I'm surprised this article didn't mention Google's Bid Simulator, which allows you to see a very good estimate for the number of clicks you can receive for a given keyword/creative pair (or adgroup, campaign) at bids you're currently not using:
Bid simulator is nice, but what I really wanted to highlight was the existence of profit maxima and the fact that it can be determined experimentally.
In retrospect, Bid simulator is probably the best thing to use for a starting estimate of the keyword profit optimum (if you don't have data and just use your global conversion rate across all keywords.)
Obviously google has the MOST information and is in the best position to optimize on advertisers behalf, but among SEMs I trust and respect, I've found that there tends to be a distrust for how Google applies the information and a belief that despite the inferior information, a more hands on approach tends to outperform Google's built in tools.
Bout 4 years ago, I was spending about $4000/mo on Adwords, with some nice results. About 2% CTR on the ads, most of which were banner ads on the content network, average about 20 cents PPC. About 50% of the people clicking on the ad would download our game. About 25% of the people downloading the game would run it+create a character. About 5% of those who ran it would eventually pay a monthly subscription. About a $35 customer acquisition cost, and I basically put all free cash into Adwords. Only thing that slowed it down was an average time of 3 months between joining the game and becoming a paying subscriber.
Fast forward to about a year or two ago, every metric for traffic from Adwords has plummeted. CTR's are 0.5%. Percent of people who hit the landing page and download are less than 1%. Percent of people who download and run the game to character creation is about 10%. Percent of people from Google who end up ever paying anything is near 0%. The cost of customer acquisition via Adwords has basically approached infinity, and I've basically stopped using Adwords altogether. Nothing I've done has seemed to get things right for Adword traffic, and that includes following all the advice from them. Meanwhile, traffic from Reddit converts just fine, there just isn't that much traffic from Reddit available.
Anyone else have experience while a huge degradation of the "quality" of traffic from Adwords? I'm at the point that this article mentions where no amount of bid manipulation can make PPC advertising via Google profitable.
Almost the exact same thing happened to us around May of last year. What had been a hugely profitable campaign for more than 8 years tanked in a matter of days. We still haven't figured out all the reasons or totally recovered, but there were a few big ones we noticed, at least for our market/campaigns.
- Content clicks on mobile devices became a huge portion of our clicks, and they're pretty much worthless (99%+ errant clicks that would immediately bounce, if they showed up in our analytics at all). We disable all mobile content clicks now until they add swipe to click or something to make them even a little useful.
- "Synonym" matching seemed to grow way out of control to things not remotely related. We mostly fixed with the broad match modifiers (+ before keyword)
Those are the only two we've really nailed down, and they helped make it at least profitable again, but still not anywhere near what is was before the week it tanked.
I've only run a couple of CPC campaigns but even as a reasonably clued up and very technical person I struggle to understand what's going on.
Is there software out there that effectively closes the loop between adwords and analytics and help me optimise my search through keyword and bid-space?
I would expect the system to work by letting me "seed" it with a few search terms and adverts, set in the acceptable margin per product and then let it search the space to find the optimal keyword portfolio and bid-amounts. Does this exist?
EDIT
As some commenters have pointed out, there are lots of bits of software out there that do this. However they all have either a snake-oil feel or a snake-oil feel combined with an enterprise price point. Is there an "Optimizely for adwords"? It all seems far too difficult at the moment.
there are general techniques for "seeding" search terms.
Obviously the best approach is to have someone manually do it who understands the space to initially seed it. However, its hard to get a really nice long tail of keywords that way.
The way to do something like this in an automated fashion would be to just find the competitor products and use something like SEM Rush to deduce what keywords they're hitting, then use a keyword expansion tool to expand that list.
After the initial setup, you should always be spending some of your adspend on expanding your keyword list. for example: 80% of your budget goes towards "exploit", which tries to perform as optimally as possible, and 20% of your budget goes towards "explore", which is basically letting Google and users give you hints about what words might be profitable.
To do this, create another campaign with broad match keywords (google matches a bunch of crap to these). Your script should regularly take the queries that match the broad match terms, then either:
(1) negative match it so you never see it again, for bad keywords
(2) exact or phrase match it to be managed by the CPC bid optimization method described in the post.
There are lots of automated bidding and keyword management tools out there on the market. The two biggest ones are Marin software and Kenshoo at the enterprise level. What they provide is both keyword analysis and expansion but also algorithmically driven bidding based on your defined goals. You can set it to maximize sales, revenue, or profitability. It does this by basically experimenting with different bids to achieve various positions for the ad and then models out the expected profit at different bid levels.
Another thing these tools do is incorporate your own sales data automatically via tags or offline data merges and use that for getting accurate revenue and gross profit metrics.
"Optimizely for Adwords" = Adwords. They have added very powerful tools for setting automated bid rules, little to no programming required. Developed ad markets mean the terrible keywords are cheap and the great keywords are expensive. If Google really is charging you 1 cent more than the next highest bidder you don't need to put in very much effort.
You can even just bid by CPA. Google figures out the rest (and they know how much you are making but that is another issue.)
The optimization that matters the most now is simply out earning your competitors.
Yeah, the trouble with CPA bidding is that you need to figure out how much you are willing to lose though. If you set your Google CPA at your actual CPA, then you won't make very much money. If you set it too low, you'll get crap traffic. The innards of these systems tend to be pretty complicated, so it can be hard to figure out where to bid.
Don't work for Google, but have worked with various bidding engines over the years.
This is a simplification of the problem. The issue with CPC bidding is that you're in a dynamic marketplace with competitors entering into it everyday. Maximizing profit is rarely what you want to do. Sometimes you want to bid before the maxima (to avoid encouraging competitors to bid it up), sometimes you want to bid after the maxima (to acquire customers before your competitors do).
It also assumes there is only one smooth curve that goes up to a maxima and then down again. In reality, there are local maximas that are worth considering.
Don't forget to segment your visitors by their CPC bid!
You'd be surprised how differently a visitor who clicked through to your site from the first ad spot behaves compared to visitor who comes from the 7th or 8th ad spot. For one thing, people who are clicking the lower ads tend to be more price conscious!
So if you have e.g., $20 per day to spend, it might be better to buy 5 clicks at $4 each than 20 clicks at $1. You could very well convert half of the expensive clicks and none of the cheap ones!
In over $5,000,000 of paid search spend managed, I have never seen this to be true.
The perception of this is universal to all clients, but the numbers never backed this up. Would love to see a case study that shows it's true.
I would think the opposite might be true. The person who is likely to click all the way down has already tried a bunch of links on top and their purchase intent may be much higher than the casual shopper who isn't as determined to find the best price. Also, the more price conscious shopper is not necessarily the worst shopper if you have the lowest price!
In theory you are right, but in practice that is rarely the case. See the link I posted above to Google's case study and explanation.
This is one of the biggest myths in the SEM world, most likely a result of paid search managers charging a percentage of ad spend, and therefore motivated to find justifications to bid more and spend more. (I am not saying this is a conscious effort, but you certainly can't deny the validity of the bias... and the fact that there is a better explanation to why the data might mislead you. (the better converting ads, end up in higher positions.)
I've not managed $5MM in ad spend, but its held true for many of the accounts that I have managed. At any rate, I'm only advocating for segmenting your visitors by CPC and looking at the results; I'm not saying that what's been true in my experience will be true in yours!
Regarding the article you shared: I assume Google is making that judgement based on AdWords' conversion tracking, which does not account for offline behavior, but rather tends to track actions like form submissions and calls. That's an incomplete picture for any business with a sales funnel that includes on an offline component (e.g., speaking to a salesperson on the phone, touring a college campus, or coming in for a test drive).
In my experience, customers who click on lower positioned ads might be as likely (or more likely) to fill out the contact form and ask for more info, but they are not as likely to convert into customers. YMMV.
Most of my experience is with offline conversions, using call tracking data, with a number switching app from mongoose metrics. Google's data has held up to my experience.
Did you track the position of the same ad, keyword, adgroup, and landing page across different positions and see a significant difference in conversions?
otherwise, there could be several explanations.
In my experience, while CTR are higher in higher positions, conversion rates stay steady across positions.
Yes - even with the same ads/ad group/keywords/landing pages, my experience has been that customers who click through from lower positions tend to be price shoppers, and are often less likely to become customers.
That said - my data is largely qualitative. My AdWords experience is primarily in selling local services online, and my customers tend to offer premium services (at higher prices) than their local competitors. All of this could be a factor.
"otherwise, there could be several explanations."
Absolutely, that's our own special kind of hell. :)
It could be that customers who see our ads in top spots recognize the brand from other channels, and so they aren't bothering with the other ads, whereas customers clicking on lower positions are meeting us for the first time. Or it could be that our landing pages speak to one type of customer (e.g., one who wants to hire the best, and is less concerned about price) but is leaving another type cold (e.g., those who want quality, but put a premium on value).
I'm just saying - don't forget to segment visitors based on their ad positions. :)
PS: thanks for the great discussion - its always wonderful to hear another perspective, and re-evaluate assumptions.
I have this discussion at least 5 times a year with clients. In industries with lots of price comparing, I have clients insist that lower positions produce better leads, cuz they already shopped around... The data just doesn't back this up in my experience.
I agree completely that you should measure and optimize around every metric you can, and bid management solutions can often do this for you.
I personally have yet to see data that supports the theory that searchers who click on different positions of the page convert at different rates...
"...on average, there is very little variation in conversion rates by position for the same ad. For example, for pages where 11 ads are shown the conversion rate varies by less than 5% across positions. In other words, an ad that had a 1.0% conversion rate in the best position, would have about a 0.95% conversion rate in the worst position, on average."
All of this is correct, but there's a monster hiding behind this paragraph:
> "It’s probable that unoptimized ads/landing pages will show that a lot of keywords have NO profitable bid opportunities. By improving conversion rate, quality score, ad copy, or any number of things, it’s possible to change the shape of the ad keyword profit curve, which necessitates recalibrating the keyword bid."
Your bid is just one of many factors that affect your cost-per-click and cost-per-conversion. When you're optimizing only a single variable you're only climbing the nearest hill, not the nearby peak. And climbing that little hill can cost you thousands of dollars in test campaigns.
One thing I'll tell you.. never ask for the Google support teams help: all they tell you to do is bid more. It's almost as if they're on commission. Google PPC is slowly becoming more and more expensive - and Google are slowly putting more and more adverts on the site.
My experience with the Google ad reps has generally been pretty positive. The only complaint I have is that it seems like they want to spend 30 minutes on the phone with me every other month.
Once they've consulted with you, they probably have their performance tied with your account and the accounts of anyone else that they have consulted with. Naturally, they would want to check in regularly if that is the case.
I developed a system called Ad-Curve that helps you crunch the numbers.
(I'm working on a better explanation, but in the interim check out the "Mendoza Pottery" case study which walks through the calculations.)
It's really handy at doing things like crunching whether or not you'll lose money with a given CTR, for instance, and also contains an explanation and suggestions for most of the key variables.
As someone with barely no knowledge of AdWords, I found this very interesting. Does anyone know of any software that helps automate / efficiently manage AdWords campaigns? I read a few years ago about this startup created by ex-Googlers that did just this, but apparently even before launching (or shortly after that) was bought by either Google or Twitter, not sure. Damn shame that never actually came out. Anyone remember which company I'm talking about?
This article does a good job of covering the basics of bidding, but most of these things should be obvious. In performance marketing, the goal is to know how much money you are going to make for a given visitor before that visitor enters your site. Then, you can use any number of bidding models to reach maximum profit.
What is really lacking here, is that Adwords is not just a basic auction system. If I have a Max CPC of $1.00, and you have a Max CPC of $0.85, it is not certain that I am going to be at a higher ad space than you.
There are a number of variables that Google takes in to account to determine the order of placements in an auction, and it obfuscates all these variables in to something called a 'Quality Score', which publicly operates on a integer scale of 1-10. Google uses the quality score to manipulate auctions to perform better for google, not the advertiser. Here's some of the most important factors:
CTR: If you're holding down position one, but nobody is clicking your ad, google is not making money. Just like you as a performance marketer, Google is not trying to get the maximum profit per individual click. They are trying to get the maximum volume of profitable clicks. One key here is to not just dip your toe in to a keyword. If you overbid (usually to a loss) out of the gate, you will get substantial volume faster, thus giving you a better CTR and a higher quality score. Once you have solidified your position in the auction, you can then ratchet down your bid to a profitable level.
Page Relevancy: It is important that your landing page is relevant to both the keyword you are bidding on, as well as your ad copy that you displayed. Google AdsBot will index your landing page using algorithms similar to it's organic bot. Page load time is a major factor here as well. This is google playing the long tail. They want to make sure that whoever clicked on the ad is happy with the experience they get on the landing page, so that they will in the future click on more ads. If a user keeps ending up on junk pages, they are less likely to click on ads or use google in the future.
Account History: This is in my opinion, one of the most frustrating parts of adwords, and it's impact seems to vary wildly from vertical to vertical. The age of your account and ads can play a major role in how much you are paying. How much of an impact? I've seen ads that are 3-4 years old cost $1-2/click, and an entirely new account at a new company cost upwards of $10-15/click on the same keyword. The more volatile the space, the bigger the disparity is. There is very little information out there about how/why this is, but here is my theory: The majority of people using adwords at scale are using it entirely for performance marketing. As such, if you get knocked out of a top position on a high volume keyword, it can have a significant effect to your bottom line. If a company has consistently been on a keyword spending money every day for 2-3 years, you would assume that they are going to stay there if they keep their position. If you allow a newcomer to come in and spend to a loss for a while, disrupting the positions of established players, there is not a guarantee that the established player will ever recover on that keyword. And since this new player has been spending so much, they can't afford to keep operating and have closed up shop. The net effect is less money for google. Again, this comes more in to play in volatile markets, but is straight up market manipulation.
----
As far as questions regarding technology to optimize SEM is concerned, Google has mostly prevented this from happening by having one of the worst API's I have ever worked with. They will make marginal changes and sunset the old version after only a couple months. Many of their practices seem somewhat openly hostile to developers working with their API. While being good performance marketers, data is our friend. We want as much of it as we can get. Google, however, wants us to only have just enough information to keep spending money. It is not their goal for you to have a 100% optimized campaign. They know there are junk queries that are worth less, and want you to keep blending those in with those that are performing, so that they can make more money.
Anybody that's curious about their tech should take a look in to what it takes to build something that uses the Adwords API. The fact that they can have such an awful API and tools for something that makes 70%+ of their revenue is amazing.
Reminds me of a story from Robert Kiyosaki. A best selling book and a best written book are two completely different things. If the goal is to make money, then you want the first one. It applies equally to software. As engineers I think we have a natural aversion to the dirty business of marketing, but if your goal is profitability it's even more important than having the best code. Just look at MongoDB for example, they're great at selling. But, in my opinion anyway, their product is inferior to some of the alternatives. That's not to say a great product shouldn't be your focus, but how you sell it is every bit as important (more so in most cases.)
This list made it seem like bidding higher makes all clicks more expensive. Would bidding $.375 CPC not also get you those 500 clicks @ $.25 CPC that you were getting before?
I was under the impression that your bid is the highest you are willing to pay, so you should bid up to your earnings per click.
Unless there was nobody bidding more than you, or nobody whose (bid * quality score) was better than yours, you won't get those 500 clicks at $0.25 anymore. Your ad will now appear higher than some other ad it appeared below in searches, and you'll pay what it costs to outbid that other advertiser. Increasing your bid almost always results in increasing your average CPC.
If you are bidding on searches where there's no competition, bidding more almost always means paying more for no benefit, too. I have bids on my searches related to my sites' brand names, which nobody else advertises on. When I set a CPC of $1 per click, Google would charge me something like $0.65 per click. When I dropped the bid to $0.50, they'd charge me $0.35 for the same clicks on the same placement. If I bid too little, they won't show the ad at all.
This is a basic application of the ideas in the first week of a microeconomics 101 course.
Maybe I'm being an arrogant dick, but if you don't already intuitively get this stuff, you really should not get involved in any kind of business that depends for its success on adwords, A/B testing, customer segmentation, and so on.
https://support.google.com/adwords/answer/2470105?hl=en
More speculatively, Keyword Planner allows you to add a bunch of keywords with bids to your account and see what kind of traffic you might expect:
https://support.google.com/adwords/answer/2999770?hl=en
Obviously this is less accurate since you don't have a lot of account history with these keywords, you haven't selected creatives yet, etc.
Disclaimer: I've worked a bit on all of these tools.