Which Logo File Format to Use for Screen, Print, and Client Delivery

You have opened the wrong file for the third time this week, and the client is waiting. The logo file formats explained here will end that hunt. The rule is simple and non-negotiable: a logo must exist in a vector format for any production use. A raster-only logo is a source-file failure, not a design asset. You do not hand a client a JPEG and call it done. You hand them a set of files, each tuned for a specific job, and you know exactly which one to reach for when the printer, the web developer, or the sign maker calls. Here is which file format to deliver for each production use and why, with the vocabulary to speak the specification language without being sold mystique.

Every professional you work with, the print estimator, the front-end developer, the large-format sign shop, expects a logo that scales. Vector graphics are resolution-independent because they are described by mathematical paths, not by a fixed grid of pixels. A raster image is a grid of coloured squares; when you enlarge it beyond its native size, you see the squares. A logo delivered only as a PNG or a JPEG will fail the moment someone needs it bigger than the file was made. The 72-ppi logical resolution for screen design is fine for a screen, but offset print demands a minimum effective resolution of 300 ppi at final size. If your only master is a 1000-pixel-wide PNG, you cannot print a business card or a billboard without interpolation artefacts. The fix is to start from a vector master in SVG, EPS, or AI, and export every other format from that master. If the only file you possess is a raster, your logo does not exist yet; you must have it redrawn or recreated as a vector path before any production can begin.

Why SVG Is the Screen Default

For anything that lives on a screen, SVG is the default. Scalable Vector Graphics is an XML-based format, which means it is a text file that describes shapes, paths, and text. It is resolution-independent. A single SVG file serves a 16-pixel favicon and a 2000-pixel hero image without re-exporting. It supports native alpha channel support, so a logo for a dark background can have a transparent background with no white box. The file size is small for simple graphics; a logo with a few paths and no photographic content will often compress to under 5 kilobytes. That matters for page weight and loading speed. For animation, SVG is the only vector web format that animates in the browser without a plugin. You can edit it in a text editor, but you should not need to; you edit the master in Illustrator or Inkscape and export the SVG from there.

Minimum Export Settings and Accessibility

The minimum viable export setting for a logo is to convert text to outlines, simplify paths by removing anchor points that do not change the shape, and remove unused defs so the file carries no dead weight. SVG accessibility is a real feature: it supports title and desc elements that a screen reader can announce. A screen-reader-accessible logo is possible, unlike a PNG where the alt text exists only in the surrounding HTML, never in the file itself.

Encapsulated PostScript used to be the standard for vector exchange, but its status is effectively deprecated by Adobe since the 2010s, and you should treat it as a legacy format. That does not mean you never use it. Sign-making software and some large-format vinyl cutters still require an EPS with outlined vector paths. The EPS file type is vector with an optional embedded raster preview, typically an 8-bit TIFF at 72 dpi, which is what you see when you place the EPS into a layout without opening it. EPS does not natively support transparency in PostScript; if you have a drop shadow or a blend mode, the file will show a white box or require flattening, which destroys the transparency. The colour model depends on the creation application and can be CMYK, RGB, or spot colours. An EPS exported from Illustrator with a Pantone reference will carry that spot colour definition. For a logo delivered as an EPS, convert text to outlines. The recipient's RIP or sign software will not have your fonts installed. The minimum viable export setting is outlines, an 8-bit preview, and a colour mode that matches the intended output: CMYK for process print or Pantone for spot colour. EPS is larger than SVG for equivalent vector data because the PostScript language is verbose, but that does not matter when a sign shop specifically asks for it.

Why PDF/X-4 Is the Print Standard

When you must send a logo to a commercial printer, PDF/X-4 is the conformance you need. PDF/X-4 is ISO 15930-7, a print-exchange format that supports live transparency and ICC colour management. Your drop shadows and gradient blends do not need to be flattened before the RIP processes them. Most commercial print providers require PDF/X-4 for colour-managed workflows. It embeds the ICC profile that tells the output device how to interpret the colour. The PDF file type can be vector, raster, or mixed, but for a logo it should be vector. The PDF colour model can be RGB, CMYK, spot, or Lab, but for print you should embed the ICC profile the printer requests, typically a coated or uncoated FOGRA or GRACoL profile. You must include a trim box that defines the final page size.

Client Delivery and Accessibility

For client delivery, a PDF is the best master. It is resolution-independent, preserves live text if you embed the fonts, and opens in any PDF reader. The minimum viable export setting for a print PDF is PDF/X-4, convert text to outlines if you do not want to embed fonts, and include the trim box. If you keep live text, you must embed all fonts; this is where font embedding permissions come in, and you will read about that in the next section. A PDF is also the right format for multi-page brand guidelines. You can put the logo, clearspace rules, and colour codes in a single file. PDF accessibility exists via tagging (PDF/UA, ISO 14289-1:2014). A tagged PDF with logical structure will read in a screen reader, which is not true of a flattened image-only file.

There are still places that do not take SVG. Email clients, some presentation software, and older content management systems will not render vector graphics, and that is where PNG comes in. PNG is a raster format with lossless compression and native alpha channel support. You get a transparent background. The colour depth is 24-bit, or 32-bit with an alpha channel, which means 16.7 million colours plus transparency levels, enough for any logo that is not photographic. The PNG file size is larger than JPEG for photographic content but efficient for flat-colour graphics. A logo with two or three colours will compress well. For a digital logo, export at 72 PPI for screen and export at 2x and 1x resolutions. A Retina display needs a 2x file to appear sharp. The minimum viable export setting is 32-bit colour with transparency, and you must generate both 1x and 2x sizes from the vector master. PNG is not a working format; you cannot edit text or paths in a PNG. Text is rasterised and there is no font data. The editability of PNG is limited to a raster editor like Photoshop or GIMP, which is fine for compositing but not for changing the logo's shape. If a client asks for a logo for a Word document or an email signature, a PNG on a transparent background is the correct fallback when SVG will not render.

Run every logo file through the favicon test before you call it done. A favicon is displayed at 16×16 pixels, and most of the detail in a complex logo will be indistinguishable at that size. If the logo does not read at 16 pixels, it will not read on a browser tab. If it does not read in one colour, it will fail when a client needs a single-colour version for a foil stamp or an engraving. The favicon test forces you to have a simplified version of the logo as part of the delivery set. This is not optional. A logo with a gradient or a drop shadow may look good on a website, but at 16 pixels the gradient is gone and the shadow is a grey blur. You need a one-colour version, typically black for light backgrounds and white for dark, with the vector paths simplified to the essential shapes. The one-colour version is also what you hand to a sign maker who cuts vinyl or a printer who does a spot colour job with a single Pantone ink. A logo that cannot survive the favicon test is a logo that has never been properly designed for production. The vector master should be structured so that a simplified monochrome mark can be derived from it without redrawing.

The fsType Bit and What It Controls

When you deliver a PDF or EPS with live text, you rely on font embedding permissions, and those permissions are not uniform. The OpenType font format encodes embedding permissions in the OS/2 table, specifically in the fsType bit. The fsType bit determines whether a font may be embedded at all, whether it may be embedded for preview and printing only, or whether it may be embedded in a document that the recipient can edit. A font with the fsType bit set to 0 is freely embeddable; a font with the bit set to 2 is embeddable for preview only. The PDF will open on screen but the text cannot be extracted or edited. If you send a client a PDF with live text and the font's fsType bit forbids editing, the client will not be able to change a spelling error, and they will blame you.

SIL Open Font License and the Outlining Trade-Off

The failure mode is worse with the SIL Open Font License, which is permissive in that it allows use, embedding, and modification. The copyleft condition is on derivative font files, not on documents. A font under the SIL Open Font License can be embedded in a PDF, but the license does not permit you to sell the font file itself or create a derivative font without opening that derivative under the same license. For logo delivery, the safest practice is to convert text to outlines so that no font data is embedded at all. Outlining a font converts each glyph into a vector path. That makes the text uneditable but also makes the logo independent of font availability. The trade-off is that outlining destroys text editability: a client cannot fix a typo in the logo without the original source file. So you deliver the outlined PDF for print, but you also keep the original AI or SVG with live text as the working master. Make sure the client knows which is which. If you keep live text in a PDF, you must embed all fonts, and you must check the fsType bit of each font before you embed it.

The RIP Failure and Preflight

A font that allows embedding for preview and printing will generate a PDF that opens on screen but fails at the printer's RIP if the RIP tries to substitute or subset the font. The print-specific failure is that the printer's RIP replaces a missing font with a default face, altering line endings and page count. That is exactly why preflight is mandatory. Preflight is the automated checking of a document against a profile of production rules before output. It is the only way to catch a font that will not embed before you send the file.

The three formats that dominate vector logo delivery are SVG, EPS, and PDF. Choosing between them depends on where the logo is going, not on which is newest. SVG is the web-native choice. It is XML-based and renders in every modern browser without a plugin. EPS is the legacy choice, a format that sign-making software still requires but that Adobe has effectively deprecated as an exchange format. PDF is the print-production and client-delivery choice. It can carry ICC profiles, trim boxes, and live or outlined text. For a client who is not a designer, a PDF is the least likely to be misused. It opens in any reader and is obviously final. For a developer, SVG is the only one of the three that can be manipulated with CSS and JavaScript for animation or theming. The EPS file type is vector with an optional embedded raster preview, but the preview is what you see when you place it into a layout; the vector paths are what the sign plotter cuts. The PDF file type is vector, raster, or mixed, but for logo delivery it should be vector-only, with no embedded photographs. The minimum viable export settings differ. For SVG: convert text to outlines and simplify paths. For EPS: convert text to outlines and include an 8-bit TIFF preview. For PDF: use PDF/X-4 and convert text to outlines if you do not want to embed fonts. There is no scenario in this comparison where the EPS wins a new project; it wins only when the downstream software demands it. All three are resolution-independent for vector content, so scalability is a given. The question is what the recipient will do with the file. If the recipient is a sign maker who has used the same vinyl cutting software for fifteen years, EPS is the safe bet. If the recipient is a web developer, SVG is the correct choice. If the recipient is a print shop, PDF/X-4 is what they will ask for.

The distinction between raster and vector is the entire basis of this guide, and it is worth stating plainly: a raster image is a grid of pixels, a vector image is a set of mathematical paths. For a logo, the vector is the source of truth and the raster is an export for a specific use. The failure mode is treating a raster as a master and upsampling it to fit a larger space, which produces a soft, pixelated result. A PNG is a raster format, and it is the right deliverable for social media, email signatures, and any digital platform that rejects SVG. The PNG export resolution standard for digital logos is 72 PPI for screen, and you should export at 2x and 1x so a retina display is sharp. But a PNG cannot be enlarged beyond its native pixel dimensions without quality loss, so the vector master must be the source. In print, a raster logo will work if it has enough pixels per inch, but it is a fallback, not a design choice. The one thing a raster has that a vector lacks is a photographic texture; a logo with a photo inside cannot be vectorised without tracing it. Tracing a photo produces many small paths that bloat the file. So when the brand includes a photographic element, the delivered set includes both the vector paths and a raster export of the photo placed on a transparent background. The vector remains the master.

When the client calls at 1 a.m. because the website is being rebuilt tomorrow, you need a delivery set that answers every possible use without a phone call. The set is four files: an SVG for web, a PDF/X-4 for print, a PNG at 2x for digital fallback, and an original vector master in AI or SVG with live text. Anything less and you will be the one explaining why the logo does not work. This is the failure case: the client has a trade show banner due in the morning, the sign maker needs a PDF with bleed, and the only file the client can find is a 72-dpi PNG from the email signature. You hand over the set once, and you add a README file that says which format is for which use. The PDF goes to the printer. The SVG goes to the web developer. The PNG goes into the email. You do not send a JPEG. A JPEG has no transparency and will show a white box on a coloured background. You do not send an EPS unless the sign maker specifically asks, because the sign maker's software may be the only one that can open it. You do not send a TIFF. A TIFF is a raster format with a huge file size and no vector paths. The name of the game is to make the right choice the obvious choice. This guide is for the professional who commissions or implements design and needs to speak the craft's specification language without being sold mystique. It is not for the person who wants colour psychology as a primary decision-making framework; colour is addressed only where evidence exists, such as legibility and accessibility research.

The cost of a logo file set is not the file itself; it is the time you will waste when the wrong format reaches the wrong destination. A commercial printer will often accept a PDF without a preflight check, but they will also charge a re-run fee if the file fails at the RIP. Preflight is the automated checking of a document against a profile of production rules before output. It is the difference between a one-hour proof and a three-day cycle of rejected files. The preflight checks for bleed, the trim box, the embedded fonts, the colour mode, and the resolution of any raster images. If you skip preflight, you are gambling that the printer will not find a problem. The practical detail is this: a PDF/X-4 file with the correct ICC profile, outlined text, and a trim box will print correctly the first time. An EPS from the same design software may have a different colour mode or a missing preview, and the transparent objects may not flatten the way you expect. You do not need to know how to fix every preflight error; you need to know enough to see the error report and ask the printer which setting to change. The timing is also practical: a print job with a hard deadline cannot wait for a font-substitution issue to be resolved at the press. The failure mode of font substitution alters line endings and page count, and the only way to prevent it is to outline the text or embed the fonts with the correct fsType bit. You can spend the extra hour outlining the text, or you can spend it explaining to the client why the business cards say the wrong phone number.

Stop optimising for the hero image and start with the favicon. A logo that works at 16×16 pixels in one colour is a logo that works everywhere else. This test is not about the file format; it is about the design's resilience. If the logo is a wordmark with a long name, it will not be legible at 16 pixels. You need a monogram version or a simplified icon. If the logo uses a gradient, the gradient will be invisible at 16 pixels. You need a flat-colour version. The favicon test is what separates a brand system from a decorative painting. You deliver the full-colour logo for the website, the one-colour version for the engraving, and the favicon for the browser tab. The favicon is a PNG, usually 32×32 or 16×16, and it is exported from the vector master. The test is not optional: hold your logo at arm's length, squint, and ask if you can identify it. If you cannot, the logo is too complex, and no amount of file format selection will fix it. The one-colour version is also what you hand to a foil stamper or a die-cutter, who cannot reproduce a gradient and will not try. A logo that fails the favicon test is the reason a client ends up with a cheap-looking business card. The details that look great on a 27-inch monitor disappear at the physical size of a postage stamp.

For a logo on a modern website, SVG wins on every technical axis except one: raster fallback. An SVG file is a text file, so it compresses well and can be cached by the browser. A PNG is a binary file that does not compress as well, and a 2x PNG for a retina display will be four times the size of a 1x PNG. The file weight, kilobytes, is lower is better for web delivery, but only up to the point where compression artefacts become visible. A hero image optimised to 40 KB may look worse than one at 120 KB with perceptually tuned compression, but a logo is a flat graphic. The PNG at 2x will be a few tens of kilobytes, which is trivial. The real difference is in the rendering. An SVG scales with the viewport, so the same file serves a 16-pixel favicon and a 2000-pixel hero. A PNG is fixed at the resolution you exported it, so you must export multiple sizes. The colour gamut also differs: SVG uses RGB with CSS colour management, while PNG is RGB or greyscale. For a brand colour that must match a Pantone or an sRGB value, the SVG can carry the CSS custom property that references the exact hex value. The PNG stores the pixel values directly. The performance argument is not about the file size of the logo itself but about the number of requests: an SVG inline in the HTML saves an HTTP request. A PNG requires a separate request. On a slow mobile connection, that is the difference between a logo that appears immediately and one that appears after the rest of the page has loaded. So the rule is: use SVG by default, and keep a PNG 2x in the delivery set for the email client or the old CMS that strips SVG.

A printer will accept a standard PDF, but they will not guarantee the colour. The PDF/X-4 conformance removes the guesswork. It is ISO 15930-7, a print-exchange format that supports live transparency and ICC colour management. When you export a PDF/X-4 from InDesign, you must choose an output intent, which is the ICC profile that defines the printing condition. The printer uses this profile to convert the document's colour space to their device. If you skip the profile and send a standard PDF, the printer's RIP will assume a default profile, which may shift your brand colour. The failure mode is the claimed versus real print colour match: the screen proof shows the intended colour, but the printed piece does not match. The file lacked an ICC profile and the RIP interpreted the RGB values under D50 lighting. The D50 lighting is specified by ISO 3664, and it is the only lighting under which printed colour must be evaluated. A passing ratio on paper may fail on a dimmed mobile screen in sunlight, which is why a hard proof is better than a screen proof. The practical rule is to ask the printer which PDF/X version they prefer and which ICC profile they recommend for the paper stock. Most will tell you, and the answer is almost always PDF/X-4. If you send an EPS, the printer can open it, but the EPS does not support live transparency. Any drop shadow will be flattened into a raster block that may print as a white rectangle. Transparency flattening artefacts are a leading cause of rejected print files. The solution is to use a format that supports live transparency natively, which is PDF since version 1.4 (2001). The PDF/X-4 standard is built on that transparency support. It is the only format you should consider for a colour-managed print workflow.

The following table is the entire guide condensed into a single reference. Use it when you are standing at the export dialog and you have forgotten which format is for what. There is no other table here because no other comparison is as clear-cut as the four-way split below.

Format	Use Case	Colour Model	Transparency	Scalability	Editability
SVG	Web, responsive, animation, favicon	RGB (CSS colour management)	Native alpha channel	Resolution-independent	Text editor or Illustrator/Inkscape
EPS	Legacy print, sign-making software	CMYK, RGB, or spot	Not natively supported	Resolution-independent for vector content	Illustrator (open but not save as EPS fully editably)
PDF	Print production, client delivery, brand guidelines	RGB, CMYK, spot, Lab, ICC-based	Native support since PDF 1.4	Resolution-independent for vector content	InDesign, Acrobat; live text if fonts embedded
PNG	Digital fallback, social media, email	RGB, greyscale	Native alpha channel	Resolution-dependent	Raster editor only (Photoshop, GIMP)

What is the difference between a font and a typeface?

A typeface is the design of the letterforms; a font is the file or physical set that delivers that design in a specific weight and style. Times New Roman Bold is a font; Times New Roman is the typeface. You install a font, but you choose a typeface.

Is 72 PPI enough for a logo in print?

No. 72 PPI is the logical resolution for screen design. Offset print requires a minimum effective resolution of 300 ppi at final size. If you only have a 72-ppi PNG, it will be pixelated in print. This is why the vector master is non-negotiable.

Why does my PDF show a different colour than my screen?

Your screen shows RGB values converted to your monitor's gamut. The PDF may contain an ICC profile that your monitor ignores. The printer uses that profile to convert to CMYK. Soft-proof with the correct ICC profile under D50 lighting to see the printed colour before you send it.

Can I edit text in a PDF with live fonts?

Only if the font's fsType bit allows editing. Many fonts set editability to 'preview and print only'. If you need the client to edit text, outline the text or deliver the original source file. Outlining destroys text editability, so keep the source as the working master.

A spot colour is a pre-mixed single ink, and the Pantone Matching System is the physical reference standard where colour is specified by swatch, not by screen value. A process colour builds hues from CMYK dot screens. For a logo, the difference matters most in print. If your brand colour is a specific Pantone, the printer must use that spot colour ink, and the PDF must carry a spot colour separation. If your PDF has the Pantone colour converted to CMYK, the print will not match the swatch. The CMYK gamut is smaller than the spot colour gamut. The practical rule is to keep the spot colour definition in the vector file, not to convert it. The ICC profile for an RGB PNG is sRGB IEC61966-2.1, and for a CMYK PDF the printer will request a specific profile, often a coated FOGRA or GRACoL. The colour gamut of a monitor is typically sRGB, which is smaller than the Adobe RGB gamut. Sending an Adobe RGB image to an sRGB-only platform will display desaturated. The same logic applies to print: if you specify a spot colour and then convert it to process, you will get a close but not exact match. The only way to guarantee the colour is to use the Pantone ink. The name of the Pantone library in Adobe software is Pantone+ Solid Coated or Uncoated. Since 2022 Adobe no longer bundles these libraries without a separate Pantone Connect license. This is a failure mode: a designer who relied on the preloaded books now sees a black colour swatch, and the logo file contains a spot colour reference that no longer exists in the software's picker.

Print layout outside North America uses the ISO 216 A-series, and the dimensions are in millimetres, not inches. A pdf page size maximum in Adobe Acrobat is 15,000,000 by 15,000,000 inches (381 km), which is absurdly large, but the practical limit is the printer's imposition software. The Ghent Workgroup is an international print-production body. Its relevance here is not its guidelines but the recurring failure mode: its export limitations and RGB-only colour model in certain workflows create print errors. The D50 lighting under which printed colour is evaluated is specified by ISO 3664, and it is not the same as the lighting in your office. A passing ratio on paper may fail on a dimmed mobile screen in sunlight, which is why the print industry uses a standard viewing booth. The value that matters is the printer-specific one: ask for the printer's specification document. Each printer publishes one, and it changes with their equipment upgrades. The bleeds, the trim box, the slug area, and the resolution of embedded raster images are all defined in that document. A printer that wants PDF/X-4 will say so, and a printer that wants a different format will say that instead. The failure mode is assuming the printer's requirements are universal; they are not. The Ghent Workgroup exists precisely because the absence of a shared standard caused so many print failures.

When you choose a font for a logo that must be localised, the glyph count is a practical constraint. A font with 400 glyphs may cover Latin Extended, which includes the accented characters used in Central and Eastern European languages. A font with 200 glyphs may fail on common diacritics like the Czech caron or the Polish ogonek. The underlying code points are stable, but the glyphs that render them may be missing. A design system that hardcodes a character by glyph appearance rather than by code point is fragile, because the glyph is a rendering, not the character. For a logo, this matters when the brand name includes an accent, a ligature, or a non-Latin script. If the font does not include the glyph, the RIP will substitute a default font, altering the line endings and the page count. The fix is to outline the text in the logo, which converts the glyphs to vector paths and removes the dependency on the font. The trade-off is that outlining makes the text uneditable, so you keep a separate working file with live text. The glyph count is also a factor for variable fonts, where a single file contains multiple axes of variation. The claimed efficiency is that one variable font file replaces a static family, but the real file is larger than the sum of the two weights actually used on the site. The unused axis data is dead weight. The efficiency claim assumes full-axis utilisation, which is rare in production. This is a caution: a font that is perfect for a long-form article may be wrong for a logo, where the priority is a consistent appearance across every production use.

Every delivery set should be accessible and failure-proof. Accessibility starts with the file format: SVG supports title and desc elements for screen-reader accessibility, and PDF supports tagging for PDF/UA conformance. For a PNG, the alt text is applied in the context of use, in the HTML or on the social media platform, not embedded in the file. The failure mode is a logo that is invisible to a screen reader because the alt text says 'logo' and the SVG has no title. The fix is to write a descriptive title for the SVG and a meaningful alt attribute for the PNG. The failure-proof part is the README file that names each format and the use it supports. The README is not optional, because a client will not remember which file is which at the point of need. The page count in print is a separate issue: a saddle-stitched booklet must have a page count divisible by 4, but a logo delivery set is not a booklet. The only rule that matters is that the vector master is the source of truth, and every other file is an export. When the client asks for a logo for a favicon and you give them the SVG, they will try to upload it to a platform that rejects SVG. The failure mode is not the file format; it is the absence of a PNG 2x in the set. You cannot predict every platform, but you can predict the 1 a.m. call. The rule is to deliver a set, not a single file, and to name the files so that the intended use is obvious. The name of the game is to make the right choice the obvious choice, and the README is the mechanism.

A logo file format is not a preference; it is a decision rule that prevents the 1 a.m. call. The call is not the file format but the absence of a decision rule.