“Uniques” Are Not That Important

If you work in online advertising (and really, who doesn’t), then you have probably been on either end of the following conversation thousands of times:

A: That website gets a ton of traffic.
B: What does “at ton” mean?
A: Last month, they had ten quatrillion pageviews.
B: How many uniques?
A: Only a hundred and fifty billion.
B: Heh. That’s what I thought.

It’s very common to talk about a site’s number of unique users as some kind of litmus test for how much traffic the site is really getting. Perhaps the overwhelmingly positive connotation comes from the world itself. That is, hearing the word “unique” reminds us of those cozy bedtime tuck-ins when our mothers told us that we were special snowflakes destined to save the world from Nazis or something.

But in reality, uniques aren’t that special. For example, my recent post about my relationship with Kari Ferrell has received over a thousand pageviews, and almost all of them were uniques. All that means is that very few people have bothered to come back (so far) to check out what else I might have to say. In other words, I got the traffic, but it’s not “sticky.” This is common. Sudden increases in traffic to a website are almost always accompanied by a sudden increase in the percentage of unique visitors. Then the moment passes, and you’re back to dealing with your regulars again (plus, hopefully, a few additions).

If you are a direct-response marketer with no desire to actually build a loyal audience, then you might be happy with a high percentage of uniques. Advertisers like that are interested in quantity, not quality, so they often figure, the more eyes, the better.

But, even in the most cynical scenario, maximizing uniques is still the wrong approach. Advertising often takes multiple exposures to take hold, with three repetitions being the magic number that was bandied about for almost a hundred years, only to be suddenly forgotten in the first era when we can finally keep track.

Old media “traditional” advertising is planned according to GRPs, or Gross Ratings Points. GRP is reach multiplied by frequency, where reach is how many people saw the ad, and frequency is how many times they saw it. GRP is a goofy metric anyway (100 people who see an ad once yields the same GRPs as 10 people who each see the same ad 10 times), but at least it represents an attempt to address the fact that frequency matters.

An ideal situation for an online marketer would be a site that has exactly one-third unique users, with each user going back to the site three times per month. You then frequency cap the ads to three. Each user gets their three exposures, and you see what happens.

But sometimes it takes more than three exposures. I’ve had the opportunity to see some fascinating data on this in my own work. It’s not uncommon for even a hardcore direct-response campaign to yield close to zero conversions for the first couple of weeks. And then, one day, and I’d be lying if I didn’t add “like magic,” the conversions start rolling in. What happened? Did the users get the magic number of exposures? Did it take that long to reach the people who just happened to also be the client’s customers? I could theorize on a case by case basis, but it’s fair to say that I wasn’t wishing for a greater percentage of uniques.

Of course, if you’re not a direct-response marketer (and why would you want to be), then you really don’t want a lot of uniques. You want an active community of users who come back to the site every day, and obsessively refresh the pages, hoping for updates.

image: A graph of the bounce rate on this blog during a recent traffic spike.