Most successful AdSense publishers will have, at some point, been approached by an ad network that guarantees to beat AdSense. The pitch is simple and sounds almost too good to be true: “Switch your AdSense code for our code and we’ll pay you more than AdSense. If we can’t do that for any ad impression we’ll just show AdSense instead”. How could you possibly lose?
As Google Certified Publishing Partners we encounter this pitch monthly. Although we’re trained and qualified by Google, we’re employed by site owners and our first allegiance is to helping them earn more through advertising. It might be surprising then that we rarely use the networks who make these claims. Whilst there are many situations where adding additional networks into the mix can result in higher yield, this particular sales approach relies on a bit of a three card trick.
So, where’s the lady?
The ruse here is that they are not offering that price on every impression – just the ones that they like the look of. A more honest way of wording their pitch would be “We’ll pay you more than your average performance for your very best impressions”, which of course is what every networks does – including AdSense.
Ad network deals are priced on RPM – the revenue generated per 1000 impressions. You are probably already familiar with your website RPM by looking at performance reports with the ‘Ad Requests’ columns selected:
This is the figure that networks will ask you to look at when setting the price that they promise to beat. All seems above board so far.
Remember though that the network isn’t promising to beat that price on every impression – just the ones that they choose. Consider how your average price is made up and this suddenly starts looking like less of a guaranteed win.
The following report shows the revenue profile for the same site above. The average RPM is just $2.11, but 50% of the impressions are paying over $3.64. More surprising still, advertisers are willing to pay over $6.54 for 20% of the impressions and over $12.15 for 5%.
This isn’t an unusual profile for any site. Your average RPM is made up of a range of prices, roughly half of which you probably wouldn’t want to sell for your average.
A publisher who is aware that their RPM is $2.11 might be very tempted with an offer of (for example) $3.50 on a first-look basis, but the chart above shows that this might not increase their earnings in the way they think and would in fact lose them money on a very significant number of impressions.
If a publisher replaces AdSense code with another network’s tags then they are giving the network first look at every ad request. This means that the network gets to choose whether or not to serve an ad each time one is requested, and passes that request back to AdSense if they choose not to.
Now, let me ask a question: If you were a network and could pay a fixed price for any of the ad requests in the chart above would you choose the low value requests to the left, or those worth 6x the average price on the right?
What happens next?
If the highest value requests are sold at a fixed price based on the average, then this can cause 2 immediate problems for the publisher:
- The fixed rate might not reflect the true value of those ad requests
- The average AdSense RPM will drop without the benefit of the high value requests
Together these issues can actually cause a drop in earnings. A “sharp” network will find this easy to blame on Google as it will be the AdSense rate dropping, but the cause is an undervalued fixed price pushing AdSense returns down.
Without the high-value ad requests there is also a risk that the site will be seen by advertisers and the AdSense system to have lower quality traffic – therefore attracting lower bids and dropping returns further.
Does this mean that publishers shouldn’t use other networks?
Not at all. AdSense is so powerful largely because of its ability to monetise every impression through its unbeatable number of advertisers. Across a large inventory I’d back AdSense to perform best against any network, but that doesn’t mean that it will win every impression.
The best results can often be achieved by combining networks in way that encourages them to compete on price. Using an ad server (such as the excellent and free DFP Small Business) you can load networks in based on their true performance and challenge AdSense to beat them on a per ad request basis to ensure the highest price for each impression.
By replacing AdSense tags with network tags you are offering the network the chance to decide who gets what ad requests. By using and ad server you are making that choice based on performance metrics.
Calculating the true Ad Request RPM
The trick to doing this effectively is to set accurate prices for each network you traffic through AdSense. This should be done based on the per request price that they actually achieve – not the price they pay for the impressions that they actually choose to serve. After all, even a £100 CPM is worthless if they take no impressions.
If you are passing any unfilled impressions from the network back to AdSense then you also need to factor in the value of that passed back inventory. This is most easily done by sending your pass backs to a separate channel for each network.
Actual CPM = (Network revenue + pass back revenue) / DFP Impressions * 1000
Simply add the network as a price priority line item based on that calculation and have an AdSense line item compete against it. Remember to update the actual CPM on a regular basis to ensure that bids are being calculated at the right price – daily or weekly depending on volume.
This seems like hard work
The popularity of AdSense is due in part to its easy “fire and forget” nature. Managing yield from multiple sources isn’t hard with an ad server, but it is a much more involved process than sitting back and allowing AdSense to do all the calculations for you.
The potential payoff is the security of broader revenue sources, potentially higher earnings and the ability to respond to opportunities. Of course, if you like the sound of the benefits but not the extra worry and work involved you can always talk to us at OKO Digital.
This article was originally published here in October 2014, but was last updated and republished September 2017.