SiteScout Display Ad Follow Along (40)
Right then, I came here to learn about display advertising, and I've spent the last month getting to know SiteScout and making campaigns there. Time to do a follow along and try and get serious. There's a big lack of media buy follow alongs here, so I hope that some of the veterans will chip in with advice and comment, I certainly need it!
I started this campaign a couple of days ago, so I'll catch up in this post. Sorry, it's going to be a long one.
Quick Backstory
A month ago I started off running a few CPL offers, mainly just to learn the platform. My initial $500 deposit and no conversions later, I did a dating campaign. I don't do dating campaigns in other media, so I don't know why I tried here, but there you go. Anyway, I started getting conversions.
Long story short, I put all my effort into dating, and spent a couple of grand testing a ton of ads and a fair few targets. Got a few campaigns starting to show positive ROI (just), and then had all my creatives retroactively pulled after a network complained about them to SiteScout. Totally understand SiteScout's position, they have to run a squeaky clean operation, so good on them for killing my ads so quickly.
I guess I could have tried another angle at dating (and probably will one day), but it's a battle getting stuff approved, and I can't be doing with that. I know there's money to be made there, but I prefer campaigns that have a little bit of staying power and don't require you to be constantly fighting to keep them running. I figure it's the same effort to make something clean and safe, so why not do that.
The Campaign
Which brings me to my new campaign, penny auctions. Not picked out of the blue, I'm doing this on a recommendation. The offer is Bid Cactus on Neverblue. Reasons for choosing this: 1) Neverblue is my preferred network, I've been with them for about four years and they've always been professional. 2) The offer page doesn't suck too much. 3) This is the main one - the offer is available both as CPL and CPS. Given a punter has to sign up (triggering a CPL lead) before they can ever buy a bid pack, running CPL means I should get conversion data more quickly.
My plan is once I have some creatives and targets that are working, I can then split test the CPS version against the CPL (and split test other offers eventually).
Day One
I started this campaign running 30th November. I picked 8 network supplied creatives, four each of leaderboard (728x90) and medium rectangle (300x250). Then I made four creatives of my own in both sizes. Uploaded all 16 for approval.
My dating campaigns all worked best with 160x600 wide skyscrapers, but the initial target batch of sites I was looking at here had very little in the way of inventory for that size. Wanting to keep the number of variables small to start with, I figure these two sizes are enough. Besides, every additional creative costs more to test.
I picked a few sites based on AdPlanner, and let her rip, with a daily budget of $50. All targets had an initial bid of $1.
Results end of day one:
Only one target got any traffic. I won 114,176 impressions (out of 186,691 bid), 16 clicks (0.01% CTR overall - yikes) and no conversions. Spend: $36.38
I paused the campaign before the before the budget was spent, as it was pretty clear this wasn't going anywhere fast.
Day Two
Creatives: I looked at what got clicks, and made a couple of variations of those. Then I made some completely new ones using a different angle. Sadly, the best performer was a network ad. Damn.
Sites: The one site that got traffic gobbled up the budget. I reduced the bid significantly to stem the flow.
Then I picked a load more targets and added them to the campaign. Again I used AdPlanner with the SiteScout list to pick targets visited by people who go to penny auction sites. There weren't many, and the correlation was low. So I added in a few more that I knew got lots of cheap traffic from my dating campaigns.
Results end of day two / start of day three:
Ok so this is bang up to date. The campaign ate my $50 budget in a couple of hours. By then it had got two conversions and I was getting data, so I added another tenner, and got one more conversion. Now in an ideal world, I would love to run this 24/7 and then day part, but my budget isn't that big! So before I went to bed, I added another tenner to the budget so it could run later. That got one more conversion.
By the time I woke up this morning, with the time difference the campaign had already been running a few hours after midnight. It had spent $55 and got another two conversions.
Total spend to date: $153.88, total revenue: $18. ROI: Eugh...
Now I have a little bit of data, I would like to try and make some informed decisions about the direction, but I fear there's not enough data to go on yet. Here is the site list for the campaign so far:
As you can see, I paused target 216032 because that's where most of the money was going without results. The CTR is very low there (I know CTR isn't the most important factor, but it's still at least an indicator of interest), and no conversions yet. I was able to reduce the bid to 22 cents and still get traffic, so I'll let it run slowly again when I've optimised the creatives a bit. For now I'd rather concentrate on targets showing more promise.
And of those, 217295 is the most promising of the bunch, with an eCPA of $12.18. Still pretty yucky, but it's a start. I spent $48 there to make back $12.
216031 I paused for similar reasons to the other one - too much traffic too quickly, eating up the budget, and a conversion rate 4 times lower than the best site. My plan for that one is the same as the other, run it slowly when the creatives are more optimised.
Here are the stats for the creatives:
(Note: You can't resize the creatives list in SS to see the whole list in one go. When you export the list, the dimensions don't get exported, which is why they're not shown).
Random thoughts, all to be taken with a large dose of salt given that six conversions is nowhere near enough to be statistically significant. But still….
- The current leader in terms of CTR, was the worst performing ad for most of the few hours of the test. (I know, CTR isn't the most important factor)
- The conversions didn't come from the ads with the best CTR, which was also my experience in the dating campaigns. Indeed, half the conversions came from ads with dreadful horrible embarrassingly low CTR. I was getting 1% - 2% on dating, these are in a different world!
- 3 conversions from leaderboards, 3 from medium rectangles.
- 4 conversions from my own creatives, 2 from network supplied creatives. Clearly, I rock at making creatives. (If only…!)
- Ad 45130 and 45126 between them account for half the conversions. They are the same ad, one's a leader board the other a rectangle.
Actions, bearing in mind this isn't statistically significant data:
- I'm pausing all network creatives except the ones that got the conversions. One is already gone after 13k impressions and no clicks.
- Where I have two or more versions of the same creative, I'm killing all but the top performer of each in terms of CTR. If I wait for conversions from every ad, it's going to cost me a fortune. I need an indicator and CTR is it.
- I will make variations of my top performing creatives. I'll also add in some completely new creatives.
- I'll pause all sites that got no traffic. If I can't get impressions at $2 (I upped a few bids) then they're too expensive.
- Not entirely sure what to do in terms of new sites to target. Suggestions? Obviously anything similar to the top performer, but it's quite specific and I've already got two other very similar sites in there.
The problem with this offer is that it appeals to a wide demo (that's a good thing too, lots of potential traffic!) It makes it hard to choose targets. Without over-thinking this, what should my strategy be?
The other thing I'm planning on is adding a couple of landing pages. I have two ideas very clear in my mind, just need to get some time to make them and throw them into the test.
Right, that's quite enough typing, time to get on with it.
PT
I tried the same thing, running penny auctions on SS. Already down $200, so i had to stop the campaign until i think of something 
Probably it would be better for me to stick to 1-2 websites and not go for a whole bunch that will suck all cash in minutes.
But the thing is that i didnt get a single conversion, so i have no idea on which websites to stick.
don't use sitescout but if you can why don't you first choose the site and then run a few offers of the demo you target. start with sites with 300x250 ads, rotate five offers with few creatives each and see what happens.
That's another tactic I plan on using later (I use it a lot in PPV). But for now, I want to get this one campaign down and showing a return
I'm new to Media Buys, so forgive me if this is a stupid question, but wouldn't Google Display Network on CPA be better to test with on small scale then go out with the tried and tested creatives further afield with SS and it's volume scaling?
I lost all my AdWords accounts to the Google ban hammer years ago. I know there are ways to get new ones, but I don't want to jump through hoops to then have to play their "we hate all advertisers with a vengeance" game. I don't see what the advantage would be. Why add in an extra step? What's the benefit to first running on a different network to the one you really want to be running on?
Day Three Update:
A few actions taken yesterday, as per the previous days notes:
- I made variations on promising banners, and added some brand new ones.
- Added 5 new sites. No feedback in this thread on ways of choosing sites, so I picked stuff that looks like it might work based on my own criteria. (I guess making banners and picking targets are the "secret sauce" of display advertising, so understand if nobody wants to share )
- Killed sites and banners as per yesterdays notes.
- I also set up a day part rule so it doesn't run midnight - 6am EST. That way I can keep an eye on it rather than wake up and the budget for the day is already gone. Next week I might make a few copies of the campaign and use day parting to run them at different times. There's no hourly budget option like in LI, so this is the only way to spread the daily budget out without manually stopping and starting all day. (Note to SiteScout: Please add an hourly budget option!)
Because most of the budget was already gone overnight, I only ran an extra $20 for the day. Some of that was swallowed up by a couple of the new sites, the others weren't yet approved. Got one more conversion, and that came from one of the new banners. But one conversion is totally meaningless. Total spend for the day was $74.16 with revenue of $9.
Actions for today:
I think I need to come up with some fixed rules for when to kill sites. Random thoughts off the top of my head:
- 10k impressions no clicks
- 2 x conversion rate but no conversions (e.g.: 5% CR = 1/20 clicks x 2 = 40 clicks)
- 3+ conversions but eCPA > n x payout
That's something to think about today. Same for creatives. One thing I learnt from my dating campaigns was at first I was killing creatives too quickly. I went back and re-enabled some of the earliest ones I'd paused, and they turned out to be winners. Still, I'd prefer to kill them too quickly than not quick enough, because most of the time the winners show themselves pretty early on.
I've lowered my bid on the existing targets so the traffic slows to a trickle, to give the new targets a chance. I'll run them for a few hours later, and keep a close eye.
Other than that, I need to get these landing pages out of my head and onto the screen, ready to whack them up for approval at the start of the week.
PT
I've been trying sitescout and am by far no expert....but I did get one campaign breaking even actually a little in profit after 5 other failed campaigns, and I've learned you really have to kill targets and banners that are just not producing, they may be producing CTR but if no conversions kill them...and play with the bids a lot, because some targets end up profitable with lower bids...problem I'm having is scaling up...
Hey polishedtured (sounds just not right greeting someone like that!!) 
Nice follow along so far, ... lets hope some of the Mods, or more experienced Media Buyers jump in and offer you some solid advice.
I'm about to dive into SS as well, so will be hoping you turn this bad boy profitable with some helpful pointers from the others.
Just a thought on the banners, ... I presume (if I remember correctly, that you can't do flash banners..???), but have you tried animated gifs etc.
Rather like Polars 'jumping' or 'twitching' banners, which would drive you nuts if you didn't click on it to escape.... That may help rocket your CTR.
Do you have WWR or Adbeat..? Maybe you can look at looooong running banners for similar penny auction sites - but I think for sure that animated banners
will certainly push your CTR through the roof.
Good luck!!!
Hi Polishedturd, great post. I totally agree that there should be more media buy related follow alongs. I currently have my testing funds tied up in SS and am confident and positive that there is much money to be made there. I haven't made any yet though but am confident...not so much on my ability but on the SS potential. Like you say, with banner advirtising, banners and traffic sources on there are closely guarded secrets but if I discover anything else of interest I'll be sure to share it on here. One of my banners so far has 0.35% CTR (3740 imps / 13 clicks / no convs). The spend so far is less than $3.00 and the offer payout is around $50 so I reckon there is potential and more testing is required. I've just submitted another campaign using a different offer at this point because I think that the offer could be the issue. Still a rookie myself though so its a big learning curve for me but I suppose you just gotta get stuck in and see what sticks. As far as my banner goes, its reasonably simple, to the point and has the brand logo on it. The main text just has a question (with an angle) that hopefully gets the visitor thinking. My initial plan was to get the ads approved and not try and pull the wool over their eyes. I respect SS so I won't try anything funny with them.
Originally Posted by condinet
they are a relevancy engine and as such they only place your ads in relevant places....so its not a good judge of whats gonna work and whats not.....almost anything will work on google
I kill banners quick on SS.....no more than 10k per banner most cases...sometimes even less if the offer is a low payout....
as far as sites are concerned I would def focus on just a few....and not tackle to many....
I am happy to give more tips but right off the bat I would say that the ctr on those ads looks kinda low for the payout....
remember the goal of the banner is to sell the click not the product.....
Originally Posted by 77erock77
Originally Posted by extremesg
Originally Posted by polarbacon
Originally Posted by polarbacon
The same thing happened with my dating campaigns, I had banners getting 2%+, but the conversions were all coming from banners with 0.2% CTR (and they were profitable). The problem is you have to run those a long time to see if they work.
That's why I'm trying to come up with a set of "kill rules" that give each banner a chance to prove itself.
All the advice in this thread so far is suggesting kill early, so I think that's what I'll do. Thanks for the feedback so far everyone!
PT
Day Four Update:
It's the weekend, but I can't work Monday so I'm doing a bit on the weekend. Besides, I need to test this baby every day of the week and buy myself some data. So:
- I moved the day part to 9am-Midnight. Figured few people are going to be signing up for auction sites before 9am. Like looking for the Higgs Boson, better to concentrate the search (and budget) in the areas of higher probability.
The campaign leapt into life as scheduled, and I let it run for $20. Got one conversion almost immediately, from one of the new targets (and one of the new "variation" banners), but nothing more after that.
One of the other new targets that I had high hopes for got 0 impressions even at $5 bid, so that's another one for the bin. There are quite a few like that, don't really know whats going on - they're listed as having plenty of impressions in the right ad sizes, and a low estimated bid, but I get 0 impressions bid (bid! not even won...) even at silly prices. Oh well, kill them and move on.
- I went through all the targets that are still live and checked the placement report, and killed any placements that are sucking money and not generating clicks, of which there were several. In fact I found that my best performing site so far has one placement that is actually profitable (with the standard disclaimer that there's not much data to base that on) and one of the others is an impression-sucking vampire, so I stuck a stake through it's heart.
- Ran another $15 of traffic a little later. Killed off one of the new targets because the CTR was dismal.
- Later on ran another $20. Lowered the bids on all but the most promising target.
- I built the first of my landers, but as SiteScout don't do approvals on the weekend that will have to wait a bit before I can test it.
Next actions:
- After killing the non-performing placements, one url shows particular promise. I'll split that out into it's own campaign so I can test it more thoroughly and with it's own budget. It also means I can tailor creatives a bit for that target. I saw in previous campaigns that sometimes a banner rocks on one target and sucks on another, so the ultimate aim here is to end up with one target per campaign.
- Another round of banner trimming, and make some new ones.
- Add some more sites to the main test campaign. Back to AdPlanner + the SS list to find "also visited" type targets for the most promising url.
Not much else to add really, this is where it gets into routine. Cut non performing sites and creatives, make new creatives, test new sites. Rinse and repeat daily.
PT
Originally Posted by polishedturd
On display its RPM(rev/1000) or ECPM(earnings/1000) is everything.....your CTR/CVR is just a part of that calculation.
when I say sell the click what I mean is you need to give them a reason to click....
and there needs to be flow to the whole thing.....something that makes sense when you dump them on your lp....
and thats where alot of affs/camps fail there is no flow....
people are like sheep...you need to lead them down a path...as stupid as it seems sometimes by connecting even the most basic of dots.....
the most basic of example follows....
hot girl on banner.....headline says "browse my pics" (is the banner selling the click? or the dating site?)
I click on link I see a dating site...
I am confused...who is this girl? Where is she from?
that causes anxiety.
I don't sign up
or
hot girl on banner.....headline says "browse my pics"
I click on a banner
I see a simple lp witha hot pic of the girl from the banner and some nav buttons to see the next pic...
I click to see the next pic....
I get taken to the dating site....
This makes sense....I want more
I sign up...
Think like your lead....ask yourself "Does this make sense to me?"
I know this example seems very basic....but thats how you need to think....think like a 2 yr old when setting up the camp....
Cool, makes sense (that's what my lander is doing so that's good news!) Just need to wait for it to go live and see how that pans out.
PT
Day Five Update:
- Made some new banners, and some variants including some animated
- Killed ALL the targets in my test campaign, and started that over with new sites. Killed a load of banners too.
- Made 2 new campaigns, each with the same target (the one which had a profitable placement from the test campaign). One of these is direct, the other uses the landing page.
As this was done on Sunday, obviously nothing went live yet, so no stats to report. Hopefully everything will get approved today, but in the past I've found often nothing gets approved on Mondays as they have a big backlog.
It completely sucks that they don't work at the weekends. I submitted stuff at the weekend and could have done with an approval seeing as it was saturday. They must earn a decent amount of money so surely they could hire someone to approve stuff on the weekend even if its just for half a day.
Day Six Update:
I was away all day so couldn't do much (well, anything really). I just made a copy of the test campaign but going via a landing page, so I can retest the other targets that showed conversions but with a lander, without having to set up separate campaigns for each. That got approved, but bizarrely, the target that I'm running as a separate campaign got denied in this campaign. I saw quite a lot of inconsistency like this from SiteScout in my dating campaigns, I guess it goes with the territory.
Because I'm stupid, and rushed setting up the single URL campaigns, I forgot to select only the placement that was previously profitable. So the first 30-40 clicks to those campaigns were to the unprofitable placement. Therefore the following stats are somewhat skewed by that, as most of the budget went on a placement that should never have run.
Here are the stats for the single url direct link campaign:
18k impressions
55 clicks, 0.31% avg ctr
0 conversions
And here is the lander version:
20k impressions
69 clicks, 0.35% avg ctr
4 LP clicks (5.8%) (really only 3, see below)
1 conversion
So right now, both single url campaigns are performing worse than when that url was part of the larger test campaign. Still not sure there's enough data to go on though.
Here are the banner stats for the landing page single url campaign (the direct campaign was using the same banners, with similar stats):
The banners themselves are getting slighter better CTR when I focus on this one site. But clearly the landing page sucks, because it only got 3 unique clicks (someone clicked twice).
Yet again, the only conversion came from a banner with one of the lowest CTRs.
The new targets in the main test campaign bombed totally (one shows perhaps a little promise…)
Actions for today:
- In the main test campaign, I'm killing off all targets save a couple which didn't get a chance to run because the crappy ones ate the budget. I won't add any new ones today, I'll let those that didn't get a chance yesterday run for a bit.
- Cull the poorest banners in all three campaigns. Make some more new banners.
- Make a totally different style landing page and run that against the current lander.
I'm a littler concerned that I'm testing different sites, banners, and landing pages, and the only variable I'm not yet testing is the offer itself. Anyone know of any good penny auction offers that have a CPL and CPS version available?
PT
Originally Posted by polishedturd
Glad to see a media-buy follow along, was looking at sitescout a long time but my inability to produce high-CTR banners and budget-limitation put me off so far. I really wish you the best with this PT!
/subscribed
Day Seven Update:
A week in and time to recap and take stock of what's what. But first, actions from yesterday:
- Fixed an issue with my lander when viewed in IE. Because of a formatting problem most of the important content was appearing below the fold. I did know about this already, but didn't have time to fix it before.
- Culled ads. Made more, added them to all campaigns.
- Found a better landing page (direct) for the offer, so switched the direct campaigns and my own lander to link to that.
- Re-ran some of the previously killed urls against the new version of the offer.
- Checked all placement reports for vampires, but none found.
- Didn't get to make a new lander as I was supposed to, so a slap on the wrist for that. However, I did make it in my head, so today I just have to get it out of there and onto the screen.
Another url started showing some vague potential. I noticed it's one of those urls that appears several times in sitescout's site list, and I've only been running one version of it. So I broke that url out into it's own campaign, and added every instance of it that SS have on offer.
At this point a recap is in order, because at this point there were five versions of the campaign running, and it's probably quite confusing if anyone is reading along and trying to follow this! These were the running campaigns so far:
- Direct linking main test campaign. Multiple targets. The place where I'm running new targets to see if they have potential.
- Lander main test. A copy of the first campaign, but linking to the lander instead (duh!)
- Single URL "alpha" direct link. One target.
- Single URL "alpha" lander. One target.
- Single URL "bravo" lander. Multiple instances of one target.
As I'm creating new banners, I'm bunging them into all the campaigns.
My single url "alpha" campaigns got more expensive to run yesterday, I guess a new bidder came along, because suddenly I had to increase my bid by 60% in order to get impressions.
Single URL "bravo" bombed totally, with 24k views and only 5 clicks, so I'm killing that, bringing it back down to four campaigns. Here's where I am at stats-wise with those since the start of this test a week ago:
Stats For The Week So Far
- Direct Link main test.
460k impressions, 366 clicks, 0.08% avg ctr
$265 spent, 8 conversions ($24), -91% ROI
- Lander main test
53k impressions, 50 clicks, 0.09% avg ctr
8 clicks on lander, 16% lp ctr
$32 spent, 2 conversions ($6), -81% ROI
- Single URL "alpha" Direct
26k impressions, 93 clicks, 0.35% avg ctr
$32 spent, 2 conversions ($6), -81% ROI
- Single URL "alpha" Lander
34k impressions, 142 clicks, 0.41% avg ctr
3 clicks on lander, 4.1% lp ctr
$43 spent, 3 conversions ($9), -79% ROI
This turd is proving difficult to polish…
But I'm encouraged by a few things:
Reasons To Be Cheerful
1) Although the CTRs look rubbish, the average figures above don't tell the whole story. As I'm testing new ads, I am getting single-ad CTRs of well over 1%. However, those are mostly meaningless (see point 3 below).
2) The best single day ROI is -57% which while obviously horrible, is better than the average over the week, so at least things are moving in the right direction.
3) Out of 15 total conversions so far, 6 came from variations of one ad. The others were varied, with no single ad-type getting more than 2 conversions. This top performing ad in terms of conversions performs equally well in the direct and the lander versions of the campaign. It is one of the worst ads in terms of ctr, by quite some margin.
4) On the "alpha" single url campaigns, and on the main lander campaign, every ad that has converted is individually profitable.
I've spent $372 so far, to make back just $45 ($327 loss, -87% roi). I'm trying concentrate more on the data than the spend, I know it costs money to test. Even so, clearly it's not possible to go on spending endlessly without some progress, and right now there's not enough. I need to get tougher on killing stuff that's just not working.
Actions:
- I'm killing the main linking campaign. It's the worst performer with no individually profitable ad.
- I'm killing every ad that hasn't yet shown a conversion, regardless of whether it has a ctr of 0.01% or 1.5%. Technically that should make all three remaining campaigns profitable because I'll only be running ads which on their own are already profitable. But in reality it's never that simple!
- I'll make copies of both Lander campaigns using a new lander, and test them side by side (or the best approximation; you can't rotate urls in sitescout so you end up bidding against yourself and it's impossible to get a true real-time comparative test).
- I'll run some more new urls through the remaining main test campaign. I need to get more organised about this. Until now I've been branching out from urls that are getting good ctr, trying to find related stuff. Now I'm going to pull high traffic urls from a wide range of demographics to widen the net.
I'm still concerned about the offer itself. The new offer page certainly improved matters yesterday, but it would be good to try another. Ziinga looks pretty good, and I believe there is a CPL version knocking around, but not in the networks I use. I'll look into this a bit more.
What else? Am I missing something obvious? Am open to suggestions 
PT
Day Eight Update:
- Killed ads, campaigns etc as per list. Actually I deviated slightly. I said I would kill anything that hasn't converted. In fact, I killed creative types that haven't converted in any campaign. So where I have 4 variations of an ad, if any one of them has converted in any campaign, all of them stay in all campaigns. Plus I left in two of the latest batch of banners as they hadn't had enough exposure to really test them. Glad I did, one of them is converting now.
- Made a new lander, copied the "alpha" lander campaign and ran it with the new lander. This wasn't a direct split test because the new campaign took all day to get approved, so the traffic to it was several hours after the traffic to the original lander campaign.
- Added 5 new targets to the main test campaign.
Results:
- Main Test Campaign, Lander 1
30k impressions, 56 clicks, 22 lp clicks (39% lp ctr), 1 conversion
$25 spent, $3 revenue
Killed 2 of the new urls quickly due to 0% ctr. One other got no impressions. The other two show promise, 1 converted and the other got 58% lp ctr.
- Single url "alpha" Direct Link
8k impressions, 32 clicks, 2 conversions
$14 spent, $6 revenue
- Single url "alpha" Lander 1
14k impressions, 78 clicks, 4 lp clicks (5.1% lp ctr), 1 conversion
$24 spent, $3 revenue
- Single url "alpha" Lander 2
14k impressions, 67 clicks, 10 lp clicks (15% lp ctr), 3 conversions
$24 spent, $9 revenue
There was one other conversion for which the pixel didn't fire, so unfortunately I have no way of knowing which campaign was to blame.
Conclusions:
- Lander 2 outperformed Lander 1 by 3:1 on the same target (95% confidence)
- However, the direct link version of that target had a lower eCPA than the lander 2 version ($7.09 direct, $8.07 lander)
- Low CTR creatives continued to perform best. One new creative got a ctr of 1.38% on one campaign, and 1.42% on another. Both got stupid low lp ctr (<5%) and neither converted. Conversions came from creatives with a ctrs ranging from 0.31% to 0.49%, i.e. horrible crappy ctrs.
Actions:
- Now the new creatives have had a few more impressions with which to prove themselves, I'll do another cull.
- The placement report for the two remaining new targets show that each have a vampire placement that requires the services of Buffy. I will split both out into their own campaigns, making a direct link and lander 2 version of each, staking the money sucking placements at the same time.
- Improving lander performance by a factor of three has encouraged me to make and test another lander variation.
This is still a hell of a long way off from getting to break even let alone profitable, but at least there is a little improvement each day. The EPC overall at the offer level has gone up from $0.10 at the start of the test, to $0.36 yesterday.
PT
polishedturd:
You're on the way man, just need to work on your creatives. CTR is the key to the heaven. This is how i do it: 1 site. Make it work. Add another site, make it work too. That's worked for me really well.
Are you saying you pick 1 offer, 1 target site, and then just work the creatives (and maybe LP) until that site is profitable? I'm up for that, not testing new sites means one less variable to consider.
Day Nine Update:
- Killed Single url "alpha" Lander 1 campaign as Lander 2 was outperforming it 3:1
- Switched the main test lander campaign to lander 2 for the same reason. Killed off the two remaining new targets and split them out into new campaigns, each using lander 2 (I guess those would be called single url "charlie" and single url "delta"!)
- Culled more banners
- Added five new targets to the main test campaign
- Made a copy of single url "alpha" and a copy of Lander 2, and pointed it to the CPA version of the offer.
- Set up a headline split test within lander 2.
- After most of the budget was already gone, I set lander 2 to rotate between the two versions of the offers own lander.
Results
- Main Test Campaign
24k impressiom, 29 clicks, 4 lp clicks, 1 conversion
$14.65 spent, $3 revenue
- Single Url "alpha" Direct Link
7k impressions. 36 clicks, 2 conversions
$12.46 spent, $6 revenue
- Single Url "alpha" Lander version
9k impressions, 28 clicks, 4 lp clicks, 1 conversion
$16.63 spent, $3 revenue
- Single Url "alpha" Lander CPA version
9k impressions, 38 clicks, 4 lp clicks, 0 conversions
$15.20 spent, no revenue
- Single Url "charlie" (uses lander 2)
18k impressions, 20 clicks, 7 lp clicks, 3 conversions
$14.64 spent, $9 revenue
- Single Url "delta" (uses lander 2)
21k impressions, 104 clicks, 49 lp clicks, 0 conversions
$17.43 spent, no revenue
- Lander 2 headline split test, almost even (2% reduction in clicks with the alternative headline, and only 35% confidence in result)
Conclusions
- Campaign "delta" sums up my frustration with this campaign. Here are the stats for the creatives:
Banner ctrs up to 0.84%, lander ctrs up to 80%. To me that says the banner is attracting interest and attention (no animation on any of those banners, and no tricks or deception to win clicks) and that the lander is congruent with the message of the banner, because that's getting high click throughs to the offer. And yet, no conversions.
That campaign sent 49 clicks to the CPL offer, and not a single conversion. Even the worse performing campaigns sent 19 clicks between them from the same lander, and managed 5 conversions.
The obvious conclusion to draw would be that my lander is not promising what the offer delivers, but I really don't believe that's the case. Quite the opposite, it's a long-form lander that clearly lays out exactly what the offer is, and how to sign up. Anyone who reads it and clicks through, should know precisely what they're going to see next.
- On a more positive note, "charlie" outperformed every other version on it's first outing (despite getting erroneously denied at first by SS!) Out of all the campaign versions, "charlie" had the lowest ctr and the highest conversion rate, which bears out the trend of this entire test.
- Of the five new urls in the main test campaign, the one that converted was technically profitable (and I only bunged it in there for a laugh). It didn't get much of a slice of the pie though, so more testing needed there,
Actions (Call for help!)
At this point I'm not sure which direction to take. Campaign "delta" has really thrown things into question. Namely:
- Trying to get better ctr on banners appears pointless. Every version of the campaign has proved that lower ctr banners convert better, and delta has taken it to a new level.
- Trying to get better lander ctr is also not helping much. On campaign alpha, the direct link version is still outperforming the lander version.
Ways I could proceed with this:
- Test a different auction offer (this is a given, I will do this no matter what, just need to sign up with another network who has a decent offer or two - I closed almost all my accounts years ago!)
- Test a direct link version of campaign "delta"
- Dump all these campaigns and just focus on testing more targets
- Dump everything except "charlie" and focus on testing tons of banners and landers on there
- Try and stay awake for 24 hours so I can test a full days traffic by manually stopping and starting the campaign every hour!
So my question to the vets out there (if any have managed to read this far, and if you did, congratulations, and also, why aren't you making campaigns instead?!) is this:
If you could spent $100 and 5 hours a day making this campaign work, where would you focus? Any of the above? Something else entirely?
For now, I think I am going to pause everything except "charlie", and whip up some more creatives for that. Plus there are other elements of the lander I can split test. Oh yeah, I'll break out that other url ("echo") and test that some more too…..
Back to polishing….
PT
Did you check the cvr on overall on neverblue for this offer?
Network epc there is $0.10, and I'm regularly over $0.50 now. Not sure that means anything though!
Day Ten Update:
So after all that amazing feedback to yesterdays question, here's what I decided to do 
- Killed everything except "charlie" and "delta"
- Unfortunately, SS had a glitch which deleted my day part rules in the other campaigns, so they already ran at midnight, spent their budgets, and got no conversions. On the plus side, this gave me more data on the single url I had planned to break out, and it was clear that was never going to work well so at least it saved an extra test.
- Made a bunch of new ads for charlie and delta.
- Made direct-link versions of both those campaigns.
Results:
- Single Url "Charlie" Lander Version
66k impressions, 81 clicks, 32 lander clicks, 6 conversions
$32 spent, $18 revenue (-44% ROI)
- Single Url "Charlie" Direct Link
28k impressions, 27 clicks, 1 conversion
$11 spent, $3 revenue (-73% ROI)
- Single Url "Delta" Lander Version
11k impressions, 58 clicks, 18 lander clicks, no conversions
$9 spent, no revenue
- Single Url "Delta" Direct Link
152 impressions and no clicks
(Not posting the others as they ran erroneously)
Conclusions:
- "Delta" direct link was competing against itself. Doesn't matter, the lander version confirmed that the previous days results weren't an anomaly. People on that target click, and they click through the lander too, but they don't convert.
-"Charlie" shows promise, it has the least bad roi of all the tests so far. In fact for a few hours it was in profit, and for a while was break even. Day parting may help, but I need lots more data for that.
- Conversions continue to come from ads with the lowest ctrs. To date, throughout all the tests, no ad with a ctr > 1% has converted.
- I noticed I was winning almost all the impressions bid, so I halved my bid on both campaigns, and continued to get clicks and conversions. Obviously this helps the bottom line.
- At last the landing page is proving useful. -44% ROI vs. -73%
Actions:
- Kill everything except single url "Charlie" Lander Version.
- Increase the budget on that campaign so it gets a chance to run over more of the day.
- Cull non performing ads.
- Perform more split tests at the lander level
- Make more creatives.
For the weekend I can only really cull ads and set up the lander tests, creatives won't get approved until Monday now.
Notes & Queries
I spent some time surfing around the target site to understand better what visitors are doing there, why they are there, who they are, and what they want. Concentrating on one site (as per bstrd's suggestion - thanks) means I can tailor my ads and lander more to those visitors.
Blending creatives to make them look like part of the site is against the TOS of SS, but that doesn't mean an ad has to look like an ad, and it doesn't mean it can't call out to the needs and desires of those specific visitors. So I'm brainstorming some ideas for how I can do that while staying within the terms. Same goes for the lander, where I do have more scope to talk directly to those visitors.
PT
Ok so...looks like from skimming most of what you have done....here is my advice.....
its the offer....
with display....it really starts and ends with the offer.....(well not always the case cheap CPM is a factor too read more below)
so....what I have done in a case like yours is go super super laser targeted....meaning don't draw a wide net and see how that offer does....
if you have done that and the offer performs awesome...then its just a matter of the finding cheap enough traffic to make it work....
I will say that on display a $3 offer can work but the CPM cost is uber critical to sucess.....
for instance....getting a $3 payout to work on a $1 cpm takes some work but can be done....but the offer needs to convert very well (preferably direct linked)....at $2 cpm its get incredibly tough as the ctr required becomes so high that it becomes hard to maintin it and banner blindness will set in very quickly and you will need creative rotations that border on excessive....
what you are trying to do with display is marry a CPM to a CVR.....so if you have a offer that converts very well 20%+ you can take risks in testing higer CPM....but if not you should be on the hunt for the cheapest traffic you can find that matches your demo..
vise-versa if the CPM is dirt cheap find offers that Can get a high enough CVR to pay for the cost of traffic.....
That is how I tend to approach display....
I know everyone says if you work hard enough you can get any offer to work anywhere....but my mind set is other wise....I look at it as a marriage...and your the match maker....
so would you pair a 20 yr hottie with some 35 yr old man who lives with his mom? God I hope not....
Think about in ways that mathematically work to your favor....hence why rebills and dating dominate display traffic....dating CPA has solid conversion rates 5-15% off banner...while rebills have such a high CPA they can monetize much more expensive CPM....due to the high EPC's
so to wrap it all up....ALWAYS start with a good offer on display....esp if its on sites you have no clue about.....ask the network...they can prob tell you if people are running it on display....if not display ask about social and see what the stats are there....if some one is running 90c epcs on fb there is a good chance you can get that offer to work on display....if they are running 20c of FB.....its prob not gonna work well...
or if you can find some dirt cheap traffic to test with throw up some very targeted banners and see what happens....if the offer converts expand from there.....if not find a new offer....
so at this stage in your camp I would be on the hunt for a good offer.....take your best performing banner/lp RPM wise and rotate offers on the backside....till you find ones that work....I would test this 2 ways
1 ) test all the lead gen penny auctions, there are a few out there some of them look really good ziinga has a 3 field $2.75 single optin that could do really well....def worth testing
2) test deal fun, ziinga and maybe 1 more with a $50-70 payout.....yes its gonna be more to test but i would def drop a few 100 or so to test it out to see if shits gonna stick....
I will end with this....if in skimming all your posts i missed that you already did all of this....I am sorry I was writing this while my 6 month old was screaming....and didn't have a chance to be a thorough as I would have liked....but I want to do what I can to help...
Originally Posted by polarbacon
Originally Posted by polarbacon
I'm up to about $0.60 epc on the offer, so yeah the math isn't in my favour. Ziinga is an offer I'm trying to get right now, and I'll check out those others as well.
My SS account just ran out of cash, so this is a good place to take the rest of the weekend off and then hit it with renewed vigour and a clearer sense of direction next week. I'm going to stop updating this thread now because it takes flipping ages and that's time better spent testing.
Thanks to everyone who gave feedback, it's all been very helpful and much appreciated. I've learned a ton from this test, and I hope the thread has been useful for others too.
PT
Sad to see you stop updating the thread but it definitely makes more sense for you to spend it on testing. It was great to have such a detailed insight into your processes, cheers! And good luck with your future SS/display ventures, let us know how it pans out!
Originally Posted by sm1810
I really just try to find cheap traffic if possible....cuz it often wont work as well on display.....not always the case but most likely
I'm going to stop updating this thread now because it takes flipping ages and that's time better spent testing.
Sorry to hear.
Maybe instead of doing huge updates just drop a couple quick sentences about how its going.
Anyways, good luck. Thanks!
Yes sorry about that, I phrased that badly. I'm not killing the thread entirely, but I won't be doing the mega daily updates.
A quick note on where i'm at then I started running this against Ziinga (at last) and had high hopes because the lander is much cleaner with the reg form at the top and only three fields as polar mentioned. So far, it's converting much worse than bid cactus. Disappointing to say the least (doubly so as the payout on Ziinga is almost dollar less).
I've made a whole bunch of new creatives which I'm working through. I also should get Deal Fun going later today. I have another offer in the mix as well, so from today I'll be rotating four offers off the back of the lander, 2 CPL and 2 CPS. That's all coming off one target url.
Frustratingly, my "charlie" url delivers conversions but very low ctr, so the clicks are expensive. The "delta" url delivers a ton of cheap clicks, and high click throughs on the lander, but absolutely sod all conversions.
That's all for now, back to polishing
PT
Interesting thread dude.
I too haven't made much money at SiteScout yet. I tried a few penny auctions but nothing too extensive. Looking forward to seeing the results.
Another quick update. Ziinga proved a disaster, conversions even lower than Bid Cactus. I also tested Arrow Outlet, but no conversions there at all. Right now I'm chucking hundreds of clicks at Deal Fun which apparently is "hot" to use AM speak. No conversions yet though (see a pattern here?)
To date, I've spent $1,154 and made back just $162, so I'm basically down a grand. For that money, I've got a handful of creatives which regularly get >1% CTR, but that's worth bog all without conversions.
I'll be running deal fun to the end of the week, then I'm going to take a break from this campaign. Not giving up (I *hate* giving up, nobody likes to fail) but there comes a point where you have to step back and rethink the thing. Can't keep throwing money down the toilet without some kind of result. The result doesn't have to be profit, but at least an improvement, and at the moment this is going backwards.
PT
Just wanted to close this off with a very short update. I stopped trying to make the penny auction thing work, and went back to first principles, c'est a dire, market first. Find a hungry crowd, feed them what they want.
I took a site with a reasonable level of traffic and a good clear demographic, spent a little time thinking about what that demo are doing on the site and what they want, then looked for some offers to match that desire. There aren't any offers that directly match it, but it was easy to use that desire as an angle, and frame some lead offers to meet it.
I tested a bunch of offers using a couple of different frames, and found an angle and offer that worked (a lead gen paying under $2). Tested the same offer on a couple of networks, and got double the conversion rate on one over the other.
By doing it this way round, market first, I've got a profitable campaign in a few days. It's returning between 90% - 110% ROI right now, and although the profit is peanuts, it can be scaled. More importantly, this is a method I'm comfortable with and I know can be replicated for pretty much any site in SS inventory.
Lessons learnt:
- KISS. Stick to first principles. Find a hungry crowd, feed them what they want.
- Frame the offer for the audience
- Test the same offer over multiple networks
- If it doesn't show immediate promise, it's not worth pursuing
- Test new banners every day
Should have done this from the outset really, but hindsight is 20/20 and all that
Thanks for all the great feedback in the thread.
PT
Glad to hear you got something working. Thanks for doing this thread.
Home > Display / Media Buying >