UpriseUp - Up
UpriseUp - Rise
UpriseUp - Up
Back to EventsBack to Blog

SEO Updates for April and May 2022

April started spectacularly with Brighton SEO bringing together SEO enthusiasts from across the globe. We witnessed a great range of talks at the biannual beachside conference, covering everything from the fundamentals of search right through to the future of SEO in the ‘metaverse’. The weeks to follow have continued to offer several other interesting developments which we’re unpacking below. 

Google Released Annual Search Spam Report

In response to a world searching for ‘how to heal’, Google highlighted how they kept 99% of searches spam-free with significant improvements in fighting link spam, scam results, and ranking manipulation in their annual search spam report. 

Google also focussed on reducing low-quality content through identifying behaviours that manipulated search rankings. These behaviours would narrowly avoid violation of the quality guidelines but negatively impact user experience. With the help of their AI-based system SpamBrain, Google stated they were able to keep 99% of searches spam-free in 2021.

As ever, websites should follow best practice guidance and steer clear of ‘black hat’ SEO techniques, such as keyword stuffing and product review manipulation, to avoid being penalised by Google’s spam algorithms. Producing high-quality, relevant content for your customers will always be the best way to help improve your search rankings. 

Google Search Parameter Tool Officially Offline

Back in March, Google announced it was going to retire the URL parameters tool, and this is the first month we can see it coming into effect. Google has now turned off support for the tool in Google Search Console. The decision was made by Google to turn off the tool due to the advancement in Google’s capabilities to decipher which parameters are useful on a site. With only a minute number of parameter configurations specified in the parameter tool deemed useful for crawling purposes, the tool was deemed unnecessary. 

Google has stated that ‘Google’s crawlers will learn how to deal with URL parameters automatically’ in the near future. We would suggest making a note of this update on your reports and keeping an eye on your analytics over the coming weeks just in case any issues arise from this change. 

 

Significant Changes to Featured Snippets being Tested

Google has started some testing that may provide a major shake-up of the featured snippets section on SERPs. Our SEO Team certainly has a lot to say about these two new features:

‘From the Web’: Traditionally, the featured snippet shown at the top is a table, a list, or a snippet of text with a link to the webpage the content comes from. For text snippets, Google is now testing short excerpts from two to three other websites in the same section, with links to the sites added after the sites’ favicons. 

 

‘Other Sites Say’: Google is planning to group at least three different sites under a new ‘Other Sites Say’ section, which shares some resemblance with the established ‘People also ask’ section. Again, this will provide more exposure for brands, but equally will create more competition in the top-ranking results. 

  

What could this mean for search?

Sites that currently hold the featured snippet position for certain keywords could face a substantial loss of traffic as more competition enters position zero in SERPs. On the flip side, if you’re not currently featuring in any snippets, this update could increase your chances and improve traffic volume to your site. 

It will be very interesting to see the impact of these tests on clickthrough rate (CTR) and visibility in the search results, and whether these updates are rolled out temporarily or permanently. One to keep an eye on!

Google PaLM: The Future of Next Generation Search

This month Google revealed a breakthrough in its efforts to create an AI architecture that can handle millions of different tasks by itself. Enter PaLM.

What is PaLM?

Google’s Pathways Language Model research (PaLM) is an AI architecture Google has been developing. PaLM can produce answers reflective of fluctuating contexts by learning how to efficiently solve millions of different tasks, including complex learning and reasoning. 

What makes PaLM special?

PaLM is a system worth recognising as it’s striving to combine the efforts of multiple existing AI systems, into a singular architecture. To achieve this, recent developments of the PaLM system have involved the scaling of the few-shot learning (FSL) process. This is a type of machine learning method that works with a limited training dataset, as opposed to deep machine learning, where an extensive amount of data needs to be manually input for the AI to learn each new ability. Essentially, FSL has the AI learning so it can make predictions based on a smaller dataset.

Recently completed was the BIG-bench benchmark, where several tasks were designed to see how large language models, such as PaLM, responded. Of the 150 strong BIG-bench tasks (relating to reasoning, translation, and question answering), PaLM outperformed many of the current state-of-the-art models. There were many notable achievements on hundreds of language understanding and generation benchmarks, including: 

 

  • Enhanced reasoning abilities 
  • Explanation generation 
  • Inference Chaining

 

This recent research shows PaLM delivers significant improvements compared to current AI systems and can even ‘outperform human benchmarks’ for certain elements of language processing and reasoning. However, humans still outperformed the new algorithm on 35% of tasks. So, whilst breakthroughs are being made, PaLM is not quite there yet. 

What could this mean for search?

Machine learning has a big impact on how search results are created, tailoring results more and more to the needs of the user. As PaLM seeks to consolidate all this machine learning into one AI system, the change to search may not be great. However, with capabilities in one place, it may mean Google can get an even greater understanding of the intent and needs of users when they use search engines. Either way, this is an update to keep an eye on.

Did we miss any SEO news?

Think we may have missed something worth exploring or if you have some thoughts you’d like to share on SEO developments? We’d love to hear from you! 

Join the conversation and tweet us @upriseUPSEM, email us at [email protected], or simply send us a message through our contact page.

    Did you enjoy this blog post?

    Share this article:

    Back to EventsBack to Blog

    How to successfully run responsive search ads (RSAs)

    Not long ago Google announced the end of an era for expanded text ads. As of June 30th 2022, you will no longer be able to create or edit expanded text ads. 

    The announcement surfaced last year and here at Uprise Up, we’ve been preparing our client’s accounts ever since. Introducing responsive search ads (RSAs) into each of our ad groups ahead of June 30th. To help you also get ahead and be fully prepared for the change, we’ve jotted down our top tips in this blog for running successful responsive search ads. 

    First things first, what are responsive search ads?

    Responsive search ads (RSAs) are another step in the direction towards automation from Google. Expanded text ads (ETAs) had a set of 3 headlines and 2 descriptions that are shown statically, whereas RSAs allow us to select up to 15 headlines and 4 descriptions. Google then automatically tests the different combinations of these headlines and descriptions to give the user the ‘best’ performing combinations. 

    Ok, so what are the potential benefits of RSAs?

    • Improved performance. According to Google, advertisers that add RSAs to their ad groups achieve up to 10% more clicks and conversions. From our experience, we’ve also seen RSAs often out-perform existing ETAs when added into our accounts. 
    • Increase ad relevance and reach. More headlines and descriptions mean Google can serve more relevant combinations to the user. With more keywords in your ad copy, you’ll be entered into more auctions for relevant searches. 
    • They’re a time saver. Instead of needing to set up multiple variations of ETAs to test and learn, you only need the one responsive search ad which will test the combinations automatically.

    One thing to note is that while Google’s auto suggestions can often be useful, they are equally often not so useful. We’d advise taking a cautious approach when applying these.

    So, on the flip side, what are the potential downsides to RSAs?

    • Less control. Your ability to specify how an ad is formatted and reads overall is limited, due to the nature of the machine learning testing various combinations. This may lead to headlines appearing together which don’t necessarily work well or make sense to a user, or for your brand.
    • Reduced learnings. You cannot see as easily which headlines and descriptions have the best CTR and conversion rate, and therefore might work well outside of Paid Search.
    • Can actually take more time to select headlines and descriptions that work well together, but are unique enough, while also assessing whether to make use of the pinning feature (discussed below) can actually be more time-consuming than creating a standard ETA.
    • Beware of auto-suggestions. Google will be missing important context, so not all suggestions will be relevant.

     

    How to Run Responsive Search Ads Successfully

    Top tip time:

    • Include keywords in your headlines. To reach those good and excellent ad strengths you’ll need to make sure you have headlines relevant to your keywords. You can also use dynamic keyword insertions to insert your keyword into headlines, from experience this will help to optimise your ad strength. 
    • Include unique headlines. To give Google the variation it needs to test and optimise your RSA, you’ll need to keep your headlines unique. Try using a variety of calls to action and offers to improve headline uniqueness. 
    • Have a combination of short and long headlines. ‘Long’ headlines being within the 30 character cap.
    • Use all the headline and description fields available. If you can aim to fill out all 15 headlines and 4 descriptions, at a minimum include 10 headlines. 
    • Pay attention to ‘ad strength’. Google will offer you suggestions to improve the ad strength of your RSAs. You’ll want to get the ad strength up to at least “Good” but ideally aiming for “Excellent”.

     

    To pin or not to pin, that is the question.

    Responsive search ads are far from perfect, we’ll still quite often see Google pair similar headlines together as the highest serving combination (e.g. two branded headlines rather than a branded headline and a CTA). There’s definitely still questions to be answered. 

    Our biggest one is around the pinning feature. When setting up your RSA, you have the option to pin a headline or description so that they only appear in a certain position. While this sounds great (especially for controlling brand messaging), the ad strength of the ad is very much affected by the use of pinning. A lower ad strength may impact your achievable impressions share and your CPC, and may result in lower impressions/clicks as a result.

    With that in mind, you may be wondering: 

    • What is the actual impact of a lower ad strength on the total impressions?
    • Does this impact outweigh the benefits of improved brand messaging?
    • How do we best use pins to balance this impact?

    Fortunately, we have sought-out to find the answers!

     

    What we’re testing

    We’ve set up an experiment to test the pinning feature specifically. We’re running A/B experiments to test RSAs with no pinning, fully pinned, and a balance of pinning. 

    Specifically one thing we’re testing, is how the number of pins effects ad strength. For example will pinning 4-5 headlines in a single headline position still allow for a stronger ad strength compared to 1-2? 

    We’ll also be testing the impact of losing an ‘excellent’ ad strength in favour of pinning, looking at the effects on impression share against conversions. 

     

    We’ll be running this test over the next few months and look forward to sharing the results once they’re in.

    Here’s some examples of the types of variations we’re testing:

    Excellent

     

    Good

    Average

    Poor

     

      Did you enjoy this blog post?

      Share this article:

      Which newsletter(s) would you like?
      Marketing Permissions

      Uprise Up will use the information you provide on this form to be in touch with you and to provide updates and marketing. Please let us know all the ways you would like to hear from us:

      You can change your mind at any time by clicking the unsubscribe link in the footer of any email you receive from us, or by contacting us at [email protected] We will treat your information with respect. For more information about our privacy practices please visit our website. By clicking below, you agree that we may process your information in accordance with these terms. We use Mailchimp as our marketing platform. By clicking below to subscribe, you acknowledge that your information will be transferred to Mailchimp for processing. Learn more about Mailchimp's privacy practices here.