This blog has been quiet for a while, and that is going to both continue and change. Which is to say that there won’t be any future posts here for a while, or ever, but the discussion of paid search marketing and SEM analytics will continue on our new blog at www.clickequations.com/blog.

That blog is live now, so if you’re an actual web page visitor, please visit and reset your bookmarks. We’ll keep the same RSS feed address so RSS subscribers will be moved automatically and not have to resubscribe. The blog there is still under-development but the basics are in place and it will shape up quickly.

This move presages a number of other significant changes and announcements which we’ve got planned and coming over the next three months. On the blog, I can promise a renewed pace and intensity of discussion regarding better ways to understand and drive PPC results.

See you there.

Dear Advertiser:

We’ve noticed that you have some money that we’re not getting. This is naturally a situation which causes us grave concern, and so we’ve assigned a team of ambitious young engineers and MBA’s to the problem, and we’re excited to tell you what they’ve come up with: Automatic Budget Depletion.

Yes, it sounds too good to be true. The best part, you don’t even have to turn it on. We’ve done that for you!

From now on, you can rest assured you’re relieved of the burdonsome task of deciding where and how you want to advertise on Google. Don’t worry about what keywords are relevant to the search queries executed by the types of customers you’d like to attract. Forget about complicated matters like match type selection and bid determination. Crafting text-ads to match the desires and intents of your prospects - those days are gone!

From now on, just give us a few clues and we’ll take care of everything else. And that terrible end-of-the-month-oh-god-we-didn’t-blow-every-penny-we-had trama will be over forever. For while we still can’t tell you the true ROI of your campaigns, we can now assure you that you won’t know the return on your *entire* budget instead of having to suffer with only that kind of partial knowledge.

While we know this is exciting, we assure you that we aren’t finished improving your marketing experience yet. Right now our very own gmail engineers are busily porting over that feature which incrementally increases the size of your gmail storage so it can work on your campaign budgets too. Pretty soon you won’t even have to lift a finger to increase those. Just in time for summer vacation.

One small word of warning, however. We’ve decided that in the best interest of our users we’ll be forced to invoke quality score penalties on any landing pages which include clashing colors, dangling participles, or excessive metaphors. To ease your transition to these new rules for the next 18 months we will allow pages with these affronts to still receive clicks, but only if the CPC to Max Bid ratio was greater than .85 else these keywords will see new higher minimum bid requirements.

Thank you again for your continued blind and unquestioning support to our programs.

- The Adwords Team

Who wouldn’t want the web to be faster? Well it’s going to get faster, because soon begins a Google-Tax on load times.

As part of our continuing efforts to improve the user experience, we will soon incorporate an additional factor into Quality Score: landing page load time. Load time is the amount of time it takes for a user to see the landing page after clicking an ad.

Why are we doing this?
Two reasons: first, users have the best experience when they don’t have to wait a long time for landing pages to load. Interstitial pages, multiple redirects, excessively slow servers, and other things that can increase load times only keep users from getting what they want: information about your business. Second, users are more likely to abandon landing pages that load slowly, which can hurt your conversion rate.

My guess is that there is a third benefit - when people click ads and get quick loading pages, they can decide that they don’t like the page and then quickly get back to Google to click another ad. Too many slow landing pages and someone could tire of browsing and either buy something or just give up. Of course, the policy change isn’t being made for that reason.

Nor is it relevant that ads are what slow down tons of pages on the web. If only they could get people to stop putting ads on every single web page.

Maybe I’m a little cynical. But why don’t they just warn people that certain pages are slow loading by placing a tiny tortoise or hare icon just below each ad?

Oh ya. Unnecessary icons would slow page load times and ruin the user experience.

The beta launch of Google’s ‘Automatic Keyword Matching’ (AKA: Ultra-Broad Match) and the ensuing reaction (here, here, here, and here ) makes it a great time to talk about Broad Match and the role Google and PPC Management should play in your campaigns.

Why Match Types

Match types exist because it’s impractical to define or purchase keywords/phrases for every search query that users are going to use when looking for your goods or services. And in theory as well as reality it makes sense for the search engines to use both their deep data and algorithms to help us avoid having to specify 97 versions of ‘organic dog food’ in our keyword lists.

The problem is that the reality is not entirely as good as the theory. The Broad Match baby comes with a lot of bath water – queries that are too far from the keyword in meaning or intent and therefore are inevitably or too frequently unprofitable.

But most people don’t have the time, skill, or tools to drain the tub properly by reducing the use of Broad Match over time and expanding their use of Phrase, Exact, and Negative matches.

And that’s the problem. It’s not that Broad Match or even Ultra-Broad Match are bad per se (although Google could certainly continue to improve it in many ways) but that many advertisers rely on Broad Match too heavily for far too long.

How To Use Match Types

Broad Match (and possibly Ultra-Broad Match) should be used in many cases to test the waters. To find out which queries people are actually typing and determine how different queries convert. Armed with this information you can build a much more focused net of exact and phrase match keywords, and filter more specifically with negatives, so that the use and bids on broad match can be significantly reduced.

To be clear, there are many very valid uses of Broad Match; I’m not at all suggesting that it’s reasonable or desirable to eliminate all use of Broad Match.

I am suggesting that Exact and Phrase are vastly under-utilized, and that the bidding structure around multiple purchases of the same keywords are almost always out-of-whack. (More on the bidding implications of Match Type in a future post.)

Expanding Match Types

It’s good to see Google expanding the ways that they use their data and algorithms to drive traffic. Unlike many in the PPC community I’m not against innovations like ‘Automatic Keyword Matching’ – there are new and small advertisers for whom this type of capability is a good match.

But I do think that for larger and more advanced advertisers Google should go the other direction – exposing more options within the Match Type controls.

• It would be great to have 3-5 grades of Broad Match with some control or indication of what kinds of matches are supported.
• It would be great to have a non-linear Phrase Match (to purchase and query that includes the exact words you buy but not necessarily in that order or without extraneous words).
• It’s a crime that Match Type is not available as a variable to be handed back in the target URL string so better analysis on the impact of Match Type can be done in search analytics.

Match types are one of the most important and least effectively utilized of the controls available in paid search. Google needs to do more with them, but so do most paid search managers.

Averages are interesting. The problem is they aren’t much more than that.

It’s interesting to know that your daily average number of visitors is 49,000. Or that the average visitor sees 4.2 pages. Or that your Yahoo Search Marketing paid search click-through-rate average is 2.1%.

But before moving these numbers from interesting to important, or accurate, or actionable, you should always take the time to look inside them. Because the average may cover up some huge difference in the numbers which lie below.

That’s exactly what is apparently happening with the age-old banner ad click-through-rate numbers now that some research has been done into exactly who was clicking on those ads. The report documents:

…just 6% of the online population yet account for 50% of all display ad clicks. While many online media companies use click-through rate as an ad negotiation currency, the study shows that heavy clickers are not representative of the general public. In fact, heavy clickers skew towards Internet users between the ages of 25-44 and households with an income under $40,000. Heavy clickers behave very differently online than the typical Internet user, and while they spend four times more time online than non-clickers, their spending does not proportionately reflect this very heavy Internet usage. Heavy clickers are also relatively more likely to visit auctions, gambling, and career services sites – a markedly different surfing pattern than non-clickers.

 The problem of averages is rampant in paid search analytics, where CPCs, CTRs, and conversion rates as well as more important numbers such as ROI are often seen in ‘roll-ups’ of campaigns or ad-groups, or some type of classifications. It’s handy to see these averages, but don’t forget to dig a little deeper before taking any resulting actions.

Bryan Eisenberg fisked my conversion rate post (which itself was spawned by a column of his) and points out what I missed - and agree with - that your conversion rate is ultimately determined by how hard you work at earning it.

I still think there are huge advantages that some have and others don’t, but as Bryan clearly points out those advantages alone aren’t enough to win.

I certainly didn’t mean to be discouraging. More important than your place on the ‘best converting site on the web’ list, and directly to Bryan’s point, most sites should be able to improve their own conversion rate dramatically, possibly by hundreds of percentage points.

If you want do that, go read a lot more of what Bryan and his team say, or better yet hire FutureNow to help.

Branded keywords play an important-often dominant-role in your paid search marketing programs.

It’s therefore vital to intelligently manage and organize branded keywords. It’s also very important, and a completely separate task, to control how branded queries are matched with other keywords in your PPC programs.

Why Branded Terms Are Different

Brand term click-through-rates and conversion rates generally exceed those of non-branded terms by quite some margin. CTR’s are often 3x, 5x or higher than their non-branded counterparts. ROI for branded keywords is commonly measured in thousands-of-percentage points.

If brand terms are mixed into ad-groups across your PPC program, the results you see at either an ad-group or campaign level will show the blended averages of the branded and non-branded terms; this does not provide an accurate view of reality and therefore makes the numbers effectively worthless.

A client for whom we recently started work, for example, had an ad-group deeply mixed between branded and category terms. Looking at a campaign report you’d see that the ad-group had a 3.90% CTR a 0.86% conversion rate - about half of other comparable ad-groups within that campaign.

A closer look inside the ad-group, however, revealed a far different picture.

The branded terms had a 16.61% CTR and 1.67% conversion rate, while the non-branded terms had only a 2.48% CTR and a 0.26% conversion rate.

Clearly these numbers give three very different impressions of how the campaign is performing and of the type and urgency of changes that may need to be made.

Organizing Branded Keywords

To gain a more clear and accurate view of campaign performance, and more actionable data, pull all branded keywords out of the ad-groups they’re scattered across, and put them into branded-only ad-groups within a purely branded campaign.

These groups should produce terrific results. Take advantage of the opportunity to really work on testing text-ads for your branded terms, and since these groups/ads usually gain great quality score rankings you should be able to cost effectively attain strong top or near-top positions.

Another way to leverage this new focus is to keep a close eye on the impression share numbers in Google as you really don’t want to miss potential inventory with this campaign. It may even be necessary to raise bids not because you want to change your average position (although that may be the results) but instead because you want ensure that your close to or achieve 100% impression share.

Managing the Resulting Non-Branded Ad-Groups

If your brand terms were outperforming (which is virtually always true) the reported results for the ad-groups from which the branded terms have been removed will usually decrease. This can be quite shocking at first, but I’ve never seen anything but positive long term results from the process.

But by seeing the remaining keywords for what they are you can make much better decisions about the work these ad-groups need moving forward.

• You may realize that working a lot harder on the text-ads is necessary to see if it’s possible to drive the CTR’s up.
• You may need to revisit and either expand or contract the keyword lists, or tune the match-types.
• It’s very possible that you’ll decide certain keywords just aren’t working and never will, and pause or delete them.

In effect, now you have to do the hard work of managing a non-branded campaign. Central to your success is a willingness to test and manipulate all of the controls which are available to you.

Query Management

When your branded keywords are all segregated within selected campaigns and ad-groups, the next challenge is making sure the search engines don’t go around you with their broad/standard matching rules.

It’s not uncommon at all to see user searches for a brand name matched to category and product names which are set on broad/standard match even though that same brand name has an exact match keyword setup in a different campaign.

There are many problems with this. The first is that you’re no longer controlling your messaging but instead letting the engine ‘decide’ which of the ad-groups and keywords (and corresponding text ads, bids, and landing pages) are the best to apply to queries containing your branded terms.

In this case, you’ve lost control of the user experience both inside the engine and on your website.

In addition, since the branded queries still click-through and convert at much higher rates than non-branded searches, your search results are once again polluted (or at least unfairly shifted) by the performance of branded terms.

Another Approach

For the more sophisticated, either in terms of thinking or tools, it should be pointed out that it is possible to classify keywords using certain reporting packages, and gain the insight and clarity of brand-segregation without completing the physical separation process.

The downside of this approach is that it may limit your ability to target your creative, monitor your impression share, and limit queries to matching only the actual branded terms via the use of negatives.

Summary

In the world of search, brand terms are from venus and all other keywords are from mars. Configuring campaigns to force their separation produces better control, better reporting clarity, and better bottom line results.

The top 10 conversion rates by online retailers in January ranged from 9.6% (Amazon.com) to 14.1% (proflowers.com), according to Nielsen/NetRating (insert huge grain of salt here).

Depending on your disposition, this is either encouraging or disheartening news.

• Does it mean that 10-15% conversion rates are a goal you should work towards?
• Does it mean your 2.4% conversion rate is a terrible embarrassment?

I don’t think it means either. Let’s take a closer look at who made this list and how they did it.

• Three firms sell flowers. Who comparison shops flowers? What would you compare? If you just did something stupid, or tomorrow is ‘the day’ and you just remembered that, you buy the flowers.
• Tickets.com. Everyone must know by now that every ticket seller on the internet sells from one database (ebay, stubhub, and craigslist excepted.). There is no point in comparison shopping. You want tickets, you buy them.
• QVC. What’s their conversion rate for TV viewers? Their website is functionally a cart, so it could be argued that they’ve got 86.2% cart abandonment.
• Coldwater Creek and Lands’ End. Huge catalog mailers. Again, many many visitors coming just to place orders considered offline. If your site dropped 4M catalogs, your conversion rates would zoom too.
• OfficeDepot.com. Many no-point-in-comparing products and I assume lots of business orders from people who have accounts and replenish online frequently.
• eBay and Amazon. These are impressive - but they’re ebay and amazon. Comparing them to almost anyone isn’t fair or informative.

The message it seems is that if you need to deliver an overall conversion rate of 10% or greater, you need 30M registered users who buy from you 3-5 times per year, a 24-hour television channel, a pattern of inflicting back pain on innocent mailmen 3-4 times each year, or to sell products which are purchased as a result of some ages-old game of emotional blackmail.

Yet for many, those methods may not be practical.

That doesn’t mean you have to simply accept the shop.org and FireClick reported broad averages of 2-3% range.

There are many things most sites can do to dramatically improve conversion rates. There are also much smarter ways to measure and consider conversion rates than the overall site average. While that may be an interesting for conference-room conversation, it’s a lot more important to break down conversion rates by method-of-contact (email vs organic vs display vs PPC), based on the place in their buying cycle where visitors engage with you, or based on user intent as evidenced in their actions/expressions.

These are big topics in and of themselves, which I’ll dive into more deeply in a future post. Hat tip to Bryan Eisenberg, whose been on conversion rate watch for longer than any of us, for bringing all this up in his recent ClickZ column.

ClickFraud tracking company Click Forensics Inc. announced today that “industry average fraud rate rose to 16.6% for fourth quarter 2007, up from the 14.2% click fraud rate for the same quarter in 2006 and 16.2 percent for third quarter 2007.”

I wish could trust these numbers. But I really don’t. They could be high or low, and as usual the averages don’t matter.

There is an interesting back and forth on the ClickFraudNetwork blog between these guys and Shuman Ghosemajumder from Google.

• I personally have no doubt there is an issue.
• Or that it’s very hard to track/detect.
• Or that the search engines aren’t being as transparent and forthcoming as they should/could be.
• Or that Click Forensics is both sincere and at the same time trying to promote a business.
• Or that if the engines believed it was as small a problem as they say, they could be much more convincing.

I applaud Click Forensics and their CEO Tom Cuthbert for the initiative and effort of setting up the ClickFraud Network. But until it’s cut loose to be a truly independent entity run by a consortium of vendors and advertisers, it’s credibility is legitimately in question, even before you get to the obvious technical problems/limitations.

At some point the industry needs to face this. Thus far it’s buried in the growth and all the other changes and issues. Perhaps it will stay that way for another few years. With so money at stake that would be unfortunate.

If the day progresses as expected, Microsoft will finally buy Yahoo. This has been an inevitability for some time.

I’m all for it, despite a general distaste for Microsoft, because Google needs a serious competitor and neither entity has been thus far.

Combining these two assets doesn’t immediately create anything particularly impressive, but the Yahoo platform will gain fortitude, a heap o’cash, and the ability to do deep far-reaching bus-dev deals. Given the size of this deal and the billions MS already put out for Aquantive it’s clear they’re serious about this.

And these are guys who know how to either get it right the third or fourth time around, or otherwise wait/beat/cheat their way into a competitive position.

We all benefit. Google needs a competitor to keep it’s priorities straight in terms of advertiser features, user experience, and business terms. The rolling-at-will through the landscape they’ve been doing for the last few years isn’t in anyone’s best interest - probably even Google.

Healthy competition between two strong, well capitalized companies with relevant business and technology vision would drive the continuing transformation of advertising in a positive way for the next decade.

I never thought I’d be hoping Microsoft aggressively joined any market, but in this case I am.

Update: Killer Fake-Steve-Jobs Analysis