Three steps make the card. The rest is busy-work.

If you have read another how-to on this, you have already seen the generic recipe: pick a template, type your message, add the recipient's email, hit send. The recipe is not wrong. It just treats every step as equally important, which is how you end up with an article that spends three paragraphs on font selection and one sentence on the cover. Font choice does almost nothing. Sticker placement does almost nothing. The cover, the first signed message, and the delivery timing do almost everything.

If you skim only this paragraph: pick a real photo of the recipient or generate a custom AI cover, sign the card yourself first with a specific real message before you invite anyone, and set the delivery time for the morning of the birthday in their time zone, not yours. The rest of this piece walks the actual steps, but if you remember those three, you have ninety percent of it.

What you actually need before you start

Less than the average article suggests. You do not need to pre-write the perfect message. You do not need a stack of stock photos. You do not need to commit to a template before you start building. What you do need is a sense of who the card is for and roughly how the recipient will read it.

Two questions, answered before you open the card tool: is this a solo card from you, or a group card that the friends or coworkers will sign? And is the recipient the kind of person who would prefer a single big photo on the cover, or the kind who would prefer a designed graphic? Solo card to your dad, photo of you and him at his sixtieth two years ago, done. Group card from twelve coworkers to a manager turning fifty, custom AI cover with a private joke about her well-known coffee habit, also done. Those decisions are the actual work. The clicking is the easy part.

The five steps that actually matter

Here it is. The shortest version of the instructions that, if you follow them, will produce a card the recipient actually keeps a screenshot of.

  1. Pick the format first. Board layout for crowds (post-it notes spread across a page, holds twenty signatures gracefully). Multi-page greeting card with envelope animation for a smaller, closer group. Past about six signers the greeting card starts to look thin; under about four signers the board looks empty. For a friend's milestone year with the whole circle invited, board. For your dad from his three kids, greeting card.
  2. Make the cover the right cover. This is the single highest-leverage step. Two real options: a photo of the recipient (or you and the recipient) from a real moment in your shared history, or a custom AI cover generated from one specific detail about who they are. Skip the stock-photo templates. The cover is what they see first and remember longest; a generic image of balloons over a wood floor is a wasted opportunity.
  3. Sign the card yourself, first, before you invite anyone. Write a real specific message. Two to four sentences. Name a moment, not a category. Your line sets the standard for every other signer. If your line is generic, the rest of the card will be too. If your line names a real Tuesday in November when the recipient did something specific, the rest of the card will rise to match.
  4. Invite signers with one sentence of guidance. Paste the share link into the group chat that already exists (Slack channel, WhatsApp thread, group iMessage). Add one sentence. Tell people the register, and ask for something specific. The word 'specific' in your invitation does roughly a third of the work by itself, because most people want to write something good and just are not sure what is expected.
  5. Schedule delivery for the recipient's morning, in their time zone. Not yours. The signatures keep collecting right up to the moment of delivery, so an invite sent a week early still picks up notes the night before. And if the birthday is tomorrow and you forgot, send it now anyway. A late card lands better than no card and far better than the awkward apology text three days later.

The AI cover step competitors can't offer

This is the step that the templated platforms genuinely cannot do, and it is the reason I switched to building cards on Reco. Most online card tools give you a library of template covers: balloons, candles, geometric confetti patterns, a watercolor cupcake. The library is large, and every cover in it has been used ten thousand times before yours. The recipient has seen most of them on every other card they have received.

The AI cover step is different. You describe the recipient in one or two sentences (her name is Priya, she is turning forty, she runs marathons and roasts her own coffee beans) and the tool generates a custom illustration that fits that description. I remember the cover I made for a friend's fortieth last year, which had a coffee bean shaped like a running shoe on it. Nobody else on earth has that cover. She kept the screenshot. Reco's online birthday ecards include this step natively; most other free group-card platforms do not.

One small caution. The AI cover works when you give it something specific about the recipient. Give it 'happy birthday' and you will get a generic cake. Give it 'happy birthday to a literature professor who collects old maps and complains about modern fonts' and you will get a cover that no template library could have provided. The more you tell it, the less generic the output gets.

When the timing matters more than the design

I have made a lot of beautifully-designed birthday cards that arrived at the wrong time and landed badly. A friend's birthday card delivered at 11pm her local time the night before, which the platform thought was 'on the morning of her birthday' because I had not set the time zone. A coworker's card delivered three days late because I had scheduled it correctly but the delivery window was set for business hours in the destination's region and the recipient was on vacation in Europe. A milestone card for my mother delivered to her work email at 6am, where she did not see it until she got to her desk and was already mid-meeting.

The timing rule that has worked for me: pick the moment in the recipient's day when they will most likely have a quiet ten minutes to read it. For most adults that is the first thirty minutes after they wake up, in their actual time zone. Back-calculate from there. If the recipient is on the East Coast and you are on the West Coast, schedule the card for 5am your time, not 8am. If they travel a lot, send to a personal email rather than a work one so it lands wherever they actually check first. If you are completely unsure of their schedule, 9am their time on the day is a safe default that almost never lands badly.

Turn it into a group card

Birthdays past about age ten are better as group cards. The geometry works: more voices, less pressure on any single signer to be profound, and a finished thing that reads like a small archive of how people feel about the recipient. A group birthday card online handles the coordination without phone trees, paper cards passed around the office, or you personally chasing eleven coworkers for signatures. One link goes to the group chat, each person writes their own line on their own time, and the card delivers itself at the moment you scheduled.

You can create a card online in a couple of minutes; the actual decisions (cover, format, delivery time) take five more minutes; the signatures collect themselves while you do something else. If you are stuck on what to put in your own first message, the cluster pieces on birthday wishes for a friend and what to write in a birthday card have lines you can adapt. For a card going to a milestone year, the milestone birthday guide has more on what makes those particular cards land.

If the recipient is somebody you do not know that well (a new neighbor, a friend-of-a-friend, a colleague you have crossed paths with maybe four times), the related piece on how to send an ecard covers the channel question (email vs text vs share link) better than this article does. The channel matters almost as much as the card.

One last thing, off-topic. I keep one folder on my laptop called 'cards sent', with a screenshot of every group card I have ever organized, going back to maybe twenty-eight of them now. I have never once gone back and looked at one. The folder exists for no reason I can articulate, and I keep adding to it anyway. There is something I cannot put into words about the small ceremony of making one of these, and the folder is the only evidence I have that it matters to me at all. Whichever platform you use, however you build the card, the actual move is just sending it. The unsent card is the only one that fails.