Email Signature Best Practices: What Belongs, What Breaks, and Whether Your Face Goes In It

Most email signature best practices reduce to one rule: the signature has to work when every image is blocked. Decide the photo question by context — client-facing roles wear a face well, internal engineering roles usually don't — then build the block as plain text that survives blocking, threading and a screen reader.

The contested part: does your face belong in your signature?

This is the only genuinely arguable question in the whole subject, and it does not have a universal answer. A portrait in a signature is completely unremarkable in some fields and quietly embarrassing in others, and the difference is not about taste. It is about whether your recipients are being asked to recognise you or to read you.

Think about what a signature photo actually does. It converts an email from a message into a small piece of identity: it says here is the person behind this address. That is valuable when the recipient will eventually meet you, choose between you and three other people doing the same job, or needs to pick your face out of a crowded inbox they did not opt into. It is noise when the recipient already knows exactly who you are and just wants the answer to their question.

Where a photo is standard, not self-promotional

In these fields, a signature photo is so common that its absence is the thing people notice. The shared trait is that the sender is, in part, the product — the recipient is evaluating a person, not just a deliverable.

• Real estate. The agent's face is already on the yard sign, the portal listing and the business card; leaving it off the email would be the inconsistency. We break down how the same portrait has to survive those three very different formats in our guide to real estate headshots.
• Recruiting and talent. You are cold-emailing strangers who are deciding in two seconds whether you are a person or a sequence. A face is a legitimacy signal.
• Sales and business development. Same logic, plus the practical benefit that a prospect who has seen your face in twelve emails recognises you on the video call.
• Client-facing consulting, law and financial advice. Relationship-led work where the client is buying judgement from a named individual.
• Freelancers, contractors and solo operators. There is no company face to hide behind. Yours is the brand.

Where it reads as self-promotional

In other contexts the same photo lands badly — not because anyone objects to your face, but because it signals that you think the email is about you. The common thread is that the work is judged on its content and the sender's identity is already established or irrelevant.

• Engineering and technical individual contributors. Your colleagues have your face in Slack already. A portrait attached to a code-review email reads as branding applied to a place that does not want branding.
• Academia and research. Correspondence conventions here are old and text-first; a photo in a signature reads as commercial in a way that a departmental affiliation does not.
• Internal-only corporate roles. If ninety-five per cent of your email goes to people on the same directory, the directory already has your photo and the signature is pure repetition.
• Government, public sector and regulated compliance functions. Formal registers where individuality in correspondence is actively discouraged.
• Anyone whose employer mandates a uniform signature. A personal photo bolted onto a corporate template usually just breaks the template.

A test that settles it

If you are genuinely on the fence, stop reasoning by category and ask a narrower question: of the last twenty emails I sent, how many went to someone who does not already know what I look like and might one day need to? If the answer is more than a handful, a photo earns its place. If it is zero or one, it does not — and you are better off putting that identity work where people actually look you up, which for most roles means your LinkedIn profile picture and your internal chat avatar.

That second point is worth holding onto. The work photo people actually see is rarely the one in your signature; it is the tiny circle next to your messages all day. We argue that case in detail in our piece on your Slack and Teams profile picture, and for anyone who ships code, the same argument applies to your GitHub profile picture. If you only have the appetite to fix one photo this quarter, fix the one that appears hundreds of times a day rather than the one that appears at the bottom of messages people scroll past.

What actually belongs in a signature — and what is filler

A signature exists to answer two questions: who sent this, and how do I reach or verify them another way? Everything that serves those two questions belongs. Everything else is decoration that costs you legibility, bytes and — in long threads — patience. The useful test is whether a recipient would ever act on the line.

The core set is short, and in this order it reads cleanly even as plain text:

1. 1Full name. The name you actually go by, spelled the way it appears everywhere else you exist professionally.
2. 2Role and organisation. One line. "Senior Product Manager, Northwind" — not a three-clause title that no external recipient can parse.
3. 3One phone number. The one you will answer. Two numbers means the recipient has to guess, which is worse than one.
4. 4A single web address. Company site, or your own if you are independent.
5. 5Anything legally required in your jurisdiction. More on this below — for most US senders, nothing.

And the filler, which is more common than the core set:

• Your email address. It is in the From field. It is in the reply. It has never been missing.
• A postal address you do not receive mail at. Unless you are sending marketing email subject to CAN-SPAM, or your jurisdiction mandates it, this is dead weight.
• Six social icons. Pick at most one platform where you are actually active and where the recipient would benefit from finding you. The rest are broken images waiting to happen.
• Inspirational quotes. They date instantly and every recipient reads them more times than you do.
• "Sent from my iPhone" left on a desktop signature. Or worse, a hand-written signature scan, which is a small security gift to anyone who wants to forge your approval.
• Environmental appeals not to print. Nobody has printed an email since roughly 2011.
• Availability banners, award badges, webinar promos. These belong in the body of the specific email that is about them, not appended to every message you send for the next two years.

The constraint that breaks most signatures: images do not always load

Here is the thing most signature advice skips. A signature is designed in a composer where every image renders perfectly, then delivered into clients that may refuse to fetch them at all. If your name, title and phone number live inside a graphic, a meaningful share of your recipients see an empty box where you thought your identity was.

Behaviour varies by client, and it is worth stating precisely rather than repeating the folk wisdom that "most clients block images":

• Classic Outlook for Windows. Microsoft's support documentation describes the setting "Don't download pictures automatically in HTML e-mail messages or RSS items" as on by default, so external pictures are blocked until the recipient chooses to download them or the sender is on a safe-senders list. Microsoft explicitly frames this as protection against tracking pixels.
• Outlook on the web and the new Outlook for Windows. Microsoft's 2026 roadmap entry for external-image handling (item 565862) describes the existing default as loading external content only for safe senders, while adding options to block all external content or to load all of it where an admin permits. Worth flagging honestly: the support article Microsoft actually publishes on blocking external images documents the Outlook apps for Android and iOS, so the web default is less firmly documented than the classic Outlook one.
• Gmail. Google's help documentation says images display automatically by default, with a user-selectable "Ask before displaying external images" option — and notes images are withheld anyway if Gmail judges the sender or message suspicious. Google separately documents that Gmail serves images through its own secure proxy servers rather than letting your host answer the recipient directly, so what renders is Google's copy.
• Apple Mail with Mail Privacy Protection on. Apple documents that remote content is downloaded privately in the background when the message arrives, rather than when you open it, and that your IP address is hidden from the sender. Apple also documents the failure case — a "Unable to load remote content privately" notice where Mail will not load the content without permission.
• Corporate security gateways. Many organisations run mail through a filtering layer that strips, rewrites or proxies remote images before delivery. Vendors here do not publish uniform, comparable defaults, so nobody — including us — can honestly tell you what share of business recipients are affected. Assume some are.

The design consequence is simple and non-negotiable: write the signature as text, and treat every image in it as optional garnish. Turn images off in your own client, send yourself a test, and read what arrives. If the version with no images loaded still tells a stranger who you are and how to reach you, the signature is correct. If it does not, no amount of visual polish saves it.

Why clients block images in the first place

It helps to understand that image blocking is not an accident of old software. It is a deliberate privacy control, and the thing it defends against is structurally identical to what a hosted signature logo does. A remote image is a request to a server, and that request tells the server something.

When a recipient's client fetches https://example.com/sig/logo.png, your server learns that this specific message was rendered, roughly when, from what IP address, and with what user agent. Make the filename unique per recipient and you have built an open-tracking system without ever calling it one. That is exactly the mechanism behind tracking pixels, which is why Microsoft names them in its own documentation on blocked pictures, why Google proxies images through its own infrastructure, and why Apple pre-loads remote content on arrival to decouple the fetch from the read.

Two practical implications follow. First, nothing you place in a signature should depend on knowing whether it loaded — no read-receipt logic, no per-recipient image names, no counting. Second, if your organisation hosts signature images on a marketing domain that is also used for campaign tracking, some recipients' filters will treat your ordinary internal email with more suspicion than it deserves. Serving signature assets from a plain, boring corporate host is the quieter choice.

Embedded images, linked images, and the weight problem

The obvious workaround to blocking — embed the image in the message itself rather than linking to it — trades one problem for a worse one. An embedded image travels with every single message, in every reply, to every recipient on the thread, and it changes how mail clients categorise your message.

There are three ways an image gets into a signature, and they fail differently:

• Remote link. Lightest possible message, but subject to every blocking behaviour above. Best default.
• Embedded as a base64 data URI. Always renders where it is supported, but base64 encoding inflates binary data by roughly a third, so a modest 10 KB logo becomes around 13,600 characters. Google's Workspace documentation puts the Gmail signature ceiling at 10,000 characters — meaning a single embedded logo can exceed the entire allowance before you have typed your name.
• Attached inline (CID). Renders reliably in many desktop clients, but the image is a genuine attachment. Some clients then show a paperclip on every message you send, which trains recipients to ignore the one email where the paperclip mattered, and some archiving and ticketing systems treat each copy as a separate stored attachment.

If you do use a logo, keep it small in both senses. Google's admin guidance for Workspace signature footers recommends an image around 70–100 pixels high by 300–400 pixels wide, with a documented maximum of 100 by 1000 pixels. That is a useful sanity bound even outside Workspace: if your signature graphic is larger than a business card at screen resolution, it is too big.

If you do include a photo, build it for a 60-pixel circle

A signature photo is one of the smallest renderings of your face anywhere in your professional life — typically 60 to 100 pixels across, often circular, and frequently sitting on a white background with no border. Photos that work at that size are not simply good photos scaled down. They are photos composed for the crop.

Three things decide whether it survives:

1. 1Crop tight. Chin to just above the head, shoulders barely present. A three-quarter-body shot at 60 pixels is a smudge with a collar. Our explainer on what actually counts as a headshot crop covers where the framing lines fall and why each one exists.
2. 2Separate from the background. The signature sits on the client's white message background, so a light, low-contrast backdrop makes your head float. A mid-tone or subtly darker background gives the circle an edge.
3. 3One face, no props, no logo overlay. Anything else at that size is visual static.

Practically, this means the signature photo is usually a tighter re-crop of your main headshot rather than a separate shoot. If you are assembling a set of images for different placements — signature, profile, speaker bio, team page — our complete personal branding photo kit lays out which crops you need and what each one is for, so you shoot once and cut many times. And whatever file you use, give it a real alt attribute; more on that below.

Pronouns: how to include them without making them a policy

Pronouns in a signature do two useful things: they reduce the chance a recipient misgenders you in a reply, and they mean colleagues who would otherwise have to disclose unilaterally are not the only ones doing it. Both benefits are real and neither requires argument. The design question is smaller than the debate around it.

Put them on the name line or immediately below it — Maya Chen (she/her) — in the same size and weight as the surrounding text. Do not set them in a different colour, do not add an explanatory sentence, and do not link them to a definitions page. A signature element that needs a footnote is not doing its job.

For organisations: guidance from inclusion practitioners is consistent that pronoun fields should be offered rather than required. Making disclosure mandatory in a company-wide template compels exactly the people the practice is meant to support. Offer the field in the template, leave it blank by default, and say clearly that filling it in is a choice.

Legal disclaimers: mostly optional, occasionally mandatory

The four-paragraph confidentiality notice at the bottom of half the corporate email in the world is largely a habit rather than a requirement. It is worth knowing which parts are genuinely mandated where you operate, because everything else is text your recipients scroll past on every message forever.

Three cases are worth separating:

• Genuinely mandated in some jurisdictions. UK companies fall under the trading-disclosure regime made under the Companies Act 2006 — though the detail sits in the Names and Trading Disclosures Regulations 2015, not in the Act, which only delegates the power. Regulation 24 requires the registered name on business letters, notices, websites and "all other forms of its business correspondence and documentation". Regulation 25 requires the further particulars — place of registration, registered number and registered office address — on business letters, order forms and websites. Neither regulation names email explicitly, so whether your footer must carry the full regulation 25 set is a question for your own advisers rather than one we can settle for you. Comparable company-particulars rules exist across the EU.
• Formerly mandated, now actively discouraged. The ubiquitous "IRS Circular 230" tax disclaimer existed because of the covered-opinion rules in former section 10.35, which Treasury eliminated in final regulations applying to written tax advice rendered after 12 June 2014. The preamble said Treasury and the IRS expected the change to end the use of a Circular 230 disclaimer in email. If you still see it in 2026, it is inertia.
• Never mandated, widely believed to be. HIPAA does not require a confidentiality footer on email. HHS's own guidance on emailing patients lists reasonable safeguards — checking the address is right, limiting how much is disclosed, meeting the Security Rule on transmission — and says nothing about disclaimer text. A footer may still be a reasonable internal control, but nobody should imagine it substitutes for encryption, access control or training, and it has limited standing against a recipient who never agreed to it.

If you must carry a disclaimer, put it last, set it smaller and greyer than the rest, and keep it to a couple of sentences. If your legal team insists on four paragraphs, ask whether it can be appended only to external mail — most enterprise mail platforms can apply footers conditionally, and internal threads are where the bloat hurts most.

The mobile signature is a different artefact, not a shrunken one

Most people either leave the factory "Sent from my iPhone" in place or paste their full desktop signature into the mobile client, where it looks enormous and often loses its formatting entirely. Neither is right, because the mobile signature is doing a different job in a different context.

Mobile email is read on a narrow screen, usually while the recipient is doing something else, and is frequently a fast reply rather than a first contact. So strip it to the minimum that still identifies you:

• Name, role, one phone number. Three lines.
• No images at all — not the photo, not the logo. Mobile clients handle signature images least predictably and the payoff at that width is nil.
• No disclaimer, unless it is one of the legally mandated cases above.
• Consider keeping a short, honest brevity note if you write terse from your phone — one clause, not a paragraph.

Deleting the default "Sent from my iPhone" is fine, but do not replace it with something arch. The stock line has one genuine function: it explains why a two-word reply arrived with no punctuation. Something plain like "Sent from mobile — brief by necessity" preserves that function without the performance.

Reply-chain bloat, and why it is your problem

Signatures are appended to every message, including the sixteen-word reply in a thread you have been trading all afternoon. By the twentieth message, a heavy signature has been serialised twenty times into the same thread, and the actual conversation is a thin ribbon of text between repeated blocks of contact details, logos and disclaimers.

This is not merely aesthetic. Long threads with repeated embedded images grow the stored message size for everyone on it, complicate quoting, and make the thread genuinely harder to read on mobile. Some ticketing and CRM systems parse threads to extract the latest reply and are confounded by ornate signature markup, filing your quick answer as an unreadable blob.

The fix is to run two signatures in every client that allows it: a full one for new messages, and a minimal one — name and phone, or nothing at all — for replies and forwards. Gmail and Outlook both let you set a separate reply signature; Apple Mail on Mac does not — Apple documents one signature per account plus a choice of where it sits relative to quoted text, so there you either keep the single signature short or delete it by hand on replies. It takes five minutes once and it is the single highest-leverage change in this article for anyone whose day is mostly threads.

Accessibility: what a screen reader hears

A signature rendered entirely as one image — a common output of drag-and-drop signature builders — is invisible. Not degraded: invisible. A screen reader announces whatever the alt attribute says and nothing more, a text-only client shows a placeholder, and a recipient trying to copy your phone number into their dialler cannot select it. Every recipient loses; some lose completely.

The rules are short and follow directly from WCAG's non-text content requirement, which asks that every informative image carry a text alternative:

• Never render text as an image. Names, titles and phone numbers must be real text. This also lets them be searched, translated and enlarged.
• Give informative images meaningful alt text. "Northwind logo" for the logo; your name for the portrait. Describe purpose, not appearance.
• Give decorative images empty alt text. An alt="" tells a screen reader to skip a divider or spacer rather than announce it.
• Meet contrast minimums. WCAG 2.1 AA asks for 4.5:1 for normal text against its background. Pale grey signature text at nine pixels fails this and also fails anyone reading on a phone in daylight.
• Use descriptive link text. "northwind.com" or "Book a call" — never a raw tracking URL, and never "click here".
• Keep the reading order linear. Multi-column table layouts read out of order. A single column is both more accessible and more robust on mobile.

Putting it together, and testing it honestly

A correct signature is boring to look at in the composer and reliable everywhere else. Once you have decided the photo question for your own field, the build is mechanical — and the test at the end is the part almost nobody does, which is why so many signatures ship broken.

The build, in order:

1. 1Write the whole thing as plain text first: name, pronouns if you want them, role and organisation, one phone, one URL. If it does not read well here, it will not read well anywhere.
2. 2Add formatting sparingly — bold on the name, one accent colour at most, a single divider.
3. 3Add images last, and only ones you can afford to lose: a small remote-hosted logo, a tightly cropped portrait if your context calls for one. Alt text on both.
4. 4Create the short reply signature and the separate mobile signature.
5. 5Add mandated legal text last and smallest, external-only if your platform supports it.

Then test it properly. Send it to a Gmail address, an Outlook address and one phone. Turn images off in at least one of those clients and read what arrives. View the message as plain text if your client offers it. If you can, have a colleague on a corporate gateway confirm it survives their filter. The whole test takes ten minutes and it is the difference between a signature you designed and a signature your recipients actually see.

One last framing. A signature is a small piece of a larger surface — the set of places where people encounter you professionally without meeting you. It is worth getting right, but it is rarely the highest-value item on that list. If your headshot in the signature is stale, the same file is almost certainly stale on LinkedIn, in Slack and on your company team page, and those get looked at far more often. Fix the source image once and the signature fixes itself along with everything else.