Embedded virality: How Mailchimp’s footer built the company

When Mailchimp went freemium in 2009, it put a small badge in the footer of every email its free accounts sent. A year later its founder reported going from 85,000 users to more than 450,000, with profit up over 650%.¹ He credited the profit jump to acquisition cost falling below $100, and said the move wasn’t a strategy at all: they did it for fun.

The badge wasn’t the only reason, but it’s the mechanism worth stealing, because it reframes what a free tier is. Not a discount, not a trial, not a loss you tolerate: a different payment method. Free users pay in impressions, and that produces the cleanest paid feature in software, which is charging people to take your name off their work.

What is embedded virality?

Embedded virality is a PLG acquisition pattern where your branding appears inside the normal output of your product, so non-users encounter your name while doing something else. A support widget, a form footer, an email signature, a scheduled post.

It is the weakest of the exposure patterns and the most widely applicable. That is the trade. Almost any product with a customer-facing surface can run it, and few get a strong signal from it.

Nobody sees “Powered by Typeform” on a form and concludes the product is good. They conclude it exists. Awareness at near-zero cost is worth having, as long as you do not mistake it for persuasion.

One boundary worth drawing, because it decides which surface you should be improving: if the non-user completes a task, you’re running co-experience, not this. DocuSign gets filed here often, but a signer uses DocuSign rather than merely noticing it, and the thing to optimize is the signing flow rather than a logo.

Why removing the “Powered by” badge is the best paid feature in software

Most freemium tiers can’t explain what the free user gives you. The honest answer is usually “nothing yet, and we hope,” which is why upgrade prompts so often resort to withholding something the user needs. Embedded virality supplies a real answer, and that changes the shape of the pricing conversation.

Mailchimp’s version was the strict one: the footer badge rode on free accounts and paying made it go away.² Nothing was crippled. The free product sent real email to real lists. The paid feature was privacy.

Intercom drew the same line, but one level up. Its cheapest paid plan still carried “Powered by Intercom,” and only higher plans could remove it.³ So it wasn’t free users paying with exposure. It was the smaller paying customers.

Buffer shows what the trade saves you when it works. Its published figures put it at $244,215 of revenue per teammate, with fewer than 90 people and no sales team. That is what a distribution model costing almost nothing looks like from the outside; what produced it is not something the numbers alone establish.⁴ When distribution is a property of the product rather than a department, the money goes into product instead.

The pattern in both cases is the same: the more a customer’s badge gets seen, the more they will pay to remove it. A hobby project doesn’t mind the badge and doesn’t generate many views. A funded startup with a widget on its pricing page minds a lot and generates a lot. So the customers costing you the most exposure are the ones most willing to pay you to stop. You would struggle to design that on purpose.

One category check belongs before any of this. A “Powered by” mark on a beautifully designed form reads as a recommendation. The same mark on a security tool, a payments flow or anything touching sensitive data reads as a question about who else is involved, and that question costs more than the impression was worth. Some categories are hurt by being badged at all, and no amount of placement fixes it.

Two consequences follow, and both are counterintuitive.

Put the badge in your pricing table, not your footer. If the exchange is real, name it: free tier includes our branding, paid removes it. Teams hide this because it feels like admitting to a tax. Say it plainly. A customer who understands what they are buying decides faster than one working it out.

Placement beats volume. Two badges with identical impression counts can produce opposite results. Mailchimp put its badge at the bottom of something the recipient had chosen to open, and its MonkeyRewards program later layered account credit on top when a badge click converted.² The passive loop generates awareness; the credit turns willing users into deliberate promoters. That’s one feature running two loops, and either can fail without taking the other down. That resilience is why it’s worth copying.

The placement principle is that the impression should land after value is delivered, never during. A badge following helpful support builds a positive association. The same badge interrupting the same interaction builds resentment. The counter is identical; what differs is what the impression teaches.

Narrow positioning makes a logo say more

There’s a second lever, and it costs nothing: what your name means to the person reading it.

“Powered by ConvertKit” on a creator’s blog wasn’t primarily advertising features. It told other creators that a tool existed for them specifically.⁵ A generic email badge in that context communicates almost nothing. A category-specific one says “this was built for you,” which is far more in the same small space.

This generalizes past email. The narrower your positioning, the more a logo carries, because the reader supplies the rest from context. If your positioning is broad, the badge says almost nothing.

The churn trap

RB2B, a website-visitor identification tool, reported reaching roughly $5M in annual recurring revenue in about 13 months on this kind of distribution. Along the way its founder published monthly churn figures in the double digits: around 20% on the entry plan in late 2024, still double digits through the $2.4M to $4M stretch, improving to roughly 7% by mid-2025.⁶

Read those together, not as a snapshot. For most of the period the growth chart was compounding while the product replaced a double-digit percentage of its customers every month, and both facts were true simultaneously. Viral distribution is excellent at putting people through the door and completely indifferent to whether the product keeps them. That is an activation failure wearing a growth chart, and it’s the first of the three ways PLG fails.

There is a version of this that runs inside a company, not out in public, and for B2B it is the more valuable one. Route the product’s output into a shared channel rather than to one person: an alert, a summary, a report that lands where a whole team reads it. The person who set it up gets their result, and everybody else gets a demonstration they did not ask for. Value visible to a team expands within that team; value delivered privately has to be described secondhand by somebody who is busy.

There’s a quieter failure too. Every free customer decides where your name appears, and at volume some of them will attach it to spam, broken pages, and dead projects. Those impressions land in your dashboard, and nothing follows them.

Three tests you can run this quarter

None of the three needs a number from outside your own company.

1. What share of new signups is badge-attributed, and is that share rising? Tag the clicks and watch the fraction across quarters. Raw counts rise with customer growth even while the loop decays; only the fraction tells you whether it compounds.
2. Do badge-sourced users retain as well as paid-acquisition users? Badge traffic always arrives with low intent; that is what a badge is. If it also retains worse, your CAC advantage is smaller than it looks and possibly negative.
3. Will anyone pay to remove it? The sharpest test available, and the one that rarely gets run. Offer removal as a paid option and watch take-up. Real demand proves the badge is prominent enough to matter. Zero demand means nobody notices it, which means it isn’t working.

Four things to change

Price badge removal explicitly, as a line item in the pricing table. If the exchange is real, say it out loud rather than hiding it.

The other three are small. Move every badge to a moment after value has been delivered, which is usually a one-line change and buys the same impressions with the opposite association. Confirm each one is a link that lands on signup, because a badge nobody can click is a decoration you are paying for in goodwill. And segment retention by acquisition source this quarter: if badge-sourced cohorts churn worse, fix retention before amplifying anything. RB2B spent its first year growing fast and churning at double digits simultaneously.

A badge reaches people who weren’t looking for you. The other end of the Evaluate stage is being there at the moment someone types exactly what you do into a search box, which sounds simpler and has far more ways to go wrong: product-led SEO.

Updated August 2026