Negative 79% ROI (18)
Hello.
I'm running a Nutra COD product with MGID in Columbia. I got my first conversion before I spent the $10.50 conversion value, and I have been getting conversions regularly. I've been pretty optimistic about the campaign
This campaign seemed to be going fairly well for the last few days, but after $300 in spend, I realized that I'm at -79% ROI!
Should this campaign be killed? What are your benchmarks for killing campaigns that are converting, based on ROI?
Thank you for reading.
Well, is there any room to optimize from your side? Can you decrease your cpc/cpm cost on MGID or you already quite low? How about the CTR on your prelander?
If those are optimized to the max you can only test other sources or other/similar offers to gain profit.
My CPC is quite low. I'd need to raise it to do any scaling at all. Right now I'm budgeted for 2000 clicks/day, and I'm raking in about 1800, which is fine with me, in terms of my testing budget.
I have a few prelanders I'm working with, as well as a few different offer pages from the affiliate network. I've killed a couple of ads that failed the statistical significance test, but I haven't killed any landers or variations of the offer yet. I have killed a few widgets, but not too many.
There is plenty I can do to optimize. However, the low ROI has me a bit worried. If I were at -40% or even -60% I'd be a bit more optimistic. Does anyone use a rule of thumb for killing a converting campaign? I've heard a new unoptimized campaign should bring between -50% and +10% ROI initially, but I know this isn't set in stone.
It looks like one of the offers you are pushing already showed some good signs right from the start.
If you haven’t killed enough low performing or non-performing widgets by now, then maybe you should start doing so. It’s really easy to ditch a promising campaign if not properly optimized.
Before killing it I would suggest trying to further optimize this campaign and see if you can increase its performance. Here are a few rules of thumb you can start with:
Block Low Landing Page CTR Widgets
Action: Block Widget
Based on: Last 7 or Last 14 days of stats
Conditions: IF Clicks > 50 and LP CTR < 10% and Revenue = $0
Block High Landing Page CTR Widgets
Action: Block Widget
Based on: Last 7 or Last 14 days of stats
Conditions: IF Clicks > 50 and LP CTR < 75% and Revenue = $0
Block No Landing Page CTR Widgets
Action: Block Widget
Based on: Last 7 or Last 14 days of stats
Conditions: IF Clicks > 50 and LP Clicks < 1 and Revenue = $0
Besides the above suggested rules, you should pay attention on your bids as well. If bidding pretty low, you risk on not getting to reach those widgets where your other competitors are focusing on. MGID is quite flexible and granular when it comes to bids, since you can tweak widget bids by increasing/decreasing their coefficients and stay more competitive on specific placements.
The fact that you have been regularly generating conversions is a good sign.
Maybe I'm wrong, but my initial thought is that the campaign has potential but simply isn't optimized well enough.
what I specifically have in mind, is that you mentioned you haven't blocked to many widgets.
In my opinion, blacklisting widgets is probably the most important part of optimizing a campaign on native... Even if you have the best flow, it simply won't convert if you are targeting crappy widgets... however even a mediocre flow can work very well when displayed on high quality widgets.
I would give a deeper look into widget performance, before getting ready to quit the campaign
This is an advertorial lander campaign with an average LP CTR of 7.35%. Because of the low CTR, I've been giving the widgets a little bit more time before killing them. I've been killing 0% CTR after 50 clicks and I've been killing widgets after $10.50 with no sales. What CTR would you use with this LP CTR to kill widgets after 50 clicks?
How many ads have you tested? Also, have you optimized ads, stopped losers and refreshed best ones with new variations of them?
Regarding ads, I have cut some loosers, but I haven't cut many yet, as I'm relying on a probability of 10% or less of being the best according to the split test calculator that's often linked on this forum before I cut any ads.
Regarding widgets, I used Excel to calculate the LP CTR percentiles and started cutting widgets that were below the 35th percentile or with a LP CTR of 2% or less. It feels a bit like splitting hairs to be cutting widgets with 2 clicks after 100 LP views and leaving those with 3 clicks after the same, but I guess I have to draw the line somewhere and stop wasting money on bad widgets.
I've heard the argument that in cases like these, where the average LP CTR is low (typical with advertorial landers and 7.35% in this case), it's best to just let the widgets run until their cost reaches 100% of the value of a conversion before cutting them. It makes sense, but it would make testing a bit more expensive.
I'm getting the idea from reading all your replies (thank you, by the way) that even with this very low ROI, it's still worth it to optimize a bit and test some more before killing a campaign like this, given that it is converting. Does this sound reasonable?
As for ads, after a week of data you should be able to cut all ROI negative ads and focus on ROI positive only, the more you refresh ads the better on native, you get more volume, better positioning and better ROI if you focus only on the best ones.
Anyway I would go testing other offers, it never hurts.
Originally Posted by alfo1324
I had been waiting to get down to the top two or three using 90% probability of the ad's being the best, before adding new ones. Unfortunately, I don't have any ads that are showing a profit yet, but there are definitely some that are losing less than the others.
I usually start with 4/5 ads in 1 campaign, the ones I see the most while spying. Usually after 1 week of data couple of those work better either because they are profitable or breakin even. I stick to those and cut everything else, adding 3/4 variations of the winners.
I usually don't measure statical significance bc if an ad works vs another it's not something it will change over time on native, so after a week of data you can be pretty sure of the direction you need to take. Plus it's vital on native to refresh ads constantly to keep your campaign alive in terms of volume + positioning.
How many other offers have you tested?
Only one offer. I found it with spy tools. The geo isn't heavily targeted with offers, so it would be difficult to find one that could be rotated in. Also, a different offer would require different landers, as these are congruent advertorials. In spying, I haven't found other offers to rotate that meet these criteria.
However, I am rotating offer pages. The network has a few different ones
Hey!
I think that in general always start with a general budget you want to spend, usually when we invest in a campaign it's the double the amount of one conversions, that is when were talking about CPA, meaning high amount of conversions. So, Let's say your running a nutra offer with a payout of 50$, Unless you spent an initial 50$ to test the offer you haven't tested enough, because in the end your trying to find the correct traffic mix. But if you want to continue talking about this and maybe test your nutra offers on our push notification traffic, feel free to PM me, Weve been seeing great results on nutra
good luck man!
Originally Posted by p rand
Spy tools can be a bit misleading in judging whether something will work, especially on lower volume geos. Could just be some other affiliate getting started there as well.
I would definitely test a handful of offers in the beginning, rather than getting tunnel visioned on one from the start. Ask your traffic rep what geos are performing well and go from there, rather than working backwards from your affiliate manager and trying to make an offer work on the source. Not that the latter isn't also a valid approach, but it's typically more costly and the goal here is to get you a profitable campaign sooner.
A great offer makes all the difference, especially on native with all the pieces you have to optimize in the beginning. The goal in my mind at least, is to find an offer where you don't have to spend a lot of time reviewing widgets to cut because sometime works well enough broadly. Of course there will be the odd worldstar.com or other sites that have low quality traffic (but also low cpcs), but in general you should shoot for being slightly below break-even across a wide amount of widgets, giving you the opportunity to either:
1. Turn things profitable by creating better performing ads or landers
2. Switching to a whitelist and blacklist campaign strategy, where your have your
a) converting widgets in one campaign, and those same widgets blocked in the blacklist campaign
b) lower bids on the blacklist, usually so you're break-even there while discovering new ids to add to your whitelist
Originally Posted by thedudeabides
Originally Posted by p rand
This thread has fast become a really great read for people looking to get into native/trying to make it work!
Originally Posted by p rand
The other day I checked one of my ads on adplexity, and according to them X widget drove the majority of the traffic to the ad....
But according to my internal stats, X widget had 14 clicks historically
Go figure...
Home > Paid Traffic Sources > Native