Rubicon Project: initial impressions

If you looked to the bottom right of my sidebar in the past couple of days, you might have noticed a new graphical ad unit. This is from the Rubicon Project, a well-funded L.A.-based start-up that does publisher-side advertising optimization to maximize yield. They’re still in beta, but seem to be pretty open – I signed up for the beta and was accepted a couple of days later, despite minimal traffic.

First off, the Rubicon Project has some quality UI designers. The site is easy to get started with and simple to use. Registration is about what you’d expect, as is the Terms of Service – I’m not a lawyer, but I didn’t see anything out of the ordinary here. You are asked to provide a Social Security Number or Tax ID, since the Rubicon Project consolidates ad network payouts into a single check. Funds received by the Rubicon Project in a calendar month are paid out midway through the next month. (Of course, the ad networks have their own terms of payment with the Rubicon Project, so impressions seen one month might not be paid for until a couple months later.)

After registration, set-up is a four-step process. Publishers are able to select the types of ads (text or image) and the sizes they want – although the number of sizes is strangely limited to four popular ones (728×90, 468×60, 160×600, and 300×250.) Publishers then self-categorize their content – for example, this is a technology blog – and call out if they feature any unreviewed user-generated content, which some ad networks won’t touch. Here publishers can also specify the color scheme of your site for their text ads, although I don’t believe the Rubicon Project automatically attempts to optimize the coloration like, say, Pubmatic. Publishers then select the types of ads they don’t want displayed – audio ads, casino/gambling ads, dating ads, download (wallpaper, smiley) ads (read ‘adware’ here), false message ads, provocative ads, shaky/flashing ads, and windows dialog ads. The Rubicon Project doesn’t take adult ads, so there’s no need to worry about porn. Once you’ve specified the ads you want, you’re provided with a standard set of JavaScript tags – one for each ad size. Stick them in your site and you’ll see ads in a few hours. (Interesting, this delay – in some places it’s near instantaneous. Probably speaks to an architecture decision or manual review process I don’t have insight into.)

The tags themselves work just how I’d expect them to – a call is made to the Rubicon Project, which in turn decides on a third-party tag and sends it back to the browser, forcing the browser to call the third-party tag by using document.write. Note that I didn’t see any IFRAME or SSL options, so I wouldn’t stick these ads on any page with secure content. (Not that you should be showing ads against secure content anyway.)

Once you’ve gotten ad tags, you can see what networks the Rubicon Project has chosen for you. Here’s where a single system-wide account with each ad network comes in handy – nine sources of ads (AdSense, Active Response Group, AdBrite, Clickbooth, CPA Empire, CPX Interactive, eType, Interclick, and Shopzilla) were selected to run on my site, and I don’t have to have an account on any of them. The Rubicon Project provides a decent description of each and lets publishers review each source. Publishers can shut off networks they don’t care for – if I didn’t want CPC-based Shopzilla affiliate ads on my site, I could just kill the network. However, there’s no evident way to add additional networks to your list. This is a bit frustrating when browsing through the Rubicon Project’s ‘Network Directory’, which shows over 130 ad networks in total, including the biggest. I’d love a taste of DoubleClick or ad.com – why can’t I have it? If the Rubicon Project’s going to tease me with a directory of the biggest ad networks, they’d do well to explain why each ad network listed isn’t eligible for my site.

The Rubicon Project’s reporting is thorough – impressions and estimated clicks by date, by ad size (by date and by ad type), by ad network (by date and by size), and by geography, all with customizable date ranges. The whole lot come with attractive charts and are exportable to Excel. One minor interface issue – I didn’t realize that I could drill down to the continent and country level by clicking on the world map until I stumbled on it by accident. Earnings are broken down by day, size, and network.

Finally, the Rubicon Project shows how they’re optimizing your inventory overall and by country. You’re able to see each ad network’s weighting in the form of a horizontal slider, and can (if you wish) override the Rubicon Project’s settings to weigh the ad sources yourself. The ‘by country’ view provides more detail than the ‘by network’ view – here, you can adjust the weighting of each individual third-party ad tag as opposed to simply by network, although the Rubicon Project could provide a bit more detail about the different tags. It’s easy enough to figure out what acronyms like ‘ROS’ mean, but how does ‘CPX_FF_160×600′ differ from ‘CPX_160×600′? Without this information, there’s not much point in having access to the slider.

Overall, I’m impressed with this product – it’s good-looking and easy to use. Of course, it’s the optimization algorithms that’ll make the product shine, and I’ve got no insight into how good these are. Even if optimization didn’t exist, the Rubicon Project would earn some of its dollars just by consolidating the payments and getting smaller publishers access to networks they normally wouldn’t run on. But I did have some minor concerns, which I hope the Rubicon Project will address in future releases.

First, the Rubicon Project is using user agent targeting to limit access to their interface – only Firefox 2.0+ users need apply – I didn’t dig in, but perhaps this is due to the complexity of the beautiful JavaScript-dependent interface. Fair enough, guys, but there are other browsers out there that use the same rendering engine as Firefox – like Camino, my browser of choice. No reason why Camino shouldn’t work too.

Next – and this is more serious – when I went through the ‘forgot my password’ process, I was e-mailed my original password instead of a link to reset it. This means the Rubicon Project stores passwords in its database in plain text, or in a format that can be decrypted into plain text when necessary. This is a very poor security decision that doesn’t inspire confidence in the development team. Please, salt and hash your passwords – it’s easy. It’s especially important for the Rubicon Project, because as a provider of consolidated payments, you’re storing Social Security Numbers. Speaking of security, this SSN is displayed in plain text within the ‘account’ tab interface. Guys, please allow people to edit the SSN through the interface, but treat it much like a password – asterisk the actual value out and use dual entry boxes to catch typos.

Finally – there’s no (easily visible) price tag for the service. The Rubicon Project has taken an interesting approach to transparency, since they’ll happily show you which ad networks are serving and the relative performance of each. However, they don’t seem to show publishers the size of their cut. If I were a publisher, I might be very tempted to forego the convenience of consolidated payments, get my own accounts with the better-performing ad networks (thanks for the analysis, Rubicon Project!), and set up an ad server to serve the ad network tags for me. Ad servers are free, and they even come with (admittedly not the best) geotargeting. If the Rubicon Project wants to keep the biggest and best publishers from taking its analysis and trying to do it themselves, they need to demonstrate that their price for their services is fair.

The Rubicon Project ad block in the sidebar probably won’t stay forever, but I’ll be watching future releases closely, and may add it back from time-to-time when I want to get data in their reporting system. For honesty’s sake, I’d also like to mention here that I work for Yahoo, which does own and offer a competing product. In order to avoid any real or perceived bias, I haven’t mentioned the competing product or compared any features of it to the Rubicon Project.