Groupon started out as one deal a day. We had a massive waiting list from merchants - 9 months, at one point.

With so much merchant demand and Groupon only running one deal a day, clones started to appear. Instead of waiting to be featured on Groupon, merchants ran with the clones.

Most customers didn't really care whether they bought from Groupon or its clones. So we realized what in effect was happening was that the model was moving to a marketplace... just one that was distributed across multiple sites.

We decided if the inventory was going to exist, it would be better on Groupon than spread across a zillion clones, so we moved to a marketplace model.

In other words it was mostly competition in a fairly commoditized space, and not scalability issues, that drove us to the marketplace.

That's not to say that scalability wouldn't have been a challenge with a daily deal model, but there are things you can do. A lot of the needs in the space actually balance out nicely. Some merchants want 20k customers, others want 20. Most merchants prefer new customers, some are indifferent. Both customers and merchants want category diversity so Groupon remains interesting and non-cannibalistic. In theory you can get all those things with smart segmentation and personalization. We experimented with that in the transition to marketplace but ultimately went ahead with moving to the "many deals" (which in the short term, is exactly what customers want) to stay ahead in the brutal competition for marketshare. Hard to say whether I'd do that part differently today... if we'd been less focused on maintaining marketshare NYTimes might have just written the obituary for Groupon instead of Livingsocial.

Once competition fragmented the market and businesses realized that 50% off coupons attracted bargain hunters rather than loyal customers, there wasn't any going back.

You could probably make a pretty good living off something like that, especially if you can scale it to several local areas.