So how hard has the recession hit online advertising? The growth of the industry has grinded to a crawl in comparison to almost every consecutive year going back to 1996. Some say 3%, some say 6% and some even say 0%. But not all sectors match this trend. Search is projected to push a 15% growth in 2009. In general, affiliate and lead-gen spending is also expected to continue to grow at a similar pace. Yes, the recession is impacting spend levels this year, but it is how advertisers are spending their dollars that is really what has changed. Continue reading
When buying CPA advertising on networks, it’s all about the acquisition of something – a click, a lead, a registration, an application or maybe a sale. So we don’t care about the impression right? We don’t care about the click-thru rate, only the end-result performance of the ads we run. We just care about the yield we are getting.
When we buy CPA advertising on a network, all we care about is that our $10,000 is generating 2000 new leads – or whatever CPA you’re targeting – and that you are able to maintain a positive ROI trend.
If you can keep the CPA targets in sight, buy leads all day long, right? Go to as many sites and networks that will do the deal and off you go.
But what about the backside? Why is that when you first run a CPA campaign you notice a growth trend? You are able to grow the lead-generation from, say 1,000 leads to 5,000 leads but then things stall out. Networks stop growing the lead flow and some even start to contract. The CPA starts to inch upward and you find yourself having to broaden your buys to more publishers and networks. What is it that’s happening here?
When you buy CPA you’re not focusing on the performance of your creative on the click-thru level, only on the acquisition level; which creative generates the best volume of acquisition. But that does not reflect the impression count. The networks and publishers are looking at click-thru rates so that they can optimize the utilization of their impressions. Your highest performing banner by lead-generation could be your lowest performing banner by click-thru and hence is getting little play on the network; whereas the lowest performing banner by lead-generation gets the highest performance by click-thru and results in a higher impression count.
This can lead to the stall-out of your lead flow growth rate, the potential contraction of your lead flow and a creep in the CPA on a specific site or network. The longer you advertise with a site or network, the more they optimize their allocation of inventory with your campaign. So they give you fewer impressions with the same acquisitions and still make the same amount of money off of you. You don’t notice the difference, or the change is marginal, and of course the impact is that your potential for growth is capped.
What do you do?
You need to have the ability to measure the effectiveness of your advertisements at the impression level so that you can determine which creative is performing well on a click-thru basis so that you can optimize the creative you start with. If you want to continue to grow CPA campaigns, you should be starting with the click-thru value of the ad and measuring all the way through to the acquisition even though you are buying on a CPC, CPA or CPL basis.
A couple of options: You can use an ad server to serve your ads, and you will be able to measure impression counts, click-thrus and post-click events (acquisitions). You can rotate the ads and maximize campaign performance with optimization. The problem is you don’t control the number of impressions you will be getting through a CPA or CPC campaign and so your ad serving volume could be off the charts. Maybe you could negotiate with the ad server and get blocks of ad serving units at low rates so that you can run on networks to keep costs low. It would be kind of like buying the bandwidth instead of buying impressions. If you are buying 100s of millions of impressions, you might be able to negotiate a different deal with an ad server rather than the standard CPM. Maybe. Even I don’t know if I could make that happen.
A second alternative would be to get the networks to cap impression delivery so that you can control the ad serving costs. I have clients that do this. They run CPC or CPA campaigns and have limits set with the networks because they are using an ad server. This keeps the ad serving costs in check, and enables them to measure impressions and click thru. It also gives them the ability to control banner selection and rotation and provides the consolidated reporting that most advertisers prefer.
A third alternative and probably a better way to go, is to partner up with a network and get them to work with you strategically. If a network can be brought to recognize that by sharing impression data with you, you can measure the effectiveness of your ads. You obviously have to commit to providing creative that will better perform based on the knowledge gained, but your effort will be empowering the network to better optimize your impression delivery. This will support the growth of your campaigns and your spend level on their network.
Networks don’t usually provide impression counts unless you buy on a CPM basis. If they did, you would see how crazy your impression counts were, how low your CT rates were and how they fall off pretty quickly. But if they did share the data with you, you could make it worth their while so long as you applied the data effectively.
Talk to your network partner and get them to work with you. Buy CPC, CPA, CPL and incorporate the impression data into your metrics. It’s okay if you get 100M impressions and your click-thru rate is 0.02% (20,000 leads) and your CPA is $20. If you know that, then you can generate creative that targets a 0.05% CT% and increase your spend level with the network by 50%. Additionally, you may also manage to drop you CPA by a couple of bucks to as your creative improves CT rate. It’s a win-win for you and the network.
Reactionary with Insight.
_uacct = “UA-980395-1″;
Mollie Spilman, Advertising.com’s Chief Sales and Marketing Officer interestingly suggested a holistic media approach in her iMediaConnect article, Search and Networks: Better Together. She urges an advertiser to combine search and network advertising together to create a more powerful, cohesive campaign. Her position is that with the rising costs of search (see my post Banners vs. Search), banner advertising can be a good offset to balance the investment. But her really convincing argument was found in the following statement:
“…while search is limited to generic text listings, behavioral targeting enables you to follow up with a more compelling sales message, using ads that are tailored to the user’s online activity. Rather than simply hoping that the consumer will return to your site after researching or comparing prices across the net, you can take action– serving up your most powerful message or promotion all across the network.”
So if you combine BT with search you can tag someone the first time they come through from search and then target them in the future based on that search. Provided that you are strategic enough to couple both the cost of the search and the cost of the BT back to the eventual sale, measuring the CPA accordingly, you may find the approach to be more effective than just one or the other on their own.
Now, Mollie’s shameful plug of a network as the Chief Sales and Marketing Office of Advertising.com is totally acceptable. But I would love to see some data behind this idea as it is compelling. It would be hard for most advertisers – let alone many agencies – to track and calculate the CPA under this method, but if someone is tagged from the initial landing page based on the search term, recognized later through the BT recognition process, retagged, and then tracked through a conversion process you could have a pretty solid argument. Would it be a cost effective acquisition? It would certainly be an accurate measurement of the acquisition. It would be a more accurate depiction of the holistic investment then just looking at search or just banner advertising with or w/o BT.
You see I have always said that the technology is here. My blog is committed to finding the technologies and then proposing other ways to using them to do more. So here goes…. First party ad serving can do this too, but we may be able to eliminate a step or two in the process.
Let’s assume that when someone lands on the site from the search term they are cookied with a first party cookie rather than a third party cookie – like the Advertising.com cookie. Then let’s use first party ad serving, like DirectServe to do behavioral targeting, using the first party cookie instead of the third party cookie. What will happen is that when the browser is recognized the next time they are encountered, the cookie recognition will be based on the advertiser’s own first party cookie and the ad served will still be event-based. So far things are pretty much the same. But when the individual clicks on the ad and is driven to the site, the advertiser can read their own cookie and will know the search term that generated the lead, the banner that generated the click and the site that generated the lead. So when it comes to calculating the CPA, all of the data will be aggregated into one place and will be readily available.
If we build off of Mollie’s model – which I originally proposed, not Mollie, so don’t shoot her – you would have to calculate the ROI from the search campaign, calculate the ROI from the network banner campaigns and then synchronize the two with some kind of algorithm.
I am liking this idea a lot. I think that Mollie is right here. I think that a great way to bring down the rising cost of search is in fact to combine it with BT banner advertising. And maybe using a network would be an affective medium since the inventory will be vast and the CPMs are lower. Premium networks also do get some good inventory as I have mentioned in the past. But we definitely need to be able to have a way to calculate the combined CPA of the lead which comes first through search and subsequently through BT and the banner. I’ve got one easy way to do it.
_uacct = “UA-980395-1″;