[MalcolmDiggs]

In my experience, there's usually two parts to this:

1. Initial external paid marketing (in the form of Adwords campaigns, email blasts, facebook ads, affiliate programs, etc).

2. Reworking the product until the "viral coefficient" is higher than 1.0. The viral coefficient(aka "k-factor" or "k value"), is basically just the likelyhood that one existing customer will refer another customer. So if your k-factor is above 1.1 for a given period, your initial userbase will grow over each period with no additional marketing needed.

In practice however, few startups ever reach a k-value of 1, so many rely on a combination of virality and traditional paid marketing in perpetuity.

[1] https://en.wikipedia.org/wiki/K-factor_(marketing)

[pesfandiar]

At the risk of sounding pedantic, K-factor is the expected value of new customers referred by an existing one; probability of 1 would be too good.

Also, for external marketing, it doesn't have to be paid. It could be in the form of time investment in content marketing and other forms of inbound (of course that's if you don't pay someone else to do it).

[MalcolmDiggs]

1. Read the link in my original comment. There's no reason values above 1 would be "too good", many people achieve it. If you're going to be pedantic, you should get your facts straight.

2. Nobody said it "has to be" paid. The OP asked how "you" market your startups, and I told him what the companies I've worked with have done, in my experience.

[pesfandiar]

I have no intention of starting an argument, but to add value to the thread. I take the bait though:

1. "likelyhood [sic] of 1" means 100% probability. Any likelihood over 1 is by definition impossible.

2. I assumed by "there's usually two parts" and "external paid marketing", you meant people have to pay for marketing. I added my personal experience to the conversation that it's not necessarily the only option.

[MalcolmDiggs]

Again, you're wrong. And yes, you clearly have every intention of starting an argument.

1. If you'd simply read the link (which you clearly have not done), you'd understand that a k-value of 2.0 means that each customer referred an average of 2 new customers. I never said "probability". You did. Yes, probability above 1 is impossible, which is why I never used that word. You clearly conflated the two, which is your error, not mine.

Here's why I said "likelyhood" (likelihood) instead of "probability": http://stats.stackexchange.com/a/2645

"...the likelihood function does not obey the laws of probability (for example, it's not bound to the [0, 1] interval). "

2. Reading-comprehension. It's not that hard.