Looking for actionable ways to consistently get high quality scores from the word go?
Look no further. Grab a coffee and take 5 minuets to read this explosive article about how quality score is really calculated, and how to smash your quality score through the roof every time. No more guess work. No more scratching your head. It’s all uncovered here.
“Quality Score is an estimate of how relevant your ads, keywords and landing page are to a person seeing your ad. Having a high Quality Score means that our systems think your ad, keyword and landing page are all relevant and useful to someone looking at your ad.”
However, once you understand vector modeling (explained in just a second) that statement can be translated to:
We look at what your selling in your ad and keywords, and using a 3D vector model check how relevant the product on your page is to your ad and keywords and give you a rounded score.
There is a lot written online about quality score, but from what we have seen and tested a lot of it’s wrong. Before I knew about vector modeling I had purchased several eBooks that proclaimed to have the answer to the mysterious quality score. Whole chapters dedicated to ‘how to build links and increase quality score’! There could be nothing father from the truth (although a method does exist for doing it, by stitching relevant keywords on to the page via links, but it’s totally unnecessary for PPC). Other far-out ideas I have read include; having over 20+ pages of content, having an aged domain, having a contact phone number, getting listed with the Y! directory and ranking on page 1 for SEO.
However they all did have one thing in common, somewhere is nearly always a step like ‘then write about topic’ with out really understanding how important this part is. This is the essence of how the score is calculated, Google even said so their self in the above quote, you just need to understand how to read between the lines of what they are saying, as of course they are not revealing the whole picture. But I’m going to get you pretty close in the rest of this post, and explain what vector modeling is so you can use it in your own campaigns to increase your own quality score, every time.
There are essentially 2 phases to quality score assessment. I’ll be looking at the first, first.
Pre-1000 impression phrase.
Before 1000 impressions is reached the only information the search engine have is taken directly from the page your ads point to. They can’t use off-page or performance factors yet, as PPC landing pages are almost always brand new, and if they are not it would be unfair to new advertisers. So they look at your ad, and keywords and then run a vector modeling algorithm on your landing page to check how relevant the ads, keywords and landing page are to each other, and give you a rounded score.
It doesn’t stop there however, it also seems they run a dampening element on the keyword, based on keyword completion and broadness. So your most fundamental broad keyword (that probably also has the highest competition) will receive this dampening the most. And your the least competitive long-tail will receive a lot less or none. This is why it makes it tough to get high quality scores for competitive broad keywords without excelling with CTR optimization. However, 80% of a good account will get most of their traffic from long-tails so it still makes this process 100% worth doing and gives you an unfair advantage over any other advertiser who doesn’t understand vector modeling.
We have halved conversion rates, and almost maxed out accounts quality scores using these methods. You can see bellow that there are some pretty high competition broad keywords like ‘office phone’ that are still getting perfect 10/10’s.
Post-1000 impression phase
After the 1000 impression phase is complete, the calculation changes to a primarily CTR focused modifier. So for example if your relevance is high and your start the pre-1000 impression phase with a above average CTR, your quality score will get modified positively. So you can expect increases. However some average, and bellow average CTRs will draw quality score down. The reason it’s only some, is the metric is extremely broad, you either get shown in the Adwords interface; bellow average, average, above average or excellent. So you can only ever be sure to the closest 25%.
This makes it extremely important to get a good CTR with in the first 1000 impressions if you don’t want to lose the gains you have already made by proving relevance. However being relevant will still always benefit you. One account we optimized by copying the whole campaign, and just vector optimizing the landing page resulted in a 50% drop in conversion costs that lasted, even tho the ads were bellow average CTR and not changed.
You account history, and Adgroup history can also play a small factor, if you constantly under or over perform you will notice CPC doing either up or down over time. However relevance and CTR are the only things that you need to worry about, and it’s easy to achieve 30-90% of your account perfect 10/10 every time if done correctly.
If you look at the main factors Google lists bellow for determining quality score found on there website here, knowing what you know now, you can see this is how it is. Even geographic performance, device and site targeting and account history is CTR based. So RELEVANCE + CTR = QS.
- Your keyword’s past click-through rate (CTR): How often that keyword led to clicks on your ad
- Your Display URL’s past CTR: How often you received clicks with your Display URL
- Your account history: The overall CTR of all the ads and keywords in your account
- The quality of your landing page: How relevant, transparent and easy-to-navigate your page is
- Your keyword/ad relevance: How relevant your keyword is to your ads
- Your keyword/search relevance: How relevant your keyword is to what a customer searches for
- Geographic performance: How successful your account has been in the regions that you’re targeting
- Your ad’s performance on a site: How well your ad’s been doing on this and similar sites (if you’re targeting the Display Network)
- Your targeted devices: How well your ads have been performing on different types of devices, such as desktops/laptops, mobile devices and tablets – you get different Quality Scores for different types of devices
So how to achieve relevance & what is vector modeling?
Relevance is calculated by assigning each word on the page a probability it is relevant to the search vector (keyword) using a specialized 3D or topic vector model. So the more highly relevant terms you use on the page, and the less non relevant terms you use, the higher the relevance.
Bellow is an example of a fairly relevant landing page:
And this one is not so relevant:
It’s also important to use the top terms with the correct weight (frequency) that correlates with the overall weight the algorithm pre-assigns during its learning scans (they way they figure our what is relevant is by running ‘learning passes or scans’ of data sets). Vectorfy does a pretty good job of reversing these learning scans and giving you a coherent output you can use for your PPC or SEO.
In short vectorfy gives you a massive competitive advantage out the gate by showing you how to generate content that fits what the search engines are looking for like a glove. Providing you use vector optimized content on your page and can write good ads with good CTR your campaigns will fly!
The full strategy for setting up a campaign is out of the scope of this post but is included in the user guide of our product here. Our tool uncovers the top 800 most powerful relevance vectors for any search, and shows you how to create one perfect document that boost the relevance for the whole campaign. It also has unprecedented results for your SEO strategy as well. Sign up today!
Or see what Rand Fishkin from moz.com is saying about relevance here.