I have to admit: I’ve been dying to write an updated piece about Quality Score (QS) since 2 years ago when we put out THIS BLOG piece. The game has changed forever and I’ve spent more time gritting my teeth and cursing at my monitor (logged into Adwords) than ever before. The reason is because we were told quality score was to help ‘reward’ advertisers for constructing highly relevant campaigns and adgroups. But its all different now. Where’s the reward?

When QS was first introduced to advertisers in 2005, it was just a static score used to determine the minimum CPC based on the ad relevancy to its keywords. Over the next five years, Google would add in: CTR, landing page relevancy, account history (a combine average of all CTR’s in an account, and (the best part) “other relevant factors.” I’ve always gotten a big laugh out of “other relevant factors” because as I would dissect QS, I could see there was much more unexplained reasoning for low quality scores.

An Illustration of Traditional Quality Score (Pre-2009-2010)

In August of 2008, Google restructured QS and made it a “real-time” score that would take effect as soon as someone searched on Google. Some of the other differences Google made were: replacment of minimum CPC  to “first page minimum bid”, landing page quality, and landing page load time. In expectation of a rough change to quality scores, we were surprised that existing advertisers who had been advertising a while, didn’t really see much change…until 2010. Now we go into the accounts and look around at QS but we’re not in Kansas no mo.

I’ve always been indifferent to conspiracy theories except when it came to my own. I had to put the pieces together myself. Google’s published material were of little help and of what Googlers would say about QS, you would have to wash the automatic procedural responses from the more helpful factual things that no one could officially comment on. And when Q4 2009 hit, something was happening. QS was changing forever.

If you currently have low quality scores then first I suggest trying these steps which have been published time and time again.

1. Increase your CTR by creating highly relevant ads to the keywords in each adgroup.
2. Make sure your ads point to landing pages that contain relevant content.
3. Use  ’call to actions’ in your ad to improve CTR.
4. Ensure your landing pages load quickly. If they timeout often, compress you page file sizes or image sizes, or increase the server bandwidth.
5. See if you get a better QS from testing  new match settings.

Prior to 2009, the above techniques probably fixed everything 95% of the time. Chances are that if you’re reading this, you have already exhausted the above resources aren’t seeing improvements. In many cases, they no longer work for your quality scores. We always start with the reasons above because they are more obvious and if they do not budge, we work our way deeper into the investigation if those don’t work.

I have spent the greater part of 2009-2010 troubleshooting quality score with my team. During the coarse of investigation and screaming bloody murder, and with the help of anonymous Googlers, I have unlocked 8 new criteria which have a certain effect on QS. Much of these points have never before been published anywhere by Google. Other points that Google has published have helped to support my investigations and case studies.

8 other things you can do to fix low quality scores:

1. Use keywords that have enough volume to be deemed relevant. If you have keywords in your mix that just don’t carry the search query volume, it is quite common that you’ll find lower QS across the board. Google demonstrates that if the keyword search volume is extremely low (less 25 impressions a day), the keyword must not be very relevant and deserves a low QS. We have found the same to be true in many instances of using very specific keywords. For our local advertisers, we used to create several keywords with city targeted names. As the search volume for “service+city” decreases in less populated areas or suburbs, so does the QS. A real problem is that the accounts were flooded with these low volume (low QS) keywords and it seemed contagious to the better volume keywords in the account. These days it is best to delete the great majority of lower volume keywords in an account and lay refuge to the keywords that substantiate for most of your per-click spend.
2. Pay close attention to your industry’s average quality score for each keyword. Google now factors in the average QS of your industry or competitors when determining your quality score. There are competitive industries where it’s hard for new businesses to get involved in Adwords. Heavy ‘barriers to entry’ protect existing advertisers but make it hard for new advertisers to perform well and keep up good QS levels. An example of this is the PPC management industry. It’s extremely difficult trying to compete for “PPC Management” keywords (as you could imagine). Somethings you can’t control. Go after other keywords that are less competitive.
3. Avoid industry keywords that Google might want to put the “SMACK” down on. Google announced last year that it would not condone advertisers who were promoting products and services which they had no hand it delivering. This is because businesses nationally complained that the Sponsored Results were cluttered with affiliate marketers and get-rich-quick schemes. If you share the same keywords with affiliate marketers, get-rich-businesses, and other banned industry types; be aware that Google could mistakenly think you are one of these business types and make your advertising days hell. (To avoid other mistakes read about Quality Components below, specifically around Navigation.)
4. Use multi-word keyphrases. One-word keywords and extremely general 2-word keywords will often lead to low QS. By nature, these terms are more vague and ambiguous. If Google finds there to be no specified meaning to your keyword, it will be faulted.  This occurs more frequently in 1-word terms rather than 2-3-word terms so weight this factor more heavily with your single-word keywords. In multiple word terms, try and judge the vagueness and compare to more specific terms to see if they are getting higher QS.
5. Make sure you are not hosted on servers that are shared with malicious phishing websites, affiliate marketers, and competitiors. There is not really a sure way to know who you are sharing a server with. Recently, we created a small targeted campaign for a client who initially had some coding problems. Once the coding bugs were fixed we saw no change in QS. Google was smacking them with 1/10 across the board with in a few minutes of launching. The diagnostic tool listed the problem to be the landing page. After troubleshooting every possible reason for this atrocious scoring, our second to last resort was to change the hosting. (The last resort would have been the change the domain name.) The hosting change worked and he was 7/10-10/10 across the campaigns.
6. Look deeper into Google’s Landing Page Quality Component. Google states, “Other than complying with our Advertising Policies, we also recommend that advertisers bear in mind the three main components of a high quality website: relevant and original content, transparency, and navigability. Maintaining a positive user experience in these areas will help improve your site’s landing page quality.” Particularly in the areas that talk about Content and Navigability. Things to look for and avoid right off are: mirrored landing pages, duplicated or non-original content on site, having a website with no link structure or navigation (it looks like and affiliate offer so sales pitch page), data collection modules that offer free stuff to harvest data, linking to shopping comparison sites, direct linking to affiliate offers, linking to redirects (what’s the point?), and similar scenarios. IMPORTANT: Google will associate poor quality score with the website and not the Adwords account so if you get in trouble here, you might try a new URL if all other techniques fail. Do so once your website and Adwords account comply with guidelines mentioned on this page.
7. Don’t build too fast. No one likes a bully. Whoever coined the term, “Patience is a virtue” should have said, “If you’re not patient, you will suck mud and die.” After countless bouts of troubleshooting QS, I have found success often when I scaled back and tried to advertise 1 or 2 small adgroups to start. One of the major observations I made is that Google might try and protect existing advertisers from new advertisers bullying their way into market. This makes a lot of sense  (from the mom and pop business owner persective). As  the smaller business owner who has been advertising on Google since Nam, how sad would it be if some schmuck-o with deep pockets came in blew them outta the water by over-bidding hundreds of keywords? That kinda of activity, unchallenged, would be the death of small business.  This goes back to ‘05 and the original principles and intentions Google had for QS which was protecting advertisers from bidding wars. To set up an optimized campaign, its gonna take time anyway. I suggest you start super slow in the first week or two and then pick up the pace if you need to from there. If you launch a campaign and you’re getting hit with poor QS, try reviewing the other factors and when you start again. Build out a new but smaller campaign/adgroup and re-launch. Delete the previous failures.
8. Investigate deeper reasoning for slow landing page load time. Now and again, we will find an advertiser who is just really having a hard time getting pages to load quickly. Here, we just want to make sure there is no coding errors which would cause server timeouts. The way to notice this is if you try and visit the landing page but it never fully loads. You notice that the browser is still loading or “thinking” (if we give it human qualities as we often do with computers). If this is the case, investigate whether it is a server load issue or whether there is something on your website that won’t allow the page the load completely every time. You might seek the help of your web developer with this. Also, avoid Flash components that are not embedded into your readable code. 100% Flash intros are a “NO-NO” for Google. Google doesn’t read Flash…yet.

This blog was inspired out of frustration with Google’s ever-changing quality score algorithm. Right when you think you have it locked down, something changes and makes it more difficult. I am NOT going to be the one who says its a profitability scheme designed to make Google more profitable but I am sure someone will.

My only advice to advertisers seeking higher quality scores and better ROI is not to give up. Be patient and explore every option. If you don’t, you’ll never reach the light. Advertisers have always stood up to Google’s challenges to guarantee their own success with quality score. Just like SEO, most advertisers eventually fulfill every criteria for it until Google comes out with some new hurdle to overcome. Its harder now than ever to overcome. This is how Google sorts out competition jockeying for position and restores order to these environments. The good news is that less people will be willing to go the distance this time; leaving you with the true opportunity. This is the next hurdle. This is the new Adwords Quality Score Frontier.

Sincerely,

Peter Dulay-Quality Score Skeptic

Directing buyers is one of the easiest things to do when writing ads because you only have so many choices of where to direct them to. When you edit ads, you have your option for destination URL. This is where you will direct your visitors once they click on the ads.

Where you direct them is for the most part, a no-brainer. Most advertisers choose to point them to the landing that has the most relevant information for their buyer. However you might run into a traffic snag here or there if you don’t mind a few things. Just as you review Bounce Rates in the Analytics settings; it is good to contemplate your landing pages if you are seeing higher percentages.

Higher bounce rate percentages doesn’t necessarily mean you have unqualified buyers or have chosen an ineffective landing page. I often find high bounce rates in adgroups where the keywords are very product specific (brands and model types) usually because they point to a very product specific landing pages. This is because all the information they are looking for surrounding that keyword and inquiry is clearly on that product landing page. They get the information they want and they leave without any need to navigate the site any further.

This is inherently dangerous. If they leave your website and they haven’t purchased anything, it might be because they received their information and realized they didn’t want that product after all. So when you see high bounce rates and low sales, you might consider changing the destination URL to something with more options. This could be a product group or category page. This way they see more options and are more likely to buy because of it. You should not always assume because you have an extensive navigation feature that visitors will always use it. Believe it or not, sometimes they don’t even recognize it as an option. You should try putting options in front of them so they are forced to pick one.

This isn’t always the right answer so testing landing pages is always a good idea. In Analytics, go to Traffic Sources> Ad Versions, you will be able to look your ads bounce rates. Remember high bounce rates can be caused by many things; some of which have nothing to do with the ad itself. So look for patterns such as the one I mentioned to decide your best course of action. If it is indeed an outside issue, it will become more apparent when you test your ads’ landing pages this way.

Testing Ad Variations (Part 4/4)

Not only should we test for landing pages, we should test the ads themselves. Writing the most effective ads will always require testing. Some things that you might test for are CTR, bounce rates, conversion rates, conversion cost, and quality score. Advertisers are often very surprised at how much a difference a minor variation in the ad structure can make. When testing ads, you can make larger, more impactful changes or you can make small changes that capitalize on previous ad performance strengths. It really all depends on the feedback you get from historical ad data. If you are getting very poor performance out of an ad, you might consider changing more of the ad’s structure. If you are already getting decent to good performance, you would probably make a minor change to improve it further.

Advertisers always ask me about how to test ads. They want to know how many ads they should run at a time and how long to test before trying new ad variations. My answer will depend on a few variables including overall traffic volume and how much data I have for those ads. The more volume an ad sees, the more impactful a new variation can be on the overall business returns. The less volume an ad sees, the longer it will take to get sufficient data used for making decisions about an ads performance.

If I have a high traffic Adgroup, I will test only 2 variations at a time and continue the test for however long it takes to get reliable information back. I want to get back a large enough population of visits and impressions to justify a change. On a high traffic ad I might wait until I have at least 100 visits.

In lower traffic Adgroups, I might try 3 ad variations so we can get more data about all the ads faster. The reason for only 2 ads is because you can’t get a true assessment of ad performance when you have more than 2.

For instance, if 45% of people prefer or purchase on AD1, 35% on AD2, and 20% on AD3, AD2 could potentially be a better performer than AD1. You would not know that the people who preferred AD3, would have chosen AD2 if AD3 weren’t an option. Less is better. But when it’s going to take a long time to gather performance data, might as well test 3 to get the answers quickly. Plus, there is only a limited effect on the total business performance when the changes are made to an Adgroup that sees little volume. On a slower traffic ad, I might make a decision after 20 accrued clicks because I don’t have weeks to wait for the other 80 clicks.

For Adgroups that see much higher volume, I will test 2 variations but clone out the existing ad 3 or 4 times so to insure the competing ad doesn’t corrupt the account performance by any noticeable standard. So in the Adgroup you will see AD1, AD1, AD1, AD1, and AD2. As you accrue ad performance data, all the AD1s should be relatively even and you can compare the AD2 with all the others. It’s important that you do it this way if there is significant volume. Say you don’t clone the ads and you write an underperforming ad which could be served 50% of the time depending on how you have your serving options set. This underperformer might sabotage the Adgroup especially if it doesn’t qualify the buys well.

I have seen terrible results happen in accounts where the advertiser was doing fine but then added a poor ad in the most important Adgroup. Sales will tank. The phone will stop ringing. And most of the time, the advertiser made several campaign changes so they don’t know what is happening. Protect yourself. Take the insurance and clone existing ads.

Finally, low CTR doesn’t mean you have a bad ad. It could just mean that you have really done a good job of qualifying your visitors. If you turn away a lot of irrelevant traffic because you have a good qualifying ad, then you are saving lots of money but you are also getting lower CTRs. The best way to look at CTR is to simply look at it, be aware of it, and take in all the other performance metrics until you get the entire story. This will lead you to making better informed decisions about your ads. At no time should you make ad decisions solely based on CTR data. It doesn’t always tell the obvious truth.

Speaking about serving options: Optimized Ad Serving vs. Rotating Ad Serving? Rotating will get you a better measurement of performance quicker and Optimized will often give you the same indications by serving certain ads more. So which should we use?

How much time you spend looking at the ads should decide which you use. If you are in the account looking every day, you can rotate them without creating a significant danger to the account. Just don’t wait too long to pull out an underperforming ad. If you are likely to just let ads run without any routine management, then I advise you to put it on the Optimized setting. And if you clone ads, use the Rotating setting to get accurate feedback without creating a risk to the account. The purpose of creating clones is so you don’t have to use Optimized settings.

I hope you enjoyed the series on writing ads. Stay tuned for more valuable lessons in SEM.

The second purpose in writing effective ads is to qualify buyers. There are two reasons why you want to qualify your buyers, 1) to make sure they are actually buyers and 2) to make sure they want something you have. Some of the actions you will take in this part may already be done by virtue of taking action to attract your buyers. Nevertheless, it is important that you make a mental note that you have qualified your visitors as your buyers before letting the ads run.

Make sure they are buyers and not just information hungry visitors. (Writing catchy ads can be very similar in task.) As an advertiser, it seems common knowledge that the sponsored links area of a search results page will hold only ads whose sponsor is looking to sell something. In other words, people searching, who are not advertisers, may not know the sponsored links are there for more than information usually. How can we make sure that people clicking on our ads are really out to buy something?

We have already done this if we inserted some “call-to-action” language or description that implies a purchase. In the example ad, there were 4 things that do this; all of which were in the description.

20% Off All Whole Bean Purchases.
Free Fast Shipping, Order Online!

20% Off means off of a price which implies a purchase. The actual word ‘Purchases’ is more littoral than implied. In case you were wondering, it states, “You will be making a purchase.” Shipping information implies purchasing or buying. And the best way to imply a necessity to buy (especially when you have more product description in the ad) is to put in a call-to-action; Order Online, Order Now, Buy Now, etc. Don’t use “click here.” Google does not allow this call-to-action and it doesn’t imply the necessity to buy.

Make sure they are looking for something you can offer.

I hope by this stage in the game, you have been particular enough to only use specific keywords that describe what you do or what you sell. If not, then look up our article on Keyword Research. Assuming all your chosen keywords are relevant to your business offerings, then there are only 2 things left for you to do. Look at Bounce Rate and Search Query.

Go to Analytics Reporting features and have a look at some keyword Bounce Rates. A bounce rate is a percentage of visitors who come to the site and exit from the exact same page they landed on. Usually if we have a high bounce rate, we would assume we are not meeting the buyer’s needs for some reason or another.

I have a client who sells electric scooters for the physically disabled and elderly. Some of the scooters have 3 wheels and some have 4 wheels. One of the client’s adgroups contains keywords “3 wheel scooter” and “electric scooter.” He had a terrible bounce rate in the beginning. Why? He was attracting buyers for children’s scooters and those scooters that actually have 2 wheels, that go very fast and jump off street curbs. You wouldn’t see a disabled person performing that stunt.

His ads were titled 3 Wheel Scooters and Electric Scooters. By simply adding some qualifying language to make them show as 3 Wheel Senior Scooters and Senior Electric Scooters, we were able to lower the bounce rate and increase his conversion rate on those key terms.

Just because you have a high bounce rate doesn’t mean you have unqualified buyers. However, it is worth investigating. Always ask yourself if you are doing a good job attracting the most relevant shoppers. You might also look at your other things like landing page effectiveness, ease in navigation and checkout process, pricing, or other things that could cause increased bounces.

Search Query Reports in the Reports feature of your Adwords account will tell you a lot about what people actually type into Google to find you. If you find a lot of irrelevant variations of your keywords in there; add and test some language in your ad that would qualify irrelevant queries better. You can also add in negative keywords or test match settings on your keywords to see if that helps. We’ll cover Match Settings and Negative Keywords in a different post.

The better you qualify online visitors, the more likely you are to have a low CTR (click-thru rate). This is okay. It is always be better to have a lower CTR than to have a high CTR and pay for unwanted traffic.

Come back tomorrow to see Qualifying Buyers (Part 3/4) and Testing Ads (Part 4/4)