To: Google Founders, From: Advertisers, RE: AdWords Rotation & Other Flaws Need Fixing
Last week, my colleague and fellow columnist, Brad Geddes outlined 8 Features Advertisers Really Need from Google AdWords and I couldn’t agree more with the points he brought up.
Back in December, I ran a wish list of my own, 7 Things On My Google AdWords Wishlist For Santa and many advertisers jumped in with their own wishlists. Actually, many of the items on my list weren’t so much features as they were requests for re-engineering of incompletely or illogically-implemented features, some so old and ingrained that many advertisers had forgotten about them until Microsoft implemented them in a more sensible way – 5 years ago!
I realize now my mistake back in December was making my appeal to Santa Claus, instead of higher powers. So, this month, I am take case right to the top, Google founders, Sergey Brin and Larry Page and hope that we can get some quicker action.
Hi Larry, Hi Sergey –
It’s been a long time. I know that things have changed a lot since 2003 when these photos were taken. Times were simpler then. There was no Bing, no Facebook. No Google+, YouTube, and no Google Images. These were pre-IPO times, when all you had to worry about was Google organic search and the new baby, Google AdWords. No wonder everyone is smiling and having a good time in these pictures!
Now, I know that things have changed quite a bit since then, and your company has grown into quite the big corporation, and so, in order to go through proper channels, I’ll put my request in memo format.
To: Larry Page and Sergey Brin
From: Matt VW and all other Adwords advertisers
RE: Very Strange Changes to AdWords in Recent Days
Can you please take a quick look inside the AdWords Engineering department and in particular, the sustaining engineering group responsible for core AdWords functionality?
There have been a number of policy and interface changes and half-implemented features coming out of there in the past few years, and the pace seems to be accelerating in pace over the past 2-3 months.
Advertisers are usually cool with changes, even rapid-fire ones, as long as:
- They make sense
- They are not rushed to market and are simply bolted into the AdWords interface without regard to tracking and reporting
- They aren’t foisted on advertisers with two weeks’ notice or less especially when these AdWords policy updates require substantial re-engineering or wipe away years of best practice development. What’s the big rush?
One recent change in particular – the removal of AdWords even ad rotation has really rankled your entire advertising community, so I’ll use it as example of why I believe things may be amiss in your AdWords engineering group. This single, highly-controversial change violates three important principles of working with business partners.
The first principle is not to pull the rug out from under your business partners. And yet, this is exactly what your engineering group did a few weeks ago. Here is the announcement made by Karen Yao, Senior Product Manager on April 30th, 2012:
“…Starting next week, the “rotate” setting for ad rotation will change…”
Next week? What? This long-standing protocol has been in place for over ten years! What’s the rush?
Advertisers pay Google for each and every click that comes our way, and we have built best practices around the way AdWords operates. It is just not right for your engineering groups to destroy all that by pronouncing a change that requires re-engineering on a moment’s notice.
Paid search is a different beast than organic search where you make algorithm changes all the time to keep ahead of spammers. Companies that optimize websites to influence rankings don’t pay you, and so you don’t owe them anything really. They can kick and scream all they want, but at the end of the day, your allegiance is solely to your best efforts to providing the best user search results.
Advertisers, on the other hand, pay you lots of money to place ads on your network and we make similar investments in best practices, ad testing and revenue-maximization. Those best practices included working with even ad rotation. Removing this functionality with no notice is incredibly disruptive to the business operations of advertisers.
The second principle is ‘don’t be an arrogant smarty-pants’ with your business partners. It sure feels like that was the prevailing attitude that resulted in this recent ad rotation change, since there was nothing but a timid ‘a priori’ declaration justifying its purpose in the announcement from the same product manager, to wit:
“…Using the “rotate” setting for ad rotation is helpful for testing new creatives, but when run indefinitely can inhibit advertiser performance and deliver less relevant ads to our users. “
Two issues here. First, Sr Product Manager Yao does not come out and say this “will” inhibit advertiser performance – she says it “can” inhibit performance. This is what we call in the copywriting business, a weasel word. To remove a long-standing AdWords protocol like removing even ad rotation should require an unequivocal statement. Second, 30 days does not equal indefinitely.
There are many ad groups throughout all of our accounts where search volume takes much longer to develop performance profiles, especially when conversions are factored in.
For example, an ad group that gets 200 clicks in a month may have enough volume to start deciding a winner based on CTR, but may only get 5-10 conversions, which is not enough data on which to act. Optimizing to CTR can actually worsen advertiser performance. The most relevant ads are the ones that lead users to successful conversion activities, not just initial clicks.
For example, an ad group that gets 200 clicks in a month is a perfectly fine ad group and one in which it is possible to determine which ad is best based on CTR. However, if conversions are factored in, it could take months. Destroying the experiment before enough conversion data is available is not in anyone’s best interest.
A third principle in working with paying customers is to finish all the engineering work before introducing new features into a general release. The recent ad rotation change is a prime example of a sloppy implementation that could have been avoided had more time been given to the engineering effort. Here is what the AdWords Online Help has to say about the new ad rotation [Bold is mine for emphasis]:
“Rotate evenly: Delivers your ads more evenly into the ad auction for 30 days, then optimizes for clicks. Even though this setting is at the campaign level, the even rotation period is tracked separately for each ad group. It starts (and resets) for an ad group whenever the ads in that ad group change: specifically, when a new ad gets added, when an existing ad is changed, or when a paused/deleted ad is enabled. When the even rotation period ends and we optimize for clicks, the campaign setting will continue to say ‘Rotate evenly’.
So, basically the documentation admits that while your ads are not rotating evenly, the campaign setting will still say they are. This is inaccurate and misleading, and very lazy engineering – and not what one would deem worthy of Google’s engineering imprimatur. What was so important and pressing time-wise that you’d leave the reporting and documentation such a mess?
On top of that, the engineering team left a big loophole in the rotation scheme which enables advertisers to keep their ads in even rotation by simply pausing them and unpausing them once every 30 days.
Advertisers are starting to use campaign automation rules to automatically pause and then unpause ads a few times a month as a workaround to continue using their existing testing protocols. Why leave this workaround in place if you truly believe that the even rotation is bad for advertisers? And is this really what you want people using automation for?
Now, to be fair, we know that the AdWords engineering group are working extremely hard to impress everyone and they are introducing useful new features and targeting options at a breathtaking rate. You’ve got some of the most talented programmers on the planet working in the AdWords engineering group.
However, it appears that the balance between new feature introduction and sustaining engineering may be out of whack and may need to be addressed. We love the new features, but we’d also love to have all of the recently introduced products working the way they should work and to get some long-standing AdWords inefficiencies brought up to 2012 standards.
Many recent columns from paid search writers have talked about these things, but two features I think are seriously overdue for engineering attention are these:
Ad Site Links: These are links that are associated with the way ads are presented, and yet they have been bolted onto AdWords at the campaign level. That makes no sense. If they are ad elements, why wouldn’t they live at the ad or ad group level? Because of the implementation at the campaign level, it has been impossible to get any decent reporting on how or if they are working. This ‘new’ feature was added two years ago, and is still miserably mired in workarounds for reporting and implementation.
Ad Scheduling: Here we are in 2012, and advertisers still have to manually train AdWords ad scheduler to know what time lunch is in Mountain View versus Chicago. As it stands now, in order to deliver ads at lunch time across the continental United States requires advertisers to duplicate the same campaign 3 times. This is so 20th century! Microsoft figured this out years ago.
So, instead of bolting new features on top of the existing AdWords platform, and adding automation onto a system that is getting a little long in the tooth, perhaps it is time to consider a major overhaul of the AdWords system? Has AdWords has become an unsustainable nightmare of a system for sustaining engineers to work on?
I get it that AdWords is incredibly complex and is reliably serving up millions of ads created by hundreds of thousands of advertisers every day at every point of the globe. A major overhaul is not an easy thing to design, develop and implement when you carry the weight of the world’s commerce on your shoulders.
But let’s face it, Larry and Sergey, AdWords was originally built when none of us had gray hair, so instead of nibbling away at the edges with silly policy things like forcing advertisers to change ads every 30 days, or adding fuzzy logic to phrase and exact match, perhaps it’s time to build a new AdWords platform from scratch.
At the very least, can you please ask your engineering department not to make any more changes that require advertisers to change all their processes, campaigns, and accounts at least until Labor Day? I’d sure like to take some time this summer, and not spend it fixing campaigns that aren’t yet broken.