Hacker News new | past | comments | ask | show | jobs | submit login
Show HN: Free Company Logo API (clearbit.com)
525 points by maccman on June 30, 2015 | hide | past | favorite | 152 comments

I honestly thought this was going to generate a free logo and got really excited for some unknown reason? I guess I was wondering how it all worked, then I clicked through and it made way more sense. That was a wild ride!

Not quite free, but this does exist at a pretty low price point: https://www.tailorbrands.com/

Very interesting indeed. Is there any portfolio, even for fake brands, so I can see the range of styles their logos may have?

Hey there. I work for Tailor. We don't have currently have a portfolio, but we are working on something akin to this. For anyone interested, you can email me at nate@tailorbrands.com, and I'll send a coupon code your way so you can give it a try.

AFAIK there is not. Maybe one of their devs will see this and can provide a link or something though.

Another great option I found recently is https://www.logaster.com/create/logo/. Free for low-res small images.

Neat! Definitely checking these guys out.

Squarespace has a logo generator: http://www.squarespace.com/logo/. Watermarked version and low-res for free, $10 otherwise.

That is very, very nice. For stylized artwork, you can search for keywords. I typed "apps" (I'm an iOS developer) and got a bunch of icons that gave me enough good ideas to open up Sketch and create one for my own one-man-company.

The post mentions Gravatar, so actually it should do the same thing and return a generated image as a function of the domain, if no known one exists.

It might be nice if that were an option, but having that as the default behaviour would be strange and unexpected IMO.

That's exactly what I thought. I told it to generate Google's logo and was then quite surprised when it actually produced Google's logo.

That would be an interesting extension to develop. I agree it should be optional.

I'm envisaging scraping colours and fonts from the target domain to recreate a text-only logo.

You could also use the first letter of the domain as the basis of the logo. (I do something a bit like that on faviconist.com)

Not an API, but I used a website to generate a free logo for an MVP I'm building. So far so good. http://www.hipsterlogogenerator.com

I thought the same thing but as a designer I had the opposite reaction. I was prepared to be offended.

Fear of change is justifiable in my opinion; offense isn't.

There are parts of any expertise that are formulaic, and by using the word "parts" I'm being generous to humankind.

Should systems/ops folks be 'offended' that PaaS are becoming the default choice for new projects?

If you DO think they should be offended, then I think you're flat wrong. Being offended by commoditisation is like being angry at entropy, or enraged by gravity.

If you DON'T think they should be offended; why not? Does their job not require thought, or creativity? I'd answer that one carefully if I were you ;)

Why in the world would this offend you?

Two reasons:

1. It degrades design as a profession. It's essentially saying that logo design can be done by an algorithm. "There's no creativity to it". Wouldn't you be offended if someone said a basic algorithm could eliminate your job? (perhaps you wouldn't)

2. It's an overly simplistic view of what a logo/brand identity is. You can't just pick out some basic shapes and text for your logo. You need color theory, contrast analysis, a brand strategy, a well developed concept, etc.

> Wouldn't you be offended if someone said a basic algorithm could eliminate your job?

Of course not. In fact, I spend quite a large portion of my time automating most of my work, so I can do other, more interesting things.

Besides, if a company looked for and used a "free logo generator", would that company really be the type of client you wanted to pick up?

> Besides, if a company looked for and used a "free logo generator", would that company really be the type of client you wanted to pick up?

Of course not. That still doesn't mean that the sentiment isn't degrading to designers. (it is)

There are are a lot of code generators (written by programmers) and they never seem to offend the programmers. I don't think everything can be automated with satisfying results or in a reasonable time but come on, why would an attempt be offending or degrading?

In fairness, I often see devs highly offended by automated output, including where it makes no difference whatsoever to performance and maintainability (and particularly so if it improves on any of those). There is plenty of frothing out there about hand-tuned SQL vs ORM and hand-crafted semantic HTML/CSS for example.

ESR's 17 Unix Rules [1], Rule of Generation:

> Developers should avoid writing code by hand and instead write abstract high-level programs that generate code. This rule aims to reduce human errors and save time.

[1] https://en.wikipedia.org/wiki/Unix_philosophy#Eric_Raymond.E...

It see you're downvoted, but I'm not going to pile on and downvote you to invisibility.

I think your perspective is interesting. Wrong, but interesting and probably shared by many others in various fields.

"We're all sorry for the other guy when he loses his job to a machine. When it comes to YOUR job, that's different. And it always will be different."

- Leonard McCoy

I'm a designer, too, and there are those of us who believe that not only is design automation possible, it is likely, and we will be much the better as contributors and society for it.

Google's Mathias Duarte advocated as such at UXPA last year in his keynote.

Think about it this way - design trends are just manifestations of a particular temporal affinity towards one aesthetic or another. Machines can measure this and expound as well as react to it at a much faster scale than the human mind can.

Design isn't art. There are way too many designers out there who think it is. Art is art. Design is about solving problems. You can and totally should automate as much of that as possible.

I've heard this anti-aesthetic argument before and I don't buy it. There's a reason companies like Stripe are lauded for their design efforts while Google comes out with material design...

If there's anything less scientific than the material design docs I want you to send it to me. Design hinges on aesthetics. You can't just ignore it (well you can, but you might end up with something that looks like material design).

Not GP, but where exactly was their argument anti-aesthetic? Art =/= aesthetics, Design =/= art. GP said: design trends are just manifestations of a particular temporal affinity towards one aesthetic or another which sounds reasonable to me, considering recent trends towards flat/ish designs

Maybe you'd like to address his/her points instead of creating strawmen?

It's only degrading to your profession if it's actually true.

I was a software developer for many years, and I've seen hundreds of claims for software to automate (in one guise or other) that job, replacing the need for coders.

None of these claims offended me. The vast majority have been comically useless. The rest have just helped remove some dull bit of grunt work out of the job, allowing me to focus on something more valuable.

Disagree. It's not degrading even if it is true. Automation has been displacing people for how long now?

I would argue it's hubris to think one brain could forever outperform the cumulative experience and information that we can automate.

That doesn't stop it being potentially degrading though.

When people, and even entire communities, define themselves by the job they do, they often end up feeling degraded and worthless as the jobs are replaced.

Doesn't mean that the automation shouldn't happen. But that sense of degradation is real for some people.

And then, we'll automate hubris.

I agree, and to amplify: while there are surely aspects of design that can be automated, the part that's difficult is that a design should actually mean something. Having an algorithmically designed logo tie the room together -- crystallize a company's business and culture, and provide a memorable rallying point -- that's the hard part. Computer-generated design is like a ghostly thread of expression of the design sensibilities of the writers of the software from which it emerges.

Consider the hidden arrow in the "Ex" of FedEx, when they rebranded from "Federal Express"; the rational propeller of BMW; even something as simple as the Burger King logo being sandwiched in a hamburger bun. Granted, logo design is just one small corner of design in general, but algorithmically generated logos do indeed result in a herd of swooshes, globes, and rings orbiting bold text (notwithstanding the similar designs of many humans who fail to recruit sufficient creativity or inspiration). Some of these can even be aesthetically pleasing, but it's so much better when a logo is retrospectively obvious and actually means something.

If someone is willing to use a free logo generator, they already don't care about design and would not be happy with any money they are spending on your services.

It's the same as a software engineer being offended that a free website generator exists.

Eh, Lucent for a while had a logo that was (nearly literally) a sample brush from Painter or something. Painty red circle. And many other logos are pretty much like that and can definitely be generated by algorithms. I'd be shocked if some large logos were picked with far less thought than you're describing here.

Also, offended? It's like photographers getting upset over people giving stuff away, or the influx of rather good amateurs killing their market.

I'm just checking out Tailor Brands, and I can say with certainty the output is similar to what I've seen professional designers offer, on occasion.

I sympathise with your perspective, but surely some of those needs (colour theory, contrast, etc) can be solved algorithmically and others (public perception and brand "feel") could be automatically tested by a site/script surveying users in the target segment.

You need someone creative to generate the good ideas to be used in the first place.

Random mutation and A/B would surely take millenia, even assuming huge traffic in the first place. And even then, one of the internal criticisms of A/B at Amazon was that it helped refine local optima (e.g. a next button in a checkout cart) but is weak on a holistic view (whole page, user experience), and making larger innovative jumps (which is where you end up getting swallowed by your competition of course).

Of course, you could just think of a seasoned creative as a really powerful software/hardware combo, with AI and heuristics light years ahead, and cheap at the price.

(My sympathy stems from me doing a lot of design work so I'm not approaching this from a "design is useless" aspect.)

Do you think that what's out there already could serve as a fairly good starting place? Run them past users to work out which out of thousands of logos are effective. Then isolate components (marks, text, etc) and test those, plus random combinations of the same.

I can appreciate the value in very high level brand creation, but I also think you can get 90% of the way there at 1% of the budget for many businesses. A remotely competent designer can slap down a shape and choose a font for the text and beat half the small business logos out there in a few minutes.

Not really, for two reasons:

1. You don't have the users to test yet.

2. This is what really makes it untenable - what is the logo for, what does effective mean, 90% of where? This is why brand creation is expensive as hell. There's a lot of work involved and by definition it is bespoke and without shortcuts. That's the whole point or it wouldn't be brand creation in any sense. When you see a logotype, it's just the tip of the iceberg.

I suspect that a lot of people might think of branding as including these cheap services where an underpaid young offshore photoshop whizz throws together an infinity symbol, dolphin, or generic swoosh and even goes so far as to hand kern some text (and bear in mind that even this kind of junior typographic task can't be automated yet).

I would argue that this isn't a ghetto version of the kind of branding that a large company does, but a cargo cult version. The logotype part is highly visible but is the tiniest component of branding - it is superficial and meaningless in isolation. Very much unlike this process, real branding is about discovering and generating an extremely high input of info and ideas and turning this into a very small and focused output, articulating the essence. (And of course a big part of the cost for a large org is that you often need to cover things like web, print, letterheads, reports, sides of trucks, and so on. All whilst maintaining brand consistency.)

Assuming you can't afford to pay for full branding, I think the best solution is to do just enough (e.g. many here might just need something for digital and business cards) and for it to be done by someone close to the company owner of even the owner themselves, as they will live and breathe what the company is for and what makes it distinct. Even poorly executed, this will communicate a much clearer story and have real value imo (trying to avoid the word authentic here but it's probably applicable).

"You can't just pick out some basic shapes and text for your logo. You need color theory, contrast analysis, a brand strategy, a well developed concept, etc."

Maybe let the entity decide for themselves what they want?

> Wouldn't you be offended if someone said a basic algorithm could eliminate your job?

Hey, some of my best friends are basic algorithms!

I am trying to think of a good word to encapsulate this concept.

Its the difference between Picasso's line drawings and a child's scribble, the difference between a maguro and otoro cut of tuna, the difference between a script kiddy's work and the sophistication of stuxnet. To a laymen they are the pretty much indistinguishable. But for those who have the experience to know the difference, suggesting the two are equivalent is fairly ridiculous.

I see two scenarios 1) If the logos themselves are as good as a real designer's work (I doubt it) great, I pick that logo. Sucks for designers but design is a much wider field than creating pretty logos.

2) If it isn't I can rapidly prototype my basic ideas and take them to a designer. This is excellent and I'd embrace this as a designer.

yes, that is what I thought too! But then again, gravatar does the same thing for users. While looking at the website I've found their lead-qualification tool for salesforce. That's much more interesting.

Killer work Alex! It's crazy you just rolled this out. We have been in development on something very similar but have yet to make a public push on the product. The main difference is we host the vector source file and build rasters from it. This helps maintain top quality at any size, and allows us to output to additional formats (i.e. PDF).

The product is still in alpha, but it's amazing how many similarities we came to with the URL scheme design. For example, image embedding:

https://img.ogol.io/<domain.com> example: https://img.ogol.io/ogol.io

we also support downloading

https://dl.ogol.io/<domain.com> example: https://dl.ogol.io/ogol.io

Each logo also has it's own page to make working with the asset outside of an API easy.


Our approach requires companies to confirm their domains and associate a vector logo with the domain. Your strategy obviously provides a lot of logos right out of the gates. Logos are such a pain to deal with, it's great to see the problem being attacked from a few different angles.

I'm curious about the claim "pixel perfect every time" seen on your home page. Do you employ some special algorithm to scale images, or do you just use the standard tools? As far as I know, scaling vector graphics isn't enough to guarantee pixel perfection (especially not at low resolutions), and low-res logos are usually hand-crafted by designers. Maybe you should also allow companies to upload raster logos?

ogol.io is a pretty awesome idea! Love it.

Instead of relying on FB, Twitter, Company's site for the logo, wouldn't it be better to create a "Company Logo Service" with an API, with the below features -

0. Change at one place, make it work everywhere. (consistent branding across the wild web).

1. Upload your company logo here with us.

2. Multiple versions.

3. Multiple sizes.

4. Control and Connect various sizes, versions with your various social accounts, newsletters, anything.

5. IFTTT support.

6. Get it printed on swag and merchandise.

7. More...?

Seems like a good way to waste money on a service that will very seldom provide value (logo redesigns) and will not work across all platforms thereby introducing an additional thing you have to worry about instead of solving it.

Better idea:

1. Take sheet of paper

2. Write down where you use your logo

Biggest fan of working on paper, to the extent of carrying a clipboard in my laptop bag all the time. Along with other gadgets.

I wouldn't pay for something like that either. I was suggesting it be offered for free, like gravtaar.

Wouldn't pay money for this. Easier to just have a list of sites where we maintain profiles since we still have to manage that various bits of other information about the company, beyond just the logo (which is probably the thing that changes the least).

Absolutely. Would pay money for this!

The question is: how much?

And when you use it, they get to collect all of the data about your visitors.

In what way?

Nothing about this API requires handing over more than the bare minimum of information (which domains you want logos for). How could they be expected to implement it without that information?

There's also the referrer header that gets sent out, which includes the URL of the page that is embedding the image. Not really "all data on your visitors" but depending on how you use it, may leak sensitive information if you're unwise enough to put it in the page urls (depressingly common), or allow them to track every visit throughout your site (on pages that embed the image)

There's no reason you have to directly embed the images.

If you want to preserve user privacy, you should definitely proxy this API.

I don't know what clearbit collects, but I would guess that this isn't the business model. Clearbit sells a B2B product - they monetize by charging money for a service directly, not indirectly via "collecting data about your visitors".

This guy knows what is up.

It's outdated. They need a version or date parameter.

This has been done before (and posted to HN), and it had the same problem. Way too many false positives, especially with older sites.


This isn't trying to scrape the page and find something vaguely logo-like. Rather, it seems to be using iOS, OpenGraph and Twitter icons. In this case, the site has a strange OpenGraph image specified in its meta tags. If it produces a weird image it's because the site owner has specified an image which isn't their logo. It's not because the API picked a random image.

For this service, if it can find an image, it's at least an image the site owner wanted to be used to represent it.

Yeah, really odd choice for an Open Graph image. They specify that burger and soda image directly in their header.

<meta property="og:image" content="//nebula.wsimg.com/7cdaeb9fbb87f3450dc65108e6d6af87?AccessKeyId=6E4E9B5CC8D37F66F76C&disposition=0&alloworigin=1">

I wouldn't have thought that the open graph image is a logo. Seems like that wouldn't be a reliable source for logos

Fair point. It's at least something they've chosen to represent themselves, however.

Interesting results with Blizzard.com: https://logo.clearbit.com/blizzard.com?size=500

Another (big) false positive is craigslist: https://logo.clearbit.com/craigslist.org

or (first semi-random one I tried): https://logo.clearbit.com/sabre.com - totally absurd

Hey Alex,

This is pretty sweet. Nice job! Is there a post where you talk about the tech behind this API? I've been working on a simple API that finds domain names from company names that I use on projects where I use business intelligence APIs like Clearbit and FullContact.

Behind the scenes we're using Clearbit's Company API which does a lot of sleuthing: scanning the page for social accounts, meta tags etc. Then we try and return the best logo we can find.


Fetch the domain at /, look for favicon/apple-touch-icon/OpenGraph image, fetch the image, store and serve.

Looks like it has twitter as a secondary source, eg.

https://logo.clearbit.com/spacex.com https://twitter.com/spacex

Maybe other fallbacks, too?

Also seems like some fairly primitive analysis would identify this as a suspect item, at least enough to be shunted off to some sort of triage. Logos, by and large, will have distinctly different data than images.

That, and it also doesn't match their favicon.

Good catch, that's pretty hard to deny

>> I use business intelligence APIs like Clearbit and FullContact.

When I saw the tagline "Business Intelligence APIs" at the top of the posted link, I totally thought Clearbit's line of business was something completely different-- more along the lines of data warehousing, analytics and reporting.

I wish Clearbit had a "very early stage" plan - say 1000 requests a month, which I could use with proof of concept or little side project systems. 50 free requests isn't much: by the time you've demo'd to 5 people for feedback they'd be gone.

There has to be some level of fair use or copyright involved, no? You are taking corporate images, modifying them, and then distributing their modified brand/logo potentially without their consent. If they don't like the results, they could sue you.

Or at least, I feel that would be the case.

'Fair use' only applies to the copyright implications of distributing these images; the far murkier bit of IP law that this gets into is trademark licensing. That's far more likely to be the issue here.

Not really modifying it. But they may have branding and licensing guidelines that a service like this might not meet which would likely result in a cease and desist rather than an immediate lawsuit.

I tested a few examples, and it looks to just be parsing the page and returning the apple-touch-icon.

That doesn't explain where https://logo.clearbit.com/debian.org comes from....

https://www.debian.org/favicon.ico can't be the source.

It looks like that's the Debian icon on Twitter.

As a developer of a URL shortening service, this is incredibly useful. Social platforms scrape the contents of all shared URLs to offer a preview in a feed. With this service I can incorporate the logo of a site's destination URL into the scraped content.

Yet another (hilarious?) false match: https://logo.clearbit.com/exxon.com

That's because Exxon specified that image. Check the HTML on exxon.com:

  <meta property="og:image" content="http://www.exxon.com/assets/imgs/en-us/energy-live-here-facebook.jpg">

That doesn't change the fact that it's a false positive…

I'd be cool if they would provide some insight on how to make an existing website to work with their API. I always wanted to know what's the "standard" way to put a logo in the webpage. Some people uses just an IMG tag, other uses an H1 with an image replacement technique.

Looks like they're just returning the apple-touch-icon if it's available.

In my case they're returning an old version of the icon.

Same here. Any way to trigger a refresh?

Worked on something similar on a local open data hackathon before, but instead I used a scraper to parse the logos.


It worked okay for like 40% of the sites, and for the rest of the sites we used Python and scikit-learn to detect the logo from the page (threw all images of the page in the script and it returned if it was a logo or not). And this actually worked quite good, irrc over 90% of the test cases worked.


But yeah, using Twitter as a source might also be a good idea.

Pretty cool! I think the API docs could possibly be a bit more precise, though. Perhaps something like this:

  You can also pass us the following optional query parameters.
  Parameter 	Default 	Description
  size 		128 	 	integer	Image size: Length of longest side, in pixels
  format 	"png"	 	string	File format, either "png" or "jpg"
  greyscale 	(not passed) 	boolean	If this parameter is passed, image will be desaturated

Good idea - we'll make this change.

Thinking perhaps we have something set up odd in our HTML that's tripping up the API, but a bit confused why this seems to be returning nothing https://logo.clearbit.com/korkapp.com

Any thoughts anyone?

This is an excellent piece of marketing.

I know Glassdoor is using Facebook as a source of their company logos

I'd like to see that open sourced. A lot of the projects I'm involved in could use it.

How would a logo image file have to be named to be found by this?

This vaguely reminds me of Picons, which was the '90s solution to this problem (Gravatar for domains, but also for newsgroups and people):


If you've ever noticed the little logos on Gmane posts, they come from Picons and favicons.

Works great! But I do wonder where it came from... parsed the HTML, grabbed the SVG and converted it to PNG... or simply got it from Twitter?

https://logo.clearbit.com/h5mag.com from... <img class="logo" src="/img/h5mag-logo.svg" alt="logo H5mag">

If you're using Twitter as a source for company logos, you might be serving a lot of rainbow icons this week!

Late to the party, but this looks interesting.

Google does something similar for extracting favicons for any domain, such as https://plus.google.com/_/favicon?domain=github.com

I noticed that you are returning the company name in the title. https://logo.clearbit.com/fb.com .Is that intentional? Asking this because you provide a Company API too which is paid.

This API call is just returning a PNG image with almost no EXIF data. In that case, the title is set by browsers to the filename, which is equal to the domain name itself. Are you seeing something else in the title? (I see fb.com)

Saw this in response headers.

Content-Disposition:"inline; filename="Facebook.png""

I am using firefox which apparently uses this header to set the title. Chrome OTOH won't do it

Interesting. I did look at headers, but missed the Content-Disposition for some reason.

Strange. I see Facebook in the title

You need a better upscaling algorithm.


Ideally all your logos should be in vector and rendered to any size (or at least powers of 2 for easier caching).

Maccman, Logaster offers own API which will generate logos. If you interested read more here https://www.logaster.com/about-logaster/api/

Hi. Nice idea. Images seem to be scaled not optimally though in same cases seeing jagged edges like here: https://logo.clearbit.com/medigo.com

Wish there was an equivalent service but with masked images of wine bottles. Always wanted to create something like Delicious Monster but for wine cellars. All the sites/apps doing it are pretty ugly or largely text based.

Are there terms of service associated with this API? I can't seem to find any.

Great API, thank you!

I made a little hubot script for this: https://www.npmjs.com/package/hubot-logo

Use it like this: hubot logo stripe.com

https://logo.clearbit.com/armadilloaerospace.com Where is the armadillo? I want my armadillo.

Doesn't work, though https://logo.clearbit.com/ycombinator.com does.

That's pretty dang cool. Might cause issues if clearbit ever looses the data and a bunch of websites link to it, but that's pretty standard.

This is neat - it could use a background color option along with the ability to fill the logo with white.

Doesn't work for my domain: zk.gd

Nice job. It would also be great if there is option to get image data that can be used in data-uri.

The trend is to have a different logo for apps and websites. another parameter might be helpful.

Where do companies like clearbit and full contact get their user data?

Isn't that what the favicon.ico in the root of your site for?

Works great! Minor nitpick: Returns the old logo for Oculus.com

Brilliant, this is helpful for building MVP products as well!

How so?

Internal error for https://logo.clearbit.com/taodyne :-( We recently chàged server, that might be why.

Seems to work fine if you put in a domain instead of just a name:


This is great I will have to remember it.

Great work! Very useful :)

Nice, seems super helpful.

seems a simple change to gravatar to allow registration and use of simply the domain name to host the Corp Logo would be a 10 minute change.

great!!!! really really great!

Awesome! Have wanted this for so long!


Well done!

Just curious.. what do you guys use for API management to handle authentications, logging, rate-limiting, etc..

1. Open-source solutions like KONG (https://github.com/mashape/kong) or similar?

2. Built in house?

3. Commercial services such as Apigee, Mashery?

Surely there's a trademark protection concern here.

Nope, trademarks are used to distinguish brands and ensure their authenticity, and that's exactly what is happening here. If people used this to impersonate another company (e.g. to pretend they had gotten their endorsement) then that's a problem. But as long as this is used as part of a directory (or similar) service there is no copyright or trademark problem.

arent there legal requirements about colors, sizes, and placement for logos? for instance you offer gray scale as a setting, i seem to recall all placement of MS logos requires approval and no altering of colors(i could be completely wrong)... are you protecting your users by disallowing them to use logos in ways that might upset the owner?

Great idea, but something tells me the companies whose logos are available won't like it. It will be interesting to see how they react. Some will probably embrace it, but I can see some of the larger corps acting territorial.

I would argue opt-in for having your logo with the service, rather than opt-out, but it might be difficult to come up with a benefit that's enticing enough.

There's maybe some value in the logo owner being able to get information on who's using their logo (service not yet provided).

From the logo-consumer standpoint, it's pretty cool, though, it would be difficult to know if the logo is current enough.

Be careful with copyright issues. How do you make it work? Do you scrape them automatically or gather them from the brands marketing materials?

I tried different logos, and I find a few issues:


Mcdonalds US logo background should be red (http://www.aboutmcdonalds.com/mcd/newsroom/image_and_video_l... )


The quality of that one is bad. Wikipedia's one is a .svg : https://en.wikipedia.org/wiki/File:Burger_King_Logo.svg

I don't think they'll have copyright issues, it'd probably be considered fair use.

As for the McDonalds logo, I suspect the turquoise is because that's the current background colour of mcdonalds.com

Or because it's from here: https://twitter.com/mcdonalds

Ah, that makes more sense. It was mentioned below that it seems to be using Twitter.

I don't see any red backgrounds in your link.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact