

Step 4: Set a Request Timeout That Matches Your Targets

timeoutSecs applies to each individual HTTP request, not to the run as a whole, and accepts 5 to 120 seconds. Fast marketing sites are comfortable at the default of 30. Enterprise portals, government sites and pages served from distant regions often need 60 to 90 before their first byte arrives. Setting the timeout too low is one of the most common causes of an error field appearing on sites that would otherwise have worked.

Step 5: Run the Logo Extractor and Watch the Log

Click Start and follow the log. Each site logs a Processing <url> line when it begins and a Found N logos for <url> line when it completes, so you can see progress in real time. Retry attempts are logged as warnings with the status code or exception that triggered them, which makes it easy to distinguish a site that is genuinely blocking from one that simply timed out on a slow first connection.

Step 6: Review the Dataset and Filter by Asset Type

Open the Dataset tab when the run finishes. Sort by logoCount to see which sites returned the richest asset set and which returned nothing. Then filter the logos array by type depending on what you are building: favicon and favicon-default for compact UI icons, og-image for social preview thumbnails, and img-logo plus svg-inline for full brand wordmarks. Rows carrying an error value are your retry queue.

Step 7: Export or Pipe the Results Downstream

Export the dataset as JSON, CSV, Excel or XML from the platform, or pull it through the API. Because every asset URL is absolute, a downstream script can download the files directly with no base-URL bookkeeping. For inline SVGs there is nothing to download at all — svgContent already holds the complete markup, ready to write to a .svg file or inject into a component.

---

🔌 API Access & Integrations

Run the website logo extractor synchronously and receive the dataset items in the same HTTP response:

curl -X POST "https://api.apify.com/v2/acts/scrapers-hub~website-logo-extractor/run-sync-get-dataset-items?token=YOUR_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "urls": ["https://apify.com", "https://stripe.com"],
    "maxConcurrency": 10,
    "timeoutSecs": 30
  }'

The same run from Python using the official client:

from apify_client import ApifyClient

client = ApifyClient("YOUR_TOKEN")

run_input = {
    "urls": ["https://apify.com", "https://stripe.com"],
    "maxConcurrency": 10,
    "timeoutSecs": 30,
}

run = client.actor("scrapers-hub/website-logo-extractor").call(run_input=run_input)

for item in client.dataset(run["defaultDatasetId"]).iterate_items():
    if item.get("error"):
        print(f"FAILED {item['url']}: {item['error']}")
        continue
    print(f"{item['url']} -> {item['logoCount']} assets")
    for logo in item.get("logos") or []:
        if logo["type"] in ("img-logo", "svg-inline"):
            print("  wordmark:", logo["url"] or "(inline svg)")

Beyond direct API calls, runs and dataset items can be wired into Zapier, Make, Google Sheets, Slack and any HTTP endpoint through Apify's webhook integrations.

---

💡 Best Use Cases for Website Logo Data

🏢 Building a Company Logo Database for a CRM or Directory

If you maintain a B2B directory, an investor database or a CRM with thousands of company records, run every stored website through this logo scraper and keep the highest-quality img-logo or manifest favicon entry per domain. The size field lets you prefer a 512×512 manifest icon over a 16×16 legacy favicon automatically, and logoCount tells you at a glance which records need a manual fallback.

🎯 Competitive Brand Monitoring

Schedule the extractor against a fixed list of competitor domains and diff the logos array between runs. A new og-image value usually signals a repositioned marketing message; a changed svg-inline wordmark or img-logo URL is a strong signal of a rebrand, often visible weeks before it is announced.

🔗 Rich Link Previews and Bookmarking Tools

Link preview cards need two things: an icon and a hero image. This actor returns both in one call — favicon or favicon-default for the small mark, og-image for the large preview. Because the URLs are absolute and post-redirect, a preview service can cache them straight away without re-resolving relative paths.

📊 Marketing and Sales Collateral Automation

Pitch decks, comparison tables and customer logo walls all need clean brand marks at speed. Feed the account list from your pipeline into the urls field and collect the img-logo and svg-inline results; the SVG markup in svgContent scales to any slide size without the halo artefacts you get from upscaling a favicon.

🧪 Web Asset and Accessibility Auditing

A site with logoCount: 0 and no error is publishing no discoverable brand assets at all — no declared favicon, no manifest, no Open Graph image. That is a genuine SEO and social-sharing defect. Running the logo extractor across a portfolio of client sites surfaces those gaps as a sortable column rather than a manual page-by-page review.

🤖 Training Data and Visual Search Pipelines

Brand-mark classifiers, logo-similarity search and trademark monitoring all need large volumes of labelled imagery. The type field gives you free weak labels, logoCount lets you drop low-signal domains, and the mix of raster URLs and svgContent vectors covers both bitmap and vector training paths.

🛒 E-commerce Vendor and Marketplace Onboarding

Marketplaces that onboard hundreds of merchants can pre-populate each storefront profile with the merchant's own brand assets instead of asking for a manual upload. Pass the merchant's website through the scraper, take the largest manifest icon by size, and fall back to og-image when no square asset exists.

---

⚙️ Tips for Better Logo Scraping Results

• Prefer manifest icons when you need square, high-resolution marks. Entries with an explicit size such as 512x512 come from the web app manifest and are almost always cleaner than the legacy /favicon.ico fallback, which is frequently a 16×16 relic nobody has updated.
• Filter svg-inline entries before storing them. Sites often contain dozens of inline SVGs — chevrons, social icons, illustration fragments — and every one is captured. Keep the entries whose svgContent includes a viewBox with a wide aspect ratio, or that sit near the top of the array, since header markup is parsed in document order.
• Raise timeoutSecs before you raise the retry count in your head. Most error values reading "Failed to fetch page after retries" on otherwise healthy sites are slow first-byte responses, not blocks. Ninety seconds usually resolves them.
• Keep maxConcurrency low when many URLs share one domain. Parallelism helps across different hosts; against a single host it just increases the chance of a 429 and triggers retries that slow the run down overall.
• Deduplicate your input list first. The actor pushes one item per input entry, so a duplicated URL produces duplicated items and consumes compute twice.
• Re-run failures separately rather than re-running the whole list. Filter the dataset for a non-empty error, feed those URLs back in with a higher timeoutSecs and a lower maxConcurrency, and you will typically recover most of them.

---

🛠️ Troubleshooting

Why does the run fail immediately with "No URLs provided in input"? The urls field is required and the actor stops as soon as it finds it empty. Blank lines and whitespace-only entries are stripped before processing, so a field that looks populated but contains only empty lines counts as empty. Add at least one valid URL and re-run.

Why is logoCount zero with an error field set? The page could not be retrieved or its body could not be decoded. "Failed to fetch page after retries" means all three attempts failed — usually a timeout, a DNS failure, or a site that returned 403, 429 or 503 on every attempt with a fresh IP. "Could not decode response body" means the response arrived but its encoding could not be read. Increase timeoutSecs, reduce maxConcurrency, and try again.

Why is logoCount zero with no error at all? The page was fetched and parsed successfully but genuinely contained no discoverable assets. This happens on single-page applications that inject their entire header after JavaScript execution, and on minimal landing pages with no declared icon, no manifest and no Open Graph tags. Because this logo scraper does not run a headless browser, client-side-rendered markup is not visible to it.

Why do I get far more entries than the site appears to have logos? Every inline <svg> element counts as an asset, and modern component libraries emit a great many of them. A logoCount in the hundreds is normal for an icon-heavy marketing site. Filter the logos array by type to get to the small set you actually care about.

Why does an img-logo entry point at something that is clearly not a logo? <img> elements are matched when the word "logo" appears in the src, data-src, data-lazy-src, class, id or alt attribute, or in the parent element's class or id. Partner badges, client logo walls and "as seen in" strips all legitimately match that rule. Combine the match with position in the array or with the image URL path when you need only the site's own mark.

---

❓ Frequently Asked Questions About Website Logo Scraping

What does the Website Logo Extractor actually do? It fetches each website you supply, parses the HTML, and returns every favicon, web app manifest icon, Open Graph or Twitter Card image, logo-like <img> element and inline SVG it can find, as absolute URLs in a single dataset item per site.

Do I need to download the logo files myself? Yes for raster and external vector assets — the actor returns URLs, not binary files, so you fetch the ones you want. Inline SVGs are the exception: their complete markup is already in svgContent, so nothing further needs to be downloaded.

Can this logo scraper handle bare domains without https://? Yes. Any entry that does not start with http:// or https:// is automatically prefixed with https:// before the request, so github.com and https://github.com behave identically.

How many websites can I process in one run? There is no fixed cap in the input schema. Runtime scales with list size and with maxConcurrency, which accepts up to 50 parallel fetches, so large lists are practical in a single run.

Does the website logo extractor use a headless browser? No. It uses curl_cffi with a Chrome 120 TLS fingerprint, which means real browser-like requests at HTTP speed without any rendering step. The trade-off is that content injected purely by client-side JavaScript is not visible.

Do I need to configure proxies? No. Proxy rotation is handled automatically inside the actor, with a fresh session identifier generated for each retry attempt. There are no proxy fields in the input and no credentials to manage.

What happens when a website blocks the request? Responses of 403, 429 and 503 are treated as retryable. The actor tries up to three times per URL, taking a new proxy session and waiting a randomised backoff between attempts. If all attempts fail, the item is still written with logoCount: 0 and an error message.

What is the difference between favicon and favicon-default? favicon entries were declared explicitly in the page — through <link rel="icon">, apple-touch-icon, mask-icon and similar, or through the web app manifest's icons array. favicon-default is the fallback found by probing /favicon.ico at the site root, and only appears when that file actually responds.

How do I get the highest-resolution logo available? Filter logos for entries with a non-null size, parse the pixel dimensions, and take the largest. Those values come from declared sizes attributes on link tags and manifest icons, which is where high-density assets such as 192×192 and 512×512 are published.

Can I extract logos from pages behind a login? No. The actor sends unauthenticated requests and there is no field for cookies, headers or credentials, so only publicly reachable pages can be processed.

Does the logo extractor follow redirects? Yes. Redirects are followed automatically and relative asset paths are resolved against the final destination URL, so a domain that now redirects elsewhere still returns correctly resolved absolute URLs.

Why do some logos entries have a url of null? Those are svg-inline entries. An inline SVG has no file of its own because it lives directly in the HTML, so the markup is returned in svgContent instead and url stays null.

Can I use this website logo extractor on a schedule? Yes. Schedule it on the platform like any other actor and diff the logos array between runs to detect rebrands, new social preview images or changed favicon sets.

What export formats are supported for logo data? The dataset can be exported as JSON, CSV, Excel, XML or RSS, or read through the API with run-sync-get-dataset-items and the apify_client libraries. Note that the nested logos array flattens awkwardly into CSV; JSON is the better choice when you need the full asset objects.

Are the extracted logos free to use commercially? No. Logos and brand marks are trademarks belonging to their owners. This tool retrieves publicly published assets; it grants no licence, and how you use them remains your responsibility.

---

🆘 Support & Feedback

Found a bug, hit an unexpected error value, or want a field added? Open a ticket on the actor's Issues tab with the run ID and the input you used — that is the fastest route to a fix, and the run log tells us exactly which stage failed.

Need something custom? Bespoke asset-extraction rules, a different logo-matching heuristic, direct upload of downloaded files to your own storage, or a private version of this website logo extractor tuned to your list — email scraperhubapi@gmail.com and describe what you need.

If the logo scraper saves you time, please leave a review on the actor page. Ratings genuinely help other people find it, and the feedback shapes what gets built next.

---

⚖️ Disclaimer

The Website Logo Extractor retrieves only publicly accessible data from public web pages. It does not bypass authentication, does not access private or password-protected areas, and does not attempt to defeat access controls.

You are responsible for how you use the extracted data. Logos, wordmarks and brand imagery are typically protected by trademark and copyright law; retrieving an asset URL does not grant you any licence to reproduce, modify or redistribute the asset. Confirm your rights before using any brand mark in a commercial context.

Where extracted material relates to identifiable individuals, applicable privacy legislation including the GDPR and CCPA may apply. Establish a lawful basis for processing, keep only what you need, and honour data subject requests. Respect each website's Terms of Service and robots.txt, and keep request volumes reasonable.

For data removal requests, or if you believe content collected by this actor should not have been retrieved, contact scraperhubapi@gmail.com and we will respond promptly.