Looking for ways to beat the competition? One way is through a combined competitor and RLSA campaign!
Read more at PPCHero.com
Navneet Panda, whom the Google Panda update is named after, has co-invented a new patent that focuses on site quality scores. It’s worth studying to understand how it determines the quality of sites.
Back in 2013, I wrote the post Google Scoring Gibberish Content to Demote Pages in Rankings, about Google using ngrams from sites and building language models from them to determine if those sites were filled with gibberish, or spammy content. I was reminded of that post when I read this patent.
Rather than explaining what ngrams are in this post (which I did in the gibberish post), I’m going to point to an example of ngrams at the Google n-gram viewer, which shows Google indexing phrases in scanned books. This article published by the Wired site also focused upon ngrams: The Pitfalls of Using Google Ngram to Study Language.
An ngram phrase could be a 2-gram, a 3-gram, a 4-gram, or a 5-gram phrase; where pages are broken down into two-word phrases, three-word phrases, four-word phrases, or 5 word phrases. If a body of pages are broken down into ngrams, they could be used to create language models or phrase models to compare to other pages.
Language models, like the ones that Google used to create gibberish scores for sites could also be used to determine the quality of sites, if example sites were used to generate those language models. That seems to be the idea behind the new patent granted this week. The summary section of the patent tells us about this use of the process it describes and protects:
In general, one innovative aspect of the subject matter described in this specification can be embodied in methods that include the actions of obtaining baseline site quality scores for a plurality of previously-stored sites; generating a phrase model for a plurality of sites including the plurality of previously-scored sites, wherein the phrase model defines a mapping from phrase-specific relative frequency measures to phrase-specific baseline site quality scores; for a new site, the new site not being one of the plurality of previously-scored sites, obtaining a relative frequency measure for each of a plurality of phrases in the new site; determining an aggregate site quality score for the new site from the phrase model using the relative frequency measures of the plurality of phrases in the new site; and determining a predicted site quality score for the new site from the aggregate site quality score.
The newly granted patent from Google is:
Predicting site quality
Inventors: Navneet Panda and Yun Zhou
US Patent: 9,767,157
Granted: September 19, 2017
Filed: March 15, 2013
Methods, systems, and apparatus, including computer programs encoded on computer storage media, for predicating a measure of quality for a site, e.g., a web site. In some implementations, the methods include obtaining baseline site quality scores for multiple previously scored sites; generating a phrase model for multiple sites including the previously scored sites, wherein the phrase model defines a mapping from phrase specific relative frequency measures to phrase specific baseline site quality scores; for a new site that is not one of the previously scored sites, obtaining a relative frequency measure for each of a plurality of phrases in the new site; determining an aggregate site quality score for the new site from the phrase model using the relative frequency measures of phrases in the new site; and determining a predicted site quality score for the new site from the aggregate site quality score.
In addition to generating ngrams from text upon sites, in some versions of the implementation of this patent will include generating ngrams from anchor text of links pointing to pages of the sites. Building a phrase model involves calculating the frequency of n-grams on a site “based on the count of pages divided by the number of pages on the site.”
The patent tells us that site quality scores can impact rankings of pages from those sites, according to the patent:
Obtain baseline site quality scores for a number of previously-scored sites. The baseline site quality scores are scores used by the system, e.g., by a ranking engine of the system, as signals, among other signals, to rank search results. In some implementations, the baseline scores are determined by a backend process that may be expensive in terms of time or computing resources, or by a process that may not be applicable to all sites. For these or other reasons, baseline site quality scores are not available for all sites.
Copyright © 2017 SEO by the Sea ⚓. This Feed is for personal non-commercial use only. If you are not reading this material in your news aggregator, the site you are looking at may be guilty of copyright infringement. Please contact SEO by the Sea, so we can take appropriate action immediately.
Plugin by Taragana
The post Using Ngram Phrase Models to Generate Site Quality Scores appeared first on SEO by the Sea ⚓.
As a society, we have been conditioned with the age-old saying “Build it and they shall come”.
However, does this hold true for the digital world and your website? And more specifically, what about Google?
In most organizations, organic search optimization becomes a layer that is applied after the fact. After the brand teams, product owners and tech teams have decided what a website’s architecture should be.
However, what if I were to tell you that if search were a primary driver in your site’s architecture you could see a 200%+ performance gain out of your organic channel (and paid quality scores if you drive paid to organic pages), along with meeting brand guidelines and tech requirements?
The top 5 benefits of architecture driven by organic search
- Match Google relevancy signals with audience segmentation and user demand
- Categorization of topical & thematic content silos
- A defined taxonomy and targeted URL naming schemes
- Ability to scale content as you move up funnel
- A logical user experience that both your audience and Google can understand
When search strategy is aligned with your architecture you gain important relevancy signals that Google needs to understand your website.
You position yourself to acquire volume and market share that you would otherwise lose out on. In addition, you will be poised for organic site links within Google, answer box results and local map pack acquisition.
Imagine opening a 1,000-page hardcover book and looking for the table of contents, only to find it is either missing completely or reads with zero logic. As a user, how would you feel? Would you know what the chapters are about? Get a sense of what the book is about?
If you want Google to understand what your website is about and how it is put together, then make sure and communicate it properly – which is the first step for proper site architecture.
Let us pick on a few common, simplistic examples:
/about-us (About who?)
/contact-us (Contact who?)
/products/ (What kind of products?)
/articles (Articles about what?)
/categories (Category about what?)
And my very favorite…
/blog (Blog? What is that about? Could be anything in the world)
These sub-directories within the infrastructure of your website are key components – they are the “chapter names” in your book. Naming something “articles” lacks the relevancy and key signals to describe what your chapter is about.
The upper level sub-directories are known as parent level pages, which means any pages underneath them are child level pages. As you build and scale child level pages, it should be categorized under the proper parent level page. This allows all of the related content of the children pages to “roll up” and become relevant for the parent level page.
Google thrives on this sort of organization, as it provides a good user experience for their users, as well as communicating systematically what the pages are supposed to be about and how they are related to each other.
Example of a proper architecture
As you can see from this example, the relevancy of the two category levels (business plan template & how to write a business plan) all have relevancy that rolls up to the term business plans.
Then as you drill down one level deeper, you can see that you would isolate and build pages that are for business plan outline and business plan samples. These both roll up to the business plan template category.
Through proper keyword targeting and research you would locate the primary keyword driver that matches the page intent and high volume for the URL naming conventions. This communicates to Google what the page will be about as well as matching high customer demand from a search perspective.
Most brand or product teams create and name a structure based on internal reasons, or no particular reason at all. So rather than applying search filters after the fact and trying to retrofit, do the research and understand the volume drivers – then apply them to the architectural plan. You will have significant gains in your rankings and share of voice.
With a structure like this, every page has a home and a purpose. This architecture not only is designed for “current state” but also will scale easily for “future state”. It becomes very easy to add child categories under the primary silo category thus allowing you to scale easily and move up funnel to capture new market share and volume.
How does user experience (UX) play a role in architecture?
A common crossroads we encounter is the UX as it relates to search, content marketing and architecture. UX typically wants minimal content, limited navigational options and a controlled user journey.
However, keep in mind that a UX journey is considered from one point of entry (typically the home page), while search if done properly – every page becomes a point of entry. So we need to solve for both.
The good news is that pure architecture structure and URL naming schemes is and can be completely different than the UX. Build the architecture the proper way and you can still apply any UX as an overlay.
Where the primary differences come in is between UX and navigation. Here again, UX typically wants to limit the choices and control the journey, which means that the navigation is reduced and not all architectural levels are available and visible.
The challenge here is that you want Google to rank you number one in the world for all of these pages; however, you are also telling Google they are not important enough to you to even be in your navigation.
A rule of thumb I learned almost 20 years ago is to make sure every page can stand on its own. A user should never have to go “back” in order to go forward. So make sure your navigation and categorical pages are available from every page, especially knowing for organic search, a user will enter your site and the journey at every level.
Now does this mean abandoning UX? No. You can still control the journey through your primary CTAs and imagery, without sacrificing navigation or architecture.
Images are the main culprit for causing oversized web pages (average size 2.2MB) that can perform slowly on mobile devices.
Last week, we looked at how to optimize your mobile site speed and test for issues that might be slowing your site down. With Google placing more and more emphasis on mobile site speed and user experience in order to achieve a good ranking, discovering issues with your mobile site speed is critical. But how can you fix those issues once you’re there?
This column, the second of three, will discuss the different ways to reduce the impact of images on the performance of your mobile site.
You need to fix your mobile image problem. You are not alone. As detailed in my previous column, the average mobile webpage is now a ludicrous 2.2MB and this can severely impede how quickly the page loads on a mobile device.
At 68% of total page weight, images are the main culprit. This column will look at the different ways to reduce the impact of image on the performance of your website.
For those you like a step-by-step approach, here are ten steps to speedy mobile web pages.
- Speed test: how fast site loads on mobile device; causes of delay.
- Test image impact: do images improve or kill user experience.
- Image policy: review company policy on images; educate everyone.
- Image audit: evaluate number, format, size, situation, impact.
- Cut the fat: remove images that do not add value.
- Image impact: make images work harder.
- Balancing images with accessibility.
- Optimize images: right format, right size.
- Alternative techniques: for icons and buttons.
- Even more speed: web design techniques.
Steps 1-4 were mostly covered in Part 1. This column will concentrate on steps 5-7. Part 3 will concentrate on steps 8-10.
Cut the fat
The first and easiest step to improving page speed is to remove unnecessary images. If images do not add value and are taking up valuable real estate get rid of them. Crappy stock image just there to fill space? Be gone.
When you are auditing your existing pages or creating a new one, ask: does this image justify:
- The space it takes up on the mobile screen?
- Adding 30KB, 50KB, 100KB etc. to the page size?
Robert Gaines, a Kansas, US-based, web and app developer, advises:
“Before you add an image to a webpage, decide if you really need it; if it really adds value. Every image that you add slows the webpage down, impacting both user experience and search rankings. If you don’t need an image, cut the fat!
Slider images are a great example of images that can be cut from a webpage. Studies have shown that sliders do not have a significantly positive impact on either conversions or user experience – they are unnecessary bloat.”
Use heatmaps, user testing and A/B testing (covered in Part 1 of this series) to assess the impact your images are having on the users. Is it positive, negligible or negative?
Images that count
Once you have illuminated the images that don’t add value, concentrate on making the remaining images work harder for their pixels/KBs.
Raluca Budiu, Director of Research at Nielsen Norman Group, says:
“As web images can take longer to load on mobile devices, it is usually a good idea if they contain information, and are not purely decorative. Users rarely appreciate a pretty picture that is unrelated to an article. Ecommerce sites, in particular need good images — people can rarely make a buying decision if they cannot see the product well.”
Case study: Unilever mobile-ready hero images
One company that has really flung its weight behind making images more useful is Unilever.
Working with Cambridge University Inclusive Design team, the company has redesigned all product shots for all its brands to make them easier for web customers to instantly recognize and pick the right product, just using visual recognition, as they speed through any retailer’s catalogue of products “Vegas style” (i.e. spinning through results like the reels on a slot machine).
The idea is that by highlighting the essential details – brand, product type, size and quantity – on the image, shoppers don’t have to read the copy for every product.
In case you were wondering, a hero image is the term given to the fat clickable/tappable banner image, often on a carousel, at the top of homepages (particularly), featuring a new product, promotion etc. and a call to action. Mobile-Ready Hero Images are the smaller thumbnail images found on a page of product listings either in a category or following a search for e.g. men’s deodorant.
Oliver Bradley, Global eCommerce Experience Design Director at Unilever, explains:
“Mobile-Ready Hero Images were created specifically for online retail to act as the primary image in search / favorites thumbnail results. They are designed to work well on all screen sizes that are typical for online retail (16mm on mobile, 23mm on tablet and 48mm on laptop / desktop).
They allow the online shopper to better recognize brand, format, variant and size as they are perform that fast vertical scroll which is typical behavior when scanning through long lists of products on a mobile or tablet.
The images can be zoomed and/or cropped to deliver better legibility & recognition of brand and variant esp. on portrait packs. If required brands can add the size of the product or number of contents in the bottom right-hand corner. If necessary, the type of product may be added as a stripe on the right-hand side (portrait packs), or as a stripe along the bottom (landscape packs). Makes better use of space than with a conventional packet shot.”
First introduced in 2014, Unilever has moved rapidly to convince retailers in 20 countries – including many of the retail giants e.g. Walmart, CVS, Walgreens, Tesco and Carrefour – to introduce the enhanced imagery to their e/m-commerce sites.
With some retailers, such as Asda.com (Walmart’s UK subsidiary), every Unilever product is now marketed with a Mobile-Ready Hero Image.
What’s more, Unilever, with the Cambridge team, have turned Mobile-Ready Hero Images into a standard: available open source and free for any competitor to pick up. And many of Unilever’s biggest competitors have done just that, says Bradley:
“We realized that we needed to create a versatile category solution with the same visual architecture and consistency across the board. Since doing this we’ve been pleased to see L’Oréal, GSK, P&G, Kellogg’s, Kimberly Clark and J&J follow our approach.”
The image below displays two screenshots. The first captures a deodorant search from Walmart.com, showing two Unilever brands with hero images with the product type and size alongside the product shot. With the two rival products, it is more difficult to establish the type of product or the size, without zooming in on the product.
As Walmart does not include these details in the blurb, this makes the hero images doubly useful.
The second image captures a shampoo search from Asda.com, showing a Unilever product (TreSemme), a P&G product (Herbal Essences), both using the hero image. The third product, also a P&G brand (Pantene) does not use a hero image. While the hero images are considerably more high-impact, Asda does include the product type and size in the blurb… assuming that the consumer reads it.
So what’s the impact on UX?
At the start of the project, Unilever and the Cambridge team conducted quantitative research (i.e. interviews and surveys) with 3000 shoppers in and then verified their findings/work by conducted eye tracking tests with more than 100 shoppers, to watch how they interacted with the site and images using mobile devices on mobile to verify.
This user testing put Unilever in a good position to rebut any doubters from the retail business.
“Some retailers argue pack shots are necessary for e-commerce, because the text description next to the pack shot provides all the information that cannot be obtained from the photograph. But we know from eye tracking that 90% of the e-commerce page is ignored.
We know that the majority of shoppers don’t bother to check the product name, size or format on retailers’ websites or apps. That’s why nearly every shopper we interviewed had a story of accidentally buying a product online that turned out to be not as expected (especially on size). If a pack looks similar to the one their used to, they don’t bother to check the text to verify.
The point of online grocery is convenience and saving time. It’s about selecting a large number of products as quickly as they can. Because you don’t have to check the text with hero images, it makes shopping even faster.
We know this is true from the conversions the hero image have delivered. The A/B test results have been amazing: a 24% lift on Magnum, a 19% lift on Simple, and a 4% lift on Ben & Jerry’s.”
The biggest issue for any sort of image, but particularly an image that contains important textual data, is that that they are invisible to people with visual impairments. People who don’t see well or at all, use screen-readers to read the text on the site aloud.
There’s nothing wrong with using images, or hero images for that point, but it is essential that the content in the image should be either written in the web copy and/or in alt tags (these tags describe the image to a screen reader, but are invisible to the able-bodied reader).
Generally, accessibility experts frown on the use of text being placed within an images, because screen readers are blind to the data it contains. This is why, for example, it is preferable from an accessibility point of view to prepare tables of data as an html table, rather than simply uploading an image of the table – but this will always take additional time and usually won’t look as good.
Marco Zehe is an accessibility evangelist and QA engineer at Mozilla:
“Mobile images need alt text, just like for the desktop, of course. But it helps if the images are responsive, because people with low vision or elderly people will benefit from them zooming better.
“However screen readers do not do image recognition on the web, yet, so if there is text on the image, this is will be inaccessible. In principle, text on images should be avoided. But if the text helps clarify a photo, especially for people with low vision, and as long as it is only duplicating what is said in the copy and in the alt text, this is acceptable.”
So, if text genuinely adds value to an image, as is the case with Mobile-Ready Hero Images, then it is safe to assume this is acceptable, as long as it is duplicated in the copy and the alt tag.
So looking back to our Walmart and Asda examples pictured above:
- The source code for the Dove deodorant on Walmart contains the alt text: “Dove Men+Care Extra Fresh Deodorant, 3 oz, Twin Pack”, which describes the product, size and quantity. The page copy describes the product, but does not include size or quantity.
- The source code for the TRESemme Shampoo on Asda contains alt text: “TRESemme Moisture Rich Shampoo Rollback”, which describes the product, and offer, but not the size. The shopper needs to click through to discover this information.
This article is Part 2 of a three-part series on how to optimize your mobile site speed. Check back next week for Part 3, which looks at how to refine and reformat your mobile site images.
Or read the previous instalment: How to optimize your mobile site speed: Testing for issues.
This column was originally published on our sister site, ClickZ, and has been reproduced here for the enjoyment of our audience on SEW.
The right picture is very useful on mobile and responsive websites. But images that are too large, too numerous and unnecessary simply slow down page load times and get in the way of the users reading and doing what they need to do.
The problem: the size of webpages sent to mobile phones has quadrupled in just five years. The main cause: images, which account for 68% of total page weight.
With mobile page speed a confirmed ranking factor in Google’s last mobile-friendly update, and Google’s mobile-first index looming large on the horizon, it’s in the interests of developers and SEOs to optimize their mobile site speed as much as possible. That means figuring out how to trim the fat from all those huge, cumbersome images.
This column will explore the issue and causes of delays in mobile page speed (i.e. how quickly pages load on a mobile device) and how to test webpages for problems with speed.
The next column will look at methods to reduce the impact of images on the performance of your mobile pages. This includes only using images that add value and making the images you do use work harder, with an excellent case study from Unilever.
This data was sourced from the incredibly useful httpArchive, which tests the top 1 million sites several times every month:
- The average transfer size (i.e. bites sent from server to device) of a webpage is 4.2 times larger than it was five years ago, rising from 521 Kilobyte (KB) in December 2011 to 2197KB or 2.197 megabyte (MB) in December 2016. N.B.: this measures compressed rather than original file sizes.
- Images are a massive amount of that bloat. Total size of images sent to mobile devices has increased 4.2 times from 352KB in 2011 to 1490KB in 2016.
- Image requests have grown from 38 to 50. JPEGs are most common, accounting for 46% of image requests.
- The one to watch is video, which was nonexistent in 2011 websites, but now averages 110KB or 5% of total page size and takes up a gigantic 542KB per request vs 43KB for a JPEG.
The most shocking discovery here is that the average size of a mobile webpage, 2197KB (2.2MB) is almost as large as the average desktop webpage at 2469 KB (2.5MB) in November 2016. We can only surmise why this might be:
- Are responsive design websites… or to be more precise inefficiently built/implemented responsive design sites to blame (because the true responsive design website is one site reformatted for different devices)?
- Has the adoption of lazy and deferred loading techniques encouraged companies to be less efficient with total page size?
Putting things right
A note before we begin:
Web/mobile images are an imprecise science. There are no hard and fast rules – different practitioners and scenarios dictate differing course of action.
There is no best format, size, content type, design, shape, placement or number of images, but there are best practices to help you make those decisions. The rule of thumb is as small, as few, as big an impact as necessary.
“Images” are not just illustrative pictures or graphs. They also include logos and icons – but these do not necessarily need to be traditional images, such as JPEGs.
- Review your policy on images, or create one, if you don’t have one. Issue guidelines for all web content creators as well as developers.
- Audit the images you are using on the site. Are they adding or taking away from user experience? Can they be improved, optimized, reduced in size (on page), pushed below the fold or removed?
- Test how effective your images are with the users. Research/test before you make any changes, test as you pilot changes and monitor results after changing.
- Work out how you will balance page speed with attractiveness, quality, impact, page speed, efficiency and accessibility.
The need for speed
Robert Gaines, a Kansas, US-based, web and app developer:
Graphics are attractive and allow users to quickly grasp concepts without reading large amounts of text, however they also slow load times. Excessive use of images or the use of especially large images will slow down webpages. Slow load times annoy both readers and search engines.
The need for graphics has to be balanced against page speed. When images are used, they should be compressed and scaled so that they load more quickly. In cases where compression and scaling aren’t enough, other advanced techniques may be needed.
There is no rule for perfect speeds that a page should download to a mobile device, how could there be – mobile connections vary massively. The rule of thumb is as fast as possible. Benchmark your performance against competitors and sector leaders.
Various studies and reports, see WPO Stats for examples, have shown that improving page speed improves conversions. For example, an FT.com study found that reducing page load by 3 second on mobile led to a 9% reduction in articles read over the month.
Google warns on its TestMySite tool that “Nearly half of all visitors will leave a mobile site if the pages don’t load within 3 seconds.” But the source of this stats is unclear.
Test, test, test
Testing is critical improving to website performance and usability.
1. Test how quickly pages download
Regularly test your mobile webpages (all new ones and all the main ones). Use different services and at different times, because tests results will differ… a lot.
For something less techie, use Google’s PageSpeed Insights (also try the simplified version: TestMySite – it sometimes surprises by offering a different results to its brother). N.B. Google doesn’t actually test page speed, it estimates page speed based on key criteria, but it is excellent at pointing out problems with the page.
httpArchive is different. It tests the home pages of the top 1 million websites, at regular times each month. It is based on WebPageTest. It’s brilliant for showing the breakdown of content types e.g. images and shows historical trends. Even if you are not in the top 1 million you can use it to benchmark against the big boys: individuals, top 100, top 1000, top 1 million.
For this random test, let’s pretend we are the biggest retailer in the world, and compare against the biggest online retailer:
Mobile speed test performance test for Walmart.com:
- httpArchive – mobile page speed (January 15, 2017): 20.6 seconds. Page weight: 1941KB. 95 requests. Images: 962KB. Image requests 53.
- WebPageTest (tested on a US-based iPhone6, cable connection, (9.00 GMT, 29-01-17) – mobile page speed: 14.3 seconds; fully loaded 17.9 seconds.
- PageSpeed Insights – mobile speed score: 45/100 (poor).
It is important to benchmark your performance against your competitors’ sites, so let’s try Amazon.com:
- httpArchive – speed (January 15, 2017): 6.9 seconds. Weight: 554KB. 89 requests. Images: 259KB. Image requests 49.
- WebPageTest – speed (9.00 GMT, 29-01-17): 2.4 seconds; fully loaded 4.8 seconds.
- PageSpeed Insights – mobile speed score: 55/100 (poor).
Google PageSpeed, pictured in the image below, estimates that Walmart could save 478KB (almost 0.5MB) simply by compressing the images on the page. As can be seen from the httpActive chart, this could save as much as half the image weight or one quarter of total page size.
2. Conduct user testing
As with all aspects of web development, user testing is critical to improving site performance and usability.
- Conduct surveys and interviews with users to discover how they use your site and any pain points with the experience.
- Test and watch users as they interact with the website. Use eye tracking to see what catches their attention and which images work.
- Use heatmaps and web analytics to track how users interact with webpages and where they look.
3. A/B test webpages with different images, numbers, placement, formats and sizes of images
A/B testing shows two different versions of the webpage to different groups of visitors. Compare the results to see which types of image work best.
As we will see in the next column when we look at Unilever’s work with mobile images, user testing your mobile website is hugely important, and small changes to images can deliver big differences.
This article is Part 1 of a three-part series on how to optimize your mobile site speed. Check back next week for Part 2, which looks at reducing the impact of images on your website’s performance.
This column was originally published on our sister site, ClickZ, and has been reproduced here for the enjoyment of our audience on SEW.
When specific people, places, and things show up in queries or in web pages, that can be a signal to search engines to do something special in the results that they show. How prepared are you to understand and anticipate how the search engines treat them? Do you have a strategy in place?
We all know some internal site search engines have a lot to be desired. Many website owners choose to use Google’s Custom Search Engine tool, but if you have a basic account, your users may see Google ads and related search results alongside your website’s pages. In addition, for many site search services, you can’t tweak or modify the way results are displayed, including the order.
Ask.fm, the Latvian startup that’s been left carrying the personal Q&A torch after its inspiration, Formspring, pivoted to pastures new, is continuing to grow at an absolute clip. It’s now at 65 million registered users, up from 8 million in June last year, and is adding about 300,000 new users per-day, seeing it garner 190 million unique visitors last month. User sessions sit at a highly engaged 15 minutes on average, too. But this growth isn’t without cost.