How To Handle AdWords With Thousands Of Keywords

[Mikkel deMib Svendsen]

For the first time I am faced with a capmpaign of this size: There is 10k keywords, one specific URL for each of them and 3-6 creatives for each (title/description combos).

Usually I set up all AdWords campaigns with one keyword (or very few close variations) in each AdGroup and multiple creatives for each, that I rotate and test to perfection. However, I just realized that the limits in Google, as far as my rep told me is:

- Max 25 campaigns in each account
- Max 100 AdGroups in each campaign
- Max 1000 keywords in each AdGroup

With these limits in mind I will not be able to set up this account the way I would usually do. As far as I see I am left with the following choces:

A) Either I limit the number of keywords used so they fit the 2500 limit (25 campaigns, with 100 adgroups that each have one keyword). This will lead to less reach in the market.

B) I add more keywords to each AdGroup. This will in my experience lead to lower CTR and most often higher CPCs and/or lower rank.


How do you handle large accounts like this (with thousands of keywords and target URLs)? If you have any experience with similar accounts to share I would appreciate it very much

[AussieWebmaster]

You can group them by category that would have much of the same creative info and then use dynamic keyword insertion to have the specific term inserted into the creative.
The separate landing page can be added to each term along with the specific tracking code for the term. Use the optimizer for the creatives and the most successful will receive the bulk of the traffic and then rotate out the low CTR ads and keep working on improving.

[Mikkel deMib Svendsen]

Yes, that is one option but still limiting to the way I really wanted to run this campaign. It's not just the leyword I change for every creative I actually do have almost 60,000 different creatives for the 10k keywords

What I do not understand is why handling 10k or 60k records is a problem for Google when the sime time they claim they can handle +6 billion undstrucured webdocuments in the web index. Maybe they could send one of the web-search engineers over to the AdWords group and get this fixed

[dhb]

I was wondering if someone might be able to lend some additional thoughts and tips as to how to start an account with thousands of keywords.

The specifics of my situation are as follows. We have about ten thousand keywords, which we've grouped into similar types. These groups of words are placed all in their own Ad Group with dynamic keyword insertion for the ad copy and individual power-posted URLs for the unique landing pages. All of these groups are in one campaign, which we have the daily budget limit set to $250,000. However, the problem we are running into is that our account is slowed before many of the words receive any impressions and corresponding CTR (based on Google's every 1000 impression check). Google then just won't show many of our words to evaluate them, even after we've deleted the "at risk" or "slowed" words and restored the campaign. We've been on the phone many times with Google's Adwords representatives, but they haven't been able to shed any light.

Would someone be willing to explain the best practice for introducing a large set of keywords using dynamic keyword insertion that allows for adequete evaluation of the words? It would be much, much appreciated.

Thank you,

David

[AussieWebmaster]

David,
You need to take some of Mike's advise.... divide the keywords into as many small groups as possible... you can have 2500 adgroups... obviously some are going to be able to hold 10-25 terms while your most successful terms and bigger traffic terms should have one word per adgroup.

Mike,
Work from a base of say 2k of individual terms and use the 12k creatives for them.... the rest you are going to have to juggle into larger groups - and when terms show solid conversion and clicks then move them into separate and clear some of the deadwood from the single groups.... it's how we do it.

[dhb]

Thanks AussieWebmaster for replying so quickly.

Can you explain further why dividing it into as many small groups as possible is beneficial? Is it to just get more targeted copy for each ad Group? Or are there other benefits to this approach? I'm just trying to understand the logic of the system.

And again, much appreciated.

~ David

[AussieWebmaster]

Originally Posted by dhb

Thanks AussieWebmaster for replying so quickly.

Can you explain further why dividing it into as many small groups as possible is beneficial? Is it to just get more targeted copy for each ad Group? Or are there other benefits to this approach? I'm just trying to understand the logic of the system.

And again, much appreciated.

~ David

Well if you place tracking to the creative you will learn which converts better and which has better CTR.... you can then use that to AB test landing pages and really tighten up your results.

[Vural Cifci]

You can divide into categories and also divide the categories in to match type. This is something that I have been trying for a while and it provides great information to analyze the broad,phrase and exact match. That helped me control the cost especially on Broad Match keywords.

[eragon]

Well this information is troubling indeed. I inherited our Adwords account, so the "fine print" never came up in my case.

Upon learning of these limitations, I immediately went to my account to see how many campaigns I have. Sure enough: I'm at 25, but 7 of them are inactive. It's strange, though, that the "Create New Campaign" link is still accessible, and even after clicking on it it lets me continue to the process. Is that because I don't have 25 ACTIVE campaigns running? Shouldn't this be deactivated if I've already reached the limit? All help is appreciated.

Maybe they could send one of the web-search engineers over to the AdWords group and get this fixed

My thoughts exactly!

[OptimizeOnline]

The number of keywords, AdGroups and Campaigns within each Google AdWords account depends, to a large extent, on how an account is structured and, as AussieWebmaster has already stated, usually you need to structure a campaign so you have as many AdGroups as possible.

But... the 'limits' that your Google rep quotes are not fixed in stone.

- 25 Campaigns in each Account
- 100 AdGroups in each Campaign
- 1,000 Keywords in each AdGroup

We have several very large Google AdWords accounts that exceed these 'limits' - although we don't usually exceed more than one of these oft-quoted 'limits' in a single account. I've just set up an account which has over 60 Campaigns, but we don't have more than 100 AdGroups in each Campaign or 1,000 Keywords in each AdGroup.

And all of these large campaigns function as well as all our other PPC campaigns.

However, you do need to be careful how much you 'push' these limits because there are implications downstream - not with respect to how the ads perform, but with reporting you may need to do. We find 'Campaign' reports are usually OK - but 'AdGroup Reports' and 'Keyword Reports' can take a bit longer to run.

[AussieWebmaster]

Originally Posted by Vural Cifci

You can divide into categories and also divide the categories in to match type. This is something that I have been trying for a while and it provides great information to analyze the broad,phrase and exact match. That helped me control the cost especially on Broad Match keywords.

If you have independent tracking you can tag them differently and find out which is actually bringing the traffic and then go through the log files and capture the actual search terms to tighten the spends.

[AussieWebmaster]

The other aspect is the disabling can be influenced by the aggregates.

[AdWordsRep]

Hi all,

I've been away from the office for the past 4 days, and am just now catching up here. I do have a few comments on the subject of account maximums. At the bottom line, most of the maximums are suggested maximums, beyond which problems may occur.

The suggested limits that I am aware of are:

* Suggested Max of 25 Campaigns.
* suggested Max of 100 Ad Groups per campaign.
* suggested Max of 750 keywords in a single Ad Group.
* suggested Max of 2000 keywords per campaign (not Ad Group).
* suggested Max of 50,000 keywords per account.

Since this differs from what others have reported hearing from AdWords support, I'll confirm my info and report back in the next few days.

Now, as noted elsewhere in this thread, many advertisers regularly exceed these recommendations by a substantial margin. Just keep in the back of your mind that it is possible to hit a 'tipping point', beyond which navigation can get difficult in your account.

I've talked to a number of advertisers who have (for example) many thousands of keywords in an Ad Group, and many hundreds of Ad Groups in a campaign - and things seem fine until, well, they simply aren't fine anymore.

And at that point, btw, it's a real pain in the neck to separate out the keywords/Ad Groups/campaigns so navigation becomes normal again.

So take the suggestions for what they're worth: suggested maximums, beyond which things may get difficult in terms of navigation within your account.

...I'm at 25, but 7 of them are inactive. It's strange, though, that the "Create New Campaign" link is still accessible, and even after clicking on it it lets me continue to the process. Is that because I don't have 25 ACTIVE campaigns running?

Exactly. The 25 campaign maximum refers to active campaigns.

BTW, on a related subject, I may have to weigh in later on a subject that is becoming a bit of a pet peeve for me: accounts loaded with thousands upon thousands of 'deadwood' keywords. More on that later.

AWR

[shorebreak]

To those of you out there who think you can manage 10,000+ keywords *and* consistently maximize volume and profits, you are seriously mistaken in my opinion. Even the agencies out there, you may have great expertise at all of the tactics necessary to do well in search, but what people don't realize is that the amount of data crunching required 24x7 to really, truly optimize a campaign to the point where no additional ounce of revenues or profit can be extracted is beyond the scope of any marketing team or agency I've ever encounter.

Large-scale paid search campaign management and optimization can only be done by algorithms. Those who say that proper tactics and campaign structure are all that is needed don't know how much they don't know.

If you have 50,000 keywords running on three search engines and there are 7 bid positions that drive meaningful volume, there are (3x7)to the 50,000th power calculations that must be done, at minimum, several times per day. Not using an algorithm to do those calculations - fully informed with actual impression, click, cost, revenue and margin data - is resulting in 40-60% of all the money spent on PPC being wasted.

I say that because I see day in and day out how much more volume and profit can be had within your typical mega-campaign when such algorithms are applied.

I won't take it the wrong way when people slam me for saying that, but it's one man's truth.

-Shorebreak

[Vural Cifci]

And you are selling......oh eFrontier.com



Anyway the only reality is whichever tool you use or so called algorithm you use, you are limited with how much the Search Engine is allowing you to do.

Anybody who doesn't agree has to expalain in details so we can pass the information to Adwords and Overture to confirm this.

Any thoughts?

[shorebreak]

I'm not selling anything, just explaining things as they are. There's a very good reason why the financial services market invests more in technology than any other sector; with billions to trillions of calculations being done by hundreds to thousands of investment firms daily in order to maximize clients' investment returns, the only way to handle that number crunching is with algorithms.

What I'm saying is that running an AdWords campaign is no different - it's a huge exercise in marketing portfolio management, and to manage such a portfolio *and* maximize returns against a defined business goal, you need to apply portfolio theory via algorithms.

True, there are constraints that Google and Overture have, in terms of their different auction mechanisms, system scalability, limited market visibility and data accuracy. We've found, though, that all of those issues can be dealt with.

[Vural Cifci]

Can you be more specific about which constraints you can deal with?
Can you bid in Overture and Adwords in every 5 mins?
Can you transfer the cost data from Overture for 1 hour before now?
Do you do anything that's more special than getting the CPC,CPA,CTR data and analyzing it?
10 keywords or 100,000 keywords. It all works the same way.
It is true that most of the investment companies do their number crunching but all can be gone with an unexpected shift in the market.(that's usually a human factor)
I believe that anybody who has an understanding of what ROI,CPC,CPA are and knows how to obtain and calculate them can manage as many listings as they want.(Guess what it is a lot easier if you know how to use excel )
It is just that I can not stand any claim that they make a BIG difference to their clients. The reason they make a difference is their clients have NO CLUE how SEM works.

[OptimizeOnline]

Originally Posted by shorebreak

To those of you out there who think you can manage 10,000+ keywords *and* consistently maximize volume and profits, you are seriously mistaken in my opinion.

Absolutely agree - running a successful PPC campaign is so much more than simply managing thousands of keywords.

Bid managment is important - although the need for which is often overtstated by companies providing bid management services, in my opinion. And, in many campaigns, a 'steady state' often occurs in which bids don't change dramatically change, or change too often - so the need for day-to-day bid management isn't so important.

Ad creative is also important and there's no substitute for testing, testing, testing. I've seen CTRs increase from 2-3% to 15-20% simply by making what seemed an small change to the ad copy.

And, to truly have an impact on ROI, you need to go well beyond Google AdWords itself. The quality of landing pages, and all stages beyond through to conversion have their part to play.

[shorebreak]

Originally Posted by Vural Cifci

Can you be more specific about which constraints you can deal with? [Budget, margin requirement, daily net profit, revenue, mutliple revenue events]
Can you bid in Overture and Adwords in every 5 mins? [On a 100,000+ keyword set, we'll change bids 1.2-1.5 times/day, on average. Some keywords require bid changes as frequently as <10 minutes, others once every week (again, on average).]
Can you transfer the cost data from Overture for 1 hour before now? [Yes]
Do you do anything that's more special than getting the CPC,CPA,CTR data and analyzing it? [Absolutely. We take all cost & revenue data, build a keyword portfolio model around it, then make bid changes where keyword-level decisions are made *in the context* of the overall portfolio of keywords, with near real-time cost & revenue data augmenting our historical model. 99% of the time an agency or internal group is just periodically loading cost & revenue data into an Excel spreadsheet that in turn kicks out suggested bid changes for each keyword. In this case the spreadsheet is making keyword-level decisions *in the context of that keyword and that keyword alone*, which the equivalent of day-trading stocks. Any savvy investment manager will tell you that making decisions on individual stocks vastly underperforms a portfolio investment strategy, yet that is what the approach agencies and internal keyword management teams boils down to. Sticky me if you want more detail, or look at the 2-keyword example on our site to see how a portfolio approach vastly outperforms the keyword-by-keyword approach the overwhelming majority of the market is taking.
10 keywords or 100,000 keywords. It all works the same way.[True, except that the computational power of 1-10 people or an agency drops off above ~1000 keywords if, in fact, the goal is to manage the keywords as a collective portfolio.]
It is true that most of the investment companies do their number crunching but all can be gone with an unexpected shift in the market.(that's usually a human factor) [Not true. Good investment firms (and keyword mgmt cos.) build historical volatility into their model (using historical keyword-level data) so that what is unexpected for most of the market is expected and planned for in the good firm's portfolio under management]
I believe that anybody who has an understanding of what ROI,CPC,CPA are and knows how to obtain and calculate them can manage as many listings as they want.(Guess what it is a lot easier if you know how to use excel )
It is just that I can not stand any claim that they make a BIG difference to their clients. The reason they make a difference is their clients have NO CLUE how SEM works. [Here again, I have to disagree with you. If all you try to achieve is efficiency within each keyword, then yes, one person can manage a million keywords; but if the goal is efficiency of the *portfolio of keywords*, then one person can only effectively manage about 500-1000 keywords. Rather than go into portfolio theory, simply check out the 2-keyword bidding example on my site (note, not trying to promote my firm, but it'll take two full pages to write the example herein). Looking at that example makes it 100% clear that no human(s) armed with Excel spreadsheets could ever manage a large keyword set *and* maximize both volume and profit for that keyword set.]

Answers to your questions:

1) Constraints: Budget, margin requirement, daily net profit, revenue, mutliple revenue events

2) Bidding frequency: On a 100,000+ keyword set, we'll change bids 1.2-1.5 times/day, on average. Some keywords require bid changes as frequently as <10 minutes, others once every week (again, on average).

3) Transferring cost data from Overture & Google - Yes

4) [Rephrased your question] Does portfolio-based keyword management involve more than just looking at CPC, CPA, CTR data & analyzing it? -
We take all cost & revenue data, build a keyword portfolio model around it, then make bid changes where keyword-level decisions are made *in the context* of the overall portfolio of keywords, with near real-time cost & revenue data augmenting our historical model. 99% of the time an agency or internal group is just periodically loading cost & revenue data into an Excel spreadsheet that in turn kicks out suggested bid changes for each keyword. In this case the spreadsheet is making keyword-level decisions *in the context of that keyword and that keyword alone*, which the equivalent of day-trading stocks. Any savvy investment manager will tell you that making decisions on individual stocks vastly underperforms a portfolio investment strategy, yet that is what the approach agencies and internal keyword management teams boils down to. Sticky me if you want more detail, or look at the 2-keyword example on our site to see how a portfolio approach vastly outperforms the keyword-by-keyword approach the overwhelming majority of the market is taking.
10 keywords or 100,000 keywords. It all works the same way.[True, except that the computational power of 1-10 people or an agency drops off above ~1000 keywords if, in fact, the goal is to manage the keywords as a collective portfolio.]

5. 10 or 10,000 keywords, it all works the same way. - True, except that the computational power of 1-10 people or an agency drops off above ~1000 keywords if, in fact, the goal is to manage the keywords as a collective portfolio and not as disconnected, day-traded stocks.

6. "It is true that most of the investment companies do their number crunching but all can be gone with an unexpected shift in the market.(that's usually a human factor)" - Not true. Good investment firms (and keyword mgmt cos.) build historical volatility into their model (using historical keyword-level data) so that what is unexpected for most of the market is expected and planned for in the good firm's portfolios under management.

7. [Rephrased question] Anyone who knows CPA, ROI and CPC can calculate and optimize to those metrics. - Here again, I have to disagree with you. If all you try to achieve is efficiency within each keyword, then yes, one person can manage a million keywords; but if the goal is efficiency of the *portfolio of keywords*, then one person can only effectively manage about 500-1000 keywords. Rather than go into portfolio theory, simply check out the 2-keyword bidding example on our site (note, not trying to promote my firm, but it'll take two full pages to write the example herein). Looking at that example makes it 100% clear that no human(s) armed with Excel spreadsheets could ever manage a large keyword set *and* maximize both volume and profit for that keyword set.

[Vural Cifci]

But how do you do this real-time again?
How do you get the data in real time such as campaign cost data and analyze it? Is your definition of real time 1 hour old (or even older)data. Because there is no way that you will be getting data from an hour ago from Overture or Google, analyze it, put it back in the process and make the necessary changes in real time. Your clients are probably not in a competitive market. Because I don't know how bid average bid change is 1.2 to 1.5 a day. I manage about 10,000 keywords and average is about 10 times a day or more depending on the day.
We can go on like this forever but again it is always easier to sell your service to the uneducated crowd. Maybe I am just used to doing this a lot and I came to the point that this is just a random function for me.