So, there were a lot of requests for how to buy on a smaller CPA and get it profitable on media buy sources. So, Im going to show you how I would go about limiting my risk for testing and then give an example of a live campaign we ran a while ago to show you guys my thought process to optimize a lower CPA offer to profitability.
So going into a smaller CPA campaign, the first thing you want to keep in mind is that youll be getting a lot faster data, so you constant optimization is NEEDED and possible. With high CPA campaigns, you need to spend more to get significant data/conversion data, because it costs more to acquire each conversion, but with smaller CPA campaigns youll get a lot faster results, so optimizing right off the bat is a must. Lets start talking about self bid CPC platforms like AOL, Pulse, Adside, Yabuka and those kind.
So to run these, you need a few things, your ads, your image (if your on adside or pulse), and your placements. Im going to use a very small old biz opp campaign (almost a year old now just fyi) for a campaign like SMC that used to exist page was similarish to THIS. Our CPA was $5, and we were direct linking. Our buying platform was adside, so we started with our ads. We used WhatRunsWhere.com to compile a few text ads we thought were doing best at the time, and then we improvised and spun off a few of our own versions of the text ads, but they still used the main parts that we found were most popular amongst a large sample size of ads we looked at. Here is what we used:
Now Hiring: (Work At Home)
Job Requires: Basic Computer Skills Your Own Home Computer. Pays Fast
Work From Home
Work 3 – 4 Hours/Day, Your Schedule Make $5K/mo, Easy Computer Work
For adside you have to have a short blurb because its a 2 step rollover process, but those were just generic stuff like Work At Home Now and Hiring: Work At Home. Ive highlighted the ad copies because across most platforms you only have that adcopy to attract a visitors eyes and get them to your landing page.
Lets say you dont have whatrunswhere.com (you should ( product plug here)). You can also find inspiration from whats running locally for you, go to the type of sites you expect to advertise your niche and look at whats running, take notes, what appears in common across various similar ads. Then write multiple starting testing ads that are different utilizing these traits, you can then optimize down and cut whats not working very quickly to find winning ad combinations.
Ok, so we have our general ads and are all ready to run besides our placements, how do we get those and make sure we arent just going to throw up something thats going to haemorrhage money. Again, I like to use whatrunswhere to look at the placements the existing advertisers are using, they dont have to be on our network, but by looking at the demographics of these placements on alexa, quantcast and compete and comparing the demographics we can get a decent idea of the type of people we should be going after. You can also do this without whatrunswhere again by tracking down placements one by one and comparing them, but it just takes a bit longer (works just fine though). Then I go to my network rep (if you dont have one, request one, they are usually invaluable help and can do a lot of the annoying setup for you), and bring them these demographics, theyll usually draw up a proposal for your campaign, you can then reference these sites to make sure youre limiting your risk by matching demographics. You can also do this by looking through network placements and checking high volume ones to see which you may have a chance of succeeding on.
NOTE: Please remember while network reps are helpful, they want your money and they want you to spend, keep this in mind if they seem overly excited to run your campaign, even if you dont succeed they still win because youve spent money on their network.
Ok now we have our placement list, what should be bid? Its all relative to the CPA of our offer, obviously if we have a $2 CPA offer, were not going to start bidding a $1.50 CPC across the board, that makes no sense. For this campaign, we had a $5 CPA, so my max bid was around the $1.50 range, because it was easily converting at around 20-40% across various other placements/sources we were buying from at the time. So, we have everything and were ready to go (make sure youve placed the network pixel in addition to your own, because they can use this to help you optimize).
For adside, we launched but only got to run for a few days before the campaign went down on the advertisers side, but I want to use this data to show you WHAT WE WOULD HAVE DONE to optimize this to profitability.
So, as you can see we only spent around $250 with a majority of the spend being used on jobs-to-careers.com. First a note, their pixel under fired a bit, our real conversions were about 1.5-2x higher (this is why we place both internal and network tracking pixels always – so around 45-55 conversions making anything under $8 profitable for us) Using my optimization rule of 3x spend before I touch anything, I only see about 6 targets I can work on. For electronista.com and mcclatchy.com their CPAs are both way too high, so Id cut those (even with the under firing their CPAs come out too high). The thing about a low cpa is you have to not be afraid to cut and tweak stuff quickly, you get a lot of usable data because of how low the CPA is so you get a lot of conversion data to optimize off of. Id cut adbr obviously because its spend $17 without a conversion. Id be close to cutting ACM Local and Macnn.com but they havent reached my optimization spend threshold yet, so Im just keeping an eye on them until then (ready to cut or tweak their bids). Since there is a low spend that is heavily weighted towards one placement, lets talk about that one. The CPA appears high but with under firing is actually within a acceptable range, here I can try to get more volume by scaling my bid up a bit, or look for similar placements on the network to test in case I missed those to try to get some more volume. There seem to be a few other targets with a few conversions and lower spend that are profitable so far, so Id let those run and just keep an eye on them. All the targets with under $10 of spend, just need more time to run and get data before I cut them, I could raise bids on them if I want more data quickly. I know there isnt much data there for examples (sorry, the offer went down as I said), but the point of this wasnt to out campaigns and just give away stuff, but more to show thought process and method.
When I optimize I compare prosper (or bevo, cpvlabs whatever you use) with the network data to try to get a match and a clear idea of whats actually happening. Pixels always break so having 2 set in place gives me a better picture of whats actually going on with the campaign.
For any other low CPA offer, the process is the same. Be sure to look at whats running on the network, if its a popular offer like IQ and its not running anywhere on the network theres probably a reason (its a popular offer, and people test it everywhere). If youre unsure, test it out, set a small budget and lower bids to try to see if it has a chance of working and if you get anything out of your test before diving in. Testing multiple things and seeing what sticks is a viable strategy to find winning offers, and its easier with small CPA offers, because a test giving reasonable data can be achieved with a lot smaller of a budget.
Also, remember certain platforms click better with certain people, it all depends on what rep you got. Personally I didnt love Adside (or my rep there at the time), and I can tell you right now, Adsonar is my favourite platform (before there are comments asking), but for each person what they like best is up to them so get out there and TEST. Most of you are going to read this and probably do nothing, you dont make money sitting around, you make money taking risks and testing. You can obviously be smart and limit those risk (*cough* WhatRunsWhere (shameless product plug) *cough*), using software and good research, but in the end theres always risk involved if there wasnt everyone would be doing it.
If this is received well Ill throw together something for display based advertising similar to this. Hope this helped some of you.
Disclaimer: this is a past campaign, is no longer running, the offer has gone down, running any data shown here is at your own risk.
Brilliant post. One more thing to ad. Once you see the placements that work don’t be afraid to go direct, cut the middle man and build a long term relationship with that pub.
I had a biz opp campaign optimized to 80% ROI on Adsonar, took it to Adside, spend almost $1k, not a single conversion. Go figure.
Awesome post, keep em coming
What would you do if your cost/conversion wasn’t below $8.00 for any of the sites? Kill the campaign/test new URLs/new creatives?
If that was the case, yeah, I’d give it some more budget, lower bids and if I couldn’t get any targets below the profitable threshold, I’d probably kill it and move on to things that make me money
i know this thread is an old thread but have to bring it back as i have a question lets say the payout was allot higher say $45 does 3 X spend rule still apply for example if my spend was $80 on site a with no conversions would i wait for the spend to be 3 X before i cut off the site
I personally would – hence why if I was to run a much larger payout offer, my testing budget would be larger too
Thank you for the insight.
The Article Published IN 06-05-2011 04:30 PM