5K members online now
5K members online now
Improve your AdWords performance and boost your ROI, CTR, and Quality Score
Guide Me

Target CPA is drastically ignored

Visitor ✭ ✭ ✭
# 1
Visitor ✭ ✭ ✭

Hello guys,


I wonder if anyone has a similar issue and found a proper workaround.

Adwords seems to have a HUGE problem with Target CPA - read about it in a few other threads already, mostly for remarketing. But not I see it also for new user acquisition.


I market an app to various countries and my conversion event is properly defined (and works well for old campaigns I set about a year ago). For India, for instance, I set a Target CPA of $4, yet the budget is spent for results around $50 (!!!). I've had similar results with CPAs of $9 and others. As if no matter what I put, Google Ads totally ignores that.


Has anyone faced this lately? How do you handle this? I reached out to an expert from Google who told me that's a known issue about a month ago, but perhaps someone here discovered anything.

Screenshots are attached.


Thank you so much Smiley HappyScreen Shot 2018-07-03 at 11.52.33.pngScreen Shot 2018-07-03 at 11.54.15.png


1 Expert replyverified_user

Target CPA is drastically ignored

Top Contributor
# 2
Top Contributor

I'm not sure how long you have been running this campaign, but there are a couple of factors that impact...


It used to be that you had to have a minimum number of conversions prior to being able to use CPA bidding - now not so... 


The rationale was that the system needed to have some benchmarks to be able to determine where to bid.


Now - the system simply seems to jump in and start bidding - the idea being that it's smart enough to learn quick and adjust it's bidding.


And in many cases it seems to manage to do this.


But here's the thing.... your conversions have a "real world" price. Itmay be $10 it may be $100... but it is your market that determines what that is, not you.


So you may feel comfortable paying (in your case) $4 - but the reality is that a "real world" conversion is going to cost $104. The CPA bid is not you saying Hey Google, get me a bunch of conversions at $4 - and Google will run off and do that.


So what happens?


In my experience there are a number of scenarios... one is that we have run a bunch of other campaigns previously and know, more or less, what the real world cost of a conversion is likely to be - we pitch our bid around there (generally a little below to give the system something to aim for) and away we go. In this case, the system seems to do fairly well and can often reduce the CPA a little.


The next is that the CPA is unachievable. In this case you tend to see some very erratic bidding as the system seems to clutch at straws to see if it can get a toehold and get some conversions (any conversions) at any cost... and it looks to see how far away from the target it is.


If it's not too far then it will make smaller adjustments as it works to get closer to the CPA.


If it is too far away, it simply tends to close up on itself - basically saying, "I can't get any conversions at this target price, so I'm not even going to waste money bidding" - and your traffic completely dries up.


I don't think that it is Google ignoring the bids  - I think it has much more to do with advertisers putting impossible bids into the system as targets.


I think it is often better to be very generous at the start of a campaign - make it easy for the system to achieve the goal - let's say you can get conversions at $10... put the CPA bid at $15 - Google will (should) hit that easily.... once it does, drop to $13 - once it is running nicely at that level, drop again to $11.


You will often find that you can get the numbers running at $8 or $9...


but remember, there is another side to this argument... do you want Google running super lean at $7 a conversion, managing to get 10 conversions a day, or running fat on $12 a conversion, but getting 200 conversions a day?


Don't forget that part of the equation. I often find that my own clients are trying to reduce the costs  - which is a laudable objective - but to do this they are sacrificing volume which is a bad mistake.

Re: Target CPA is drastically ignored

Visitor ✭ ✭ ✭
# 3
Visitor ✭ ✭ ✭
Hey! thanks so much for the elaborated response.

I'd share a few details to complete the picture.

1. This particular campaign, for India, runs for the first time. So it does
need some time to learn.

2. The campaign started off with more than a doubled Target CPA $9. Once
the learning phase ended (needless to say, cost per conversion went CRAZY),
it showed the campaign is limited by budget. Researching, I found out it's
in fact limited due to a counterintuitive reason - the bid is too high, and
I'm missing out on possible results. I then reduced it to $4. And still,
even if you don't agree with the action item (which was recommended by
Google), it started off with $9 and had similar, disturbing performance.

3. I'm perfectly fine with the campaign freezing its traffic due to
unachievable results. I prefer this than losing lots of money, creating ROI
negative scenario. Facebook, for instance, operates this way; it wouldn't
serve your ad in the first place. Here, Google serves it generously,
completely ignoring the boundaries I set.

4. Since the change from $9 to $4, the campaign went back to learning
status. I'm willing to leave it this way for 1-2 more days until it'll
finish learning, but the question is: will it stop serving my campaign at
these costs post-learning? Is this status, of ignoring the cap, relevant
only to the learning phase and is expected to get fixed later? This will
make it justified, even if hard.

Thanks again.

Target CPA is drastically ignored

Top Contributor
# 4
Top Contributor

Couple of follow up points...


How did you decide upon the $4v or $9 CPA - did you have data from previous campaigns to suggest this?


Also - if we try to think like the machine (!!!) the target CPA is just that - it¡s a target - it is not a limit, and AFAIK there is no multiple that acts as a ceiling either... it might make sense for Google to say, OK - you say $4 - we won't spend more than $16 because, well, because that would be silly when you set a $4 limit. But then it cannot learn as effectively as it should


Consider this... let's say your cpc was low 10cents.... but your conversion rate was also low.... 1%.... it would cost you $10 for a conversion.


now consider an alternative where your CPA is $5 but your conversion rate is 50% - CPA still $10


If you were the learning machine, which would you prefer to work with? I know which one I would choose, and why.

Target CPA is drastically ignored

[ Edited ]
Visitor ✭ ✭ ✭
# 5
Visitor ✭ ✭ ✭

Thing is that I have almost 40 campaigns on different GEOs. Most were set a year ago, and they all "respect" the limit they were configured to work with. 

The learning, after the learning period at least, needs to consider CVR when bidding - so I'd expect. So it would truly be CPA - cost per acquisition, and not per impression/click. Should the algorithm learn the conversion rate is 50%, I'd simply expect it to bid for $2.5 instead of $5, for instance. That's what I saw so far on my campaigns, and that's the same way FB Ads is working for us. :/
As for how I came up with the CPA - that's the highest we can pay as we get $9.1 revenue from such users, ±.
That's how we decide on the bids for all of the Geos: we know the very true limit, above which we lose money. We then set the CPA to never exceed the limit, and when possible lower it even more to increase the revenue.