ClickEquations Blog

• Retweeting @imc_marketing: Can we use web analytics on whitehouse.gov? Yes we can! http://tinyurl.com/bjcne9 #
• Retweeting @Szetela: watch for PPC/AdWords contest this afternoon. $5,000-value prizes. #
• Retweeting @Szetela: RT @KISSmetrics: 20 PPC Tips For 2009 http://bit.ly/TNwp #
• Retweeting @bgtheory: It’s Official – only one Display URL per Ad Group in AdWords: http://cli.gs/2W0U4U (Versions of same domain OK) #
• Now recording a PPC Rockstars episode with @Szetela – talking about Impression Share. Airs before SMX. #
• RT @Szetela: I love it when guests divulge info to makes listeners rich. Tune in 2/ 2 for interview w/ @clickequations: http://bit.ly/1oJcLm #
• Same keyword, multiple match types, same ad group = Bad Idea. #
• Is it ad group, AdGroup, or ad-group. Can we get a definitive ruling? I write all three ways all the time. Would like to settle on one. #ppc #
• Another post on match type. This time on organizing the copies of a keyword. http://bit.ly/1OfzBf We split them out. What do you do? #
• ClickEquations site and blog down – the officials have been notified. ClickEquations PPC software platform NOT effected. #
• More ClickFraud Hype. http://bit.ly/SCpG This is taken seriously? An offiical sounding name + monthly charts & echo-chamber takes over. #ppc #
• Yahoo Explains their ‘forced optimizations’ http://bit.ly/49lcvU – Sorry Yahoo, good but not good enough. It has to be opt-in not opt-out. #
• Interesting metric idea – the number of different products purchased after clicks on a specific keyword during the selected timeframe. #

If you’re going to buy the same keyword multiple times with different match types assigned, how should you organize them?

Buying the same keyword more than once, with different match type settings, is an idea we like, as explained in our Match Type Keyword Trap series.

But this practice begs the question – should the same keyword appear more than once in the same ad group, or should you split them into different ad groups?
.

Separate But Equal
In terms of the effectiveness of the keywords at their match types it doesn’t matter. Google will match them appropriately no matter where you put them.

But I favor splitting them into separate ad groups for five reasons.

1. It’s easier to match search queries to text ads. This is the name of the game, and each keyword will attract different queries based on the different match types. So can you write better ads knowing that some of these queries will be exact, some will use the phrase, and some will be all over the broad-match-place? Probably.
   .
2. Reporting is easier to digest (pt 1). If you’re a search query freak like me, and have a great tool like ClickEquations that shows you nearly every search query, it’s easier to scan the queries in an ad group to see if they’re all appropriate and uniform in content and nearly so in performance if they’re segregated by match type.
   .
3. Reporting is easier to digest (pt 2). The roll-up data and averages of any ad group are only as worthwhile as the consistency of the performance of the keywords that make it up. Diverse keyword groups produce statisics-of-questionable-value (SOQV as it’s known in the trade). Broad match keywords perform very differently than exact match keywords and I don’t find it useful to see the average CTRs or CPCs or CPAs of them rolled-up together.
   .
4. Quality Score should be better. By the letter of the law on QS, we want high-as-possible CTRs and tight query-keyword-adgroup-landing page relevance. Both should be slightly better with segregated ad groups – although as with all quality score details, there is no way to prove this!
   .
5. Reporting is easier to produce. Google does not provide a macro to automatically tell you the match type of a keyword as part of the destination URL. This is one of the few areas where Yahoo and MSN have something Adwords does not (intentionally on the part of Google we can be sure). Therefore if you want to track, measure, report on the performance differeces of your various match types, it’s a lot easier if they’re in separate ad groups. There are other solutions, but this one is the simplest and most robust.

This is not a big deal. For many people, or even in certain situations within a campaign, repeating the keyword in a single ad group makes sense. But if and when possible, I split them out.

Note: This post was inspired by comments made on a recent PPC Rockstars with David @Szetela Podcast. These shows have become a regular part of my commute, and I recommend them highly! (Even the occasional ones when I’m an guest.)

• New video on Impression Share in ClickEquations Analyst – just posted on our YouTube channel. http://www.youtube.com/clickequations #
• Our company holiday party is tonight. MUCH easier to hire a Santa this time of year. #
• Retweeting @erez_z: Google showing ads on image searches – just the top 3. I think it won’t do very well for advertisers – time will tell #
• Marketers say people don’t need drills, they need holes. You don’t need keywords, you need search queries. Why do KW’s get all the press? #
• Another QuickEquations in 90 Seconds Video: http://bit.ly/Sf0e See campaign & keyword growth or decline reported versus the prior period. #
• Last post in our Impression Share Series is now liive. http://bit.ly/TKM2 This one focuses on how to get back lost impression share. #
• Retweeting @bgtheory: New blog post: Conventional Wisdom Failing when Search and Content are Not Treated Separately http://cli.gs/e0B8Sb #
• What better post-inauguration event than a ClickEquations Webinar? Today at 1pm EST. http://bit.ly/3vG08U Nearly 2 million joining from DC! #
• Google kills those mucho-ugly-o print ads they were offering/running. http://bit.ly/lZMo The design/execution was wrong. They’ll try again. #
• Why do YouTube Insight Analytics lag by 2-3 days? #
• RT @rimmkaufman: Google Broad Match: A Change for the Worse: Google’s Broad Match changed over the summer — unb.. http://tinyurl.com/cb7ecx #
• Surprisingly large % of PPC conversions come from landing page bookmarks – might be same person, emailed, tinyURL’s etc. Huge attribution ? #
• Anyone know how any specific PPC tool handles bookmarked PPC URLs? They probably toss them & do not accrue revenue, but I’d like to be sure. #
• Forbes article on Omniture latency woes. http://bit.ly/GqPi Over-states tech problem but not the service one. Comments reinforce this. #
• New post on Revenue Attribution with rarely seen data of actual click-purchase sequences. http://bit.ly/dCq1 How’d you spread credit? #ppc #
• Retweeting @szetela: trying to solve the problem w/negs = “bop a mole” – get hip: http://bit.ly/CsXe #
• ClickEquations Client / Trial User Training Today at 1pm EST. http://bit.ly/1TLr – Includes preview of new editing and bid automation. #

Get our tweets in real-time: follow ClickEquations on Twitter at www.twitter.com/clickequations

Revenue attribution is how you (or the software tracking your online marketing activities) decides where to place the credit for the sales that occur on your website.

If someone who has never been to your website before does a paid search, clicks an ad caused by a keyword you bought, then makes a purchase, the attribution is easy. They keyword that you paid for to get the click, should get credit for the revenue generated by the purchase.

But very often, this is not the scenario that leads to conversions which take place on your site.

• People come multiple times before purchasing.
• They often come from different sources each time the come, occasionally repeating sources along the way.
• They sometimes make a series of purchases, either after all of their visits or interlaced among their visits.

These are just a small sampling of the issues and don’t begin to define or describe the complexities.

And this is not the post where I’ll try to do either. (Those will come.)

But was we work to sort out the right way to handle revenue attribution within ClickEquations, we’re capturing some data that I thought it would be interesting to share.

The following images document real-life ‘click-chains’ – sequences of visits to a website with resulting or intersperced conversions. They are a tiny tiny fraction of the sequences found in one account in a 30 day period.

• Rows with green ‘P’ cells are visits that came from paid search keyword clicks.
• Rows with white ‘N’ cells are visits from organic search, email, affiliate links or other sources.
• Rows with yellow ‘C’ cells are conversion events.
• The number in the first column represents the visit number for that person over all time.
• Only visits within a 30 day window are included although the visit count may have begun far earlier.

And if you’re into this kind of thing, they’re very interesting.

Click-Chain Histories

#1 – Our first contestant is a frequent visitor (note we’re starting with visit 37), loves those paid search ads, but does buy at least occasionally.

Four more after the jump (as they say)

Continue Reading |Comment

Months ago we were sharing our development progress with our advisor Avinash Kaushik and asking him for input in terms of the kinds of search analytics capabilities we should include in ClickEquations.

Avinash, like another charismatic leader, shared his passion for change.

“It’s fine to see the top 50 keywords by clicks, or the top 10 ad groups by revenue” he said, but what is much more powerful is to see what’s changed -- the keywords making money today that were not making money yesterday. Or the campaigns that performed well last month, and are not performing well today.”

There is a great post on the subject of this approach to reporting on his well respected blog, written about the time this conversation took place.

In our talk he went on to explain that “the top ten of anything rarely changes. With the Delta Reports you can truly see “what’s changed” and what changes is what’s actionable. If keywords were suddenly producing clicks or conversions, a marketer can and should go figure out why.”

This leads to another word you hear often when talking to Avinash -- “insight”.

“When they go and figure out why a keyword or campaign is suddenly performing or failing to perform, they’ll likely learn something about their business or market. Something they can capitalize on.”

ClickEquations Analyst Delta Reports
The conversation inspired us to add a major new capability to our ClickEquations Analyst Excel Plug-in.

The Delta Report tables make it possible to request information about any PPC metric, and get back a sorted list based on the difference between two time periods.

Or as Avinash would say, we show you what’s changed:

• So you can request the top 100 keywords making more money this week than last.
• Or the 5 campaigns whose click-through-rates have dropped the most dramatically between January and last September.
• Or the worst 5 ad groups in terms of declining impressions.
• Or just about any period-to-period comparison pre-sorted by the amount of change.

See These Reports In Action
The Delta Report tables in ClickEquations Analyst allow you to request any data for any time-frame, and build whatever report or dashboard you need. They automatically compare the selected timeframe to the prior period (this week to last week, last month to the month before it) or you can specify any two arbitrary periods to compare.

This is very powerful.

But we include pre-built reports that are ready-to-use and take advantage of these features too.

One is our Growth/Decline Report which shows you twelve different views of your campaign and keywords based on the amount of growth and decline against a variety of metrics including clicks, revenue, and profit.

This report is the subject of the next video in our ‘ClickEquations in 90-Seconds’ series -- which is now live on YouTube and below.

Video: ClickEquations In 90 Seconds -- Growth/Decline Report

.

PS: Thanks to Avinash for the idea and inspiration.

Earlier we looked at the Google Adwords Impression Share metrics. These tell you if your ads are running when people type search queries that match the keywords you’re bidding on.

Very rarely will you find that the ads in your campaign are running anywhere near 100% of the time. Often you will find that they’re not running 25%, 50%, even 75% of the time when you probably expect that they’ll appear.

This will be shocking to some, and should be considered a huge problem.

The only reason to bid on keywords is if you want your ads to run when matching queries are typed. There is no logic to the idea that missing impression share is ‘ok’ because you don’t need the ‘extra impressions’.

• Isn’t it possible that the impressions you’re missing are the best – meaning highest converting – impressions? Or the most competitive impressions – those others are trying the hardest to take away from you? Do you really want to buy only the remnant impressions?
  .
• Or it could be that you’re getting the best ones, and missing the worst impressions – particularly if you have much lower impression share than impression share exact match (and if you’re keywords are well chosen). It could be that you’re missing lots of wierdo-broad-match Google Gumbo queries that you wouldn’t want anyway.

The point is that lost impression share is an uncontrolled mystery.

If your campaigns have high amounts (say over 30%) lost impression share you’re letting Google decide how and when to advertise your site and spend your money.

Shouldn’t you decide?

Divide And Conquer
As discussed in post II in this series, your first step is to break down your campaigns into logical units for which IS becomes meaningful. IS metrics across campaigns with dozens of dis-similar ad-groups aren’t actionable.

Of course, re-organizing campaigns is a large and difficult process. Adwords Editor makes it possible in a simpler matter than before, but it’s still a lot of work.

At a minimum your ‘must win’ ad groups should be isolated in ways that give you good visibility into their IS performance. Your core brand terms, which we’ve written about before in terms of organization, are a good place to start.

Then I’d suggest creating a slum for your losers, misfits, and keywords of questionable origin. Every campaign has them, ad groups that are a bit of stretch, a test, perform terribly but are hung onto for sentimental value, whatever.

Get anything you really don’t care about, or know deep down isn’t likely to work moved out of your bread-and-butter campaigns and onto ‘short bus’ campaigns.

You can let them run there, work on improving them, ignore them, whatever. But they will no longer be mucking up the impression share metrics in your more meaningful campaigns.

Now Do Everything Right
Once you have reasonably tight campaigns, and clear IS metrics for these cleaned-up campaigns, you can start working on a fix to the real problem(s).

Except for one tiny problem: You can’t fix what’s causing lost impression share.

Lost impressions are a symptom of a much larger disease – the overall quality of just about every aspect of your campaigns design and performance.

So if you want to eliminate lost impression share, you’re just going to have to improve nearly every aspect of your campaigns:

• Build out your match type keyword traps. Increasing coverage of exact and phrase match terms, and bidding them properly, should garner more impressions for those terms for broad-match heavy campaigns.
• Harvest search queries to increase negatives and add new phrase/exact match keywords. Every step to remove excess and intelligently expand your keywords improves the value of the IS measurement and hopefully the number as well.
• Check and address quality score across your campaign. Ad Rank = bid x QS, and often QS isn’t thought of enough.
• Write and test more text ads. This is the most overlooked effort in PPC, can drive quality score which drives ad-rank, and more importantly can multiple CTR by many times which grows everything positive.
• Bid differently. As a component of ad-rank, which plays a huge role in Impression Share, bids are a factor. Notice that bids don’t have to be your first or only lever (And watch for our upcoming blog post series on bidding.)

Impression Share is an interesting, and perhaps unexpected, broad measure of the quality of our campaigns because of how it’s influenced by the wide range of factors suggested above. Paid search is way too complex, and still to opaque (and perhaps inconsistent and imperfect) to pretend that it’s a clear measure that will track ‘campaign quality’ in any precise way – but it is an indicator and one we can use in surprisingly far-reaching way.

Impression Share Wrap Up
A lot of the paid search process happens without enough feedback or context.

Any available metrics that help us understand and measure the funnel we’re trying to push people through, therefore, is very important.

Other than the laughably inaccurate traffic/click estimates in the keyword tool, impression share is our only way to get critical visibility into the size of the audience we’re aiming at and keep a scorecard of our progress toward reaching it.

Bonus Link: Watch our ‘ClickEquations in 90 Seconds’ video on how ClickEquations Analyst enables you to track and report on Impression Share.

The post a few days ago about Impression Share included a few screen shots from one of the reports included with ClickEquations, which provides a graphic view of Impression Share.

This new video from our ‘ClickEquations in 90-Seconds’ series provides a full tour of that report and the benefits it provides.

• Retweeting @SEMOE: Using ’5s’ to manage paid search (ppc) keyword organization. http://tinyurl.com/74534p #
• Adwords ‘Suspended – low traffic keyword’. What’s that? Why can’t ads run on infrequently searched terms? How does this help them/us? #ppc #
• No Google results for the exact adword phrase ‘suspended – low traffic keyword’. What does that say about this Adwords Status Message? #ppc #
• Campaigns which organize all adgroups based on category give the appearance of organization and useful roll-up reports, but undermine both. #
• Better to group tight adgroups into campaigns based on similarities in people targeted or performance characteristiccs. Better #’s & trends. #
• Proofof PPC echo chamber? Google ‘Adwords Impression Share’. Nearly every hit is reprint of Goog post. Not good for this important metric. #
• I’ve never really understood Exact Match Impression Share. I do now. I think many are confused by it. Working on a blog post to explain.. #
• Just heard a great term that applies to #PPC and #WA – False Precision. All those numbers which seem specific but are really approximate. #
• Pls take a survey about a specific revenue attribution question – how should keywords get credit for #ppc conversions – http://bit.ly/s3Qs #
• Looks like just two majorpaid search platforms soon. http://bit.ly/HUJ1 I hope they get this over with quickly. #
• Walking downtown Palo Alto. Nice change from east coast. Finished demo with another huge advertiser who really liked clickequations. #
• 2nd post in what’s becoming a series on Google Adwords Impression Share posted. http://bit.ly/176nK #ppc #

What do you do to fix low Impression Share? That’s the question we were left with at the end of the last Impression Share post.

But before we get to that, there is something else about Impression Share that should be discussed.

Does It Matter?
Impression Share is only provided at the campaign level.

In most accounts, campaigns are roll-ups of many ad groups, and ad groups are roll ups of many keywords. Usually keywords and ad groups are not all of the same type or importance.

So before getting too flustered about missing impression share it’s worth stopping to decide if it matters, or more precisely if you can actually tell if it matters.

Suppose we have a campaign called ‘Bedroom Furnishings’ which contains 27 ad groups for everything from ‘nightstands’ to ‘sheets and pillow cases’. Within each ad group are 50 to 500 keywords, of various levels of importance and at various match types and bids.

For this business, suppose that within Bedroom Furnishings, 70% of sales are bedroom sets, 10% are headboards, 8% are lamps and the remainder are all kinds of little things. (assume all of these sales are profitable.)

In other words, only 3 of the 27 Ad Groups represent 88% of the company sales and profit.

In this case all the Impression Share metrics are useless.

The campaigns and ad groups are not organized in a way that allows us to use the IS information as it is provided.

There are too many different types of targets mixed into a single campaign. For some of the ad groups it contains we really want all the impressions we can get. For others, there are more firm ROI targets and beyond a certain point we can’t afford to bid. Still others just don’t matter much.

If we want to use and benefit from IS metrics, we need to reorganize so that one campaign holds the large volume (and profit) ad-groups, and within those ad-groups only the successful corresponding keywords.

Move the marginal keywords and ad-groups into their own campaign that can be tracked separately. And move all the other ad groups and keyword into a third campaign.

This is the minimum reorganization to make IS useful.

• At this point we can look at the IS metrics for our ‘large volume and profitable’ campaign and reasonably obsess about every % we miss.
• We can watch and work on the ‘marginal keywords and groups’ for these high profit categories, and make smart choices to improve them both in performance and IS.
• And we can watch the IS for all our other categories but probably not do too much about them.

A Bag of Rocks and Diamonds
Let me try and make the whole point another way.

Pretend you had a bag filled with 10,000 rocks and 100 diamonds.

If you knew the bag had a hole and a few dozen things had fallen out, you’d be concerned – but really not know how serious the problem was. Maybe all you lost was a few rocks.

Wouldn’t you feel better though if you could put the diamonds in their own little bag and really make sure that nothing fell out?

Keywords and ad groups are the same way. You can’t take great care of the good ones when they’re mixed in with all the junk. Separate and segregate.

A little bit of a big topic for another time, but the use of Impression Share highlights the need.

I’ve written about the problem of averages before. Impression Share is another place that getting average data for a disparate set of things can greatly diminish the value of the information. It’s up to you to organize so that the metrics provided are useful.

End of Part II
In the next post we’ll leave this issue behind and assume you’ve organized your campaigns in ways that make the IS metrics meaningful, and talk about what to do to fix what you find.