Following on from our interview with Andrew Smith, this week we bring you his...
Or get it on:
In this episode you'll hear Mark Williams-Cook talking about:
GSC page and speed reports: The new reports in Google Search Console showing rich product SERPs and CrUX speed reports.
Political ads: Google's new policies on showing political ads on their platform.
AlsoAsked.com: An introduction to our People Also Asked tool.
Listener Q&A: A dive into possible uses of canonical tags on e-commerce sites.
Show note links
Limits on political advertising https://blog.google/technology/ads/update-our-political-ads-policy
New reporting for Products Results in Search Console https://webmasters.googleblog.com/2019/11/products-rich-results-search-appearance.html
Google Search Console Speed report https://webmasters.googleblog.com/2019/11/search-console-speed-report.html
MC: Welcome to Episode 37 of the Search with Candour Podcast! Recorded on Friday the 22nd of November 2019. My name is Mark Williams-Cook, you're just stuck with me this week but I hope I can make your search lives a little bit easier.
In this episode we are going to be talking about the new Google Search Console releases, looking at product results and the recent experimental speed report in Google Search Console, the news on Google's limits on political advertising across their platforms, our new tool AlsoAsked.com and I've got some Q&A from a listener about how to best use canonical tags on an e-commerce site which we'll go through, explain and just delve into that a little bit more.
Since the last episode we've had a couple of new updates to Google Search Console and it’s actually a trend we've been seeing now for the last few months, which is Google does seem to be pushing all of these new features, filters and different reports into Search Console which is really great for us working in SEO and there's just been a new feature released which Google has written a blog post about, which they've titled ‘New reporting for product results in Search Console’, so what we've now got inside Search Console, if you go and log in and you look at the performance menu on the left, so that's probably one of the parts of Search Console that you look at the most and it's the part that gives you the line graph that defaults to show you the total amount of clicks and impressions your website's had in Google, along with the average click-through rate/ CTR and average position, so we're all probably quite familiar with that. And if you log in, click on performance, below that you’ll have the filters, again you're probably quite aware of things like queries, pages, countries, devices and there's going to be a new one there under search appearance, which is product results. So what this is doing is it's an extra filter that's been added, where you can now segment and look at the results of product results in Google Search. So this means where you've had a search ranking that Google's pulled up a product entity for, where they have attached information, so you've probably seen the rich results before where it has things like whether it's in stock, the price, star rating of that product, that kind of thing, so it's really interesting now that we can pull that data up separately in Google Search Console.
So Google says ‘the new product results search appearance will help website owners understand their search performance for products rich results. For example, they'll be able to answer the following questions; how much traffic comes from experiences like rich data?’ like price availability and the second question ‘how does shopping traffic change over time and for what shopping search queries is the website shown?’
As we know and as I've mentioned a few times on this podcast, it’s rare that Google does one of these moves that isn't playing into a larger strategy and I definitely think what we're seeing here is, Google know and as many SEO agencies have rightly professed and SEO professionals who said, if you get these rich results it will help you drive more traffic because the rich results are, you know they're better for users, they tend to get more engagement, so I think Google knows two things. One; if they can segment this in the report, you can then demonstrate that, look at the results we've had that have rich data associated with them like the product results get for instance, 25 percent more clicks than the equivalent position like normal vanilla result. So the outcome of this is that, it will give objective data to allow people to make business cases to provide Google with the information they require to make these rich results, which is normally schema and again that plays into Google's strategy of we all know we've got less results people are clicking on so we've got more of these what we call zero click results, where Google's are showing your data at the SERP, Google's bigger quest to become an answer engine, to not have to rely on their sort of heuristic reading and understanding of content, if they can get webmasters to label that information for them through schema, it really helps accelerate their plans in in this regard because it drives accuracy. So I think there's a definite win for Google, in that they've got a way now to help marketers, product owners, webmasters, sell that internally and get money for it.
Something I think that’s worth mentioning and I wasn't aware that a lot of people didn't know this and it's been I think most of the year, I think it was February this started, which is apart from using schema to get rich results in Google - if you're struggling to do that, so you can't get the development resources to do that, there is another way basically which is, you can actually submit a product feed through Google Merchant Center; so Google Merchant Center has historically been where you put your product data in a feed and it will then become eligible for Google Shopping ads, so that's when you do a Google search and you see all the shopping results where that structured information is coming from.
Now Google will actually use that shopping feed information to enrich organic results and in some cases it can be easier to generate a feed than insert schema, so if you've got like a magenta or WordPress site for instance, an e-commerce site and you haven't got a way out of the box to label all of the product scheme and there is normally a plugin that will either be free or very cheap that will generate these product feeds from the database for you so just doing that and submitting that feed to Google Merchant Center and I'll highlight you don't actually have to do any kind of ad spend, just giving Google that verified and labeled information can actually give you those rich results, so that’s a shortcut if you're having issues, so that's a cool new feature in Google Search Console.
Before this, so this was like in the last week, Google released this and just over two weeks ago now, so I think it's the 6th of November, I saw Google announced they have released some experimental speed data within Google Search Console and I mentioned this really quickly because I've just seen it in the podcast before - so the episode before last when we had a chat with Andrew Smith, the ex Expedia, cheap flights, eDreams SEO, that's talking about a big site SEO, I mentioned this to him and asked him if he had seen it and this was to do with the new speed reports in Google Search Console and I really really like these, so site speed I think to be honest is maybe a little bit over egged in terms of SEO, certainly not in terms of the user experience, I think site speed is really really important. There’s a really neat site called WPOstats.com - which is web performance optimization stats.com - WPOstats.com and this gives loads of soundbites of case studies of how revenues and conversions were increased by improving site performance, ie site speed and there’s some really interesting cases from big companies there, where they've increased performance page load speeds by fractions of a second and seen objective measurable improvements in conversion and sales.
From my own experience and what I've heard Google say is, they will use site speed on a preliminary basis ie if your site is awful, so it's taking like 20 seconds to load, of course that's gonna have an impact, if it's slow but still within the realm of reason, it really seems to be used as a tiebreaker; so if two sites you know the backlink profiles are quite equal, the content seems to be on par with each other, they might start using things like HTTPs signals or site speed to get that result off the fence because they can deliver the best result and the user, but as I say it is more important overall with the user experience.
So obviously everyone's been focusing on site speed performance I think really heavily the last twelve months in the SEO community and it's a really thorny issue because firstly, there's a whole selection of different tools we can use to measure site speed and all of those tools whether you use GT metrics, whether you use lighthouse, whether you use Pingdom, all of these tools will give you different results in terms of speed and actually the same tools will give you different results, especially if you're just doing like spot checking, because certain times a day, sometimes the server's a bit grumpy and you'll get a much slower return. I mean if you run the same lighthouse report a few times over a few hours, you'll see you get varying results. The location of course as well, certainly the cloud based tools will vastly change the results you get. This month I was dealing with a case with a client where they had done some speed tests and everything was reported fine, but interestingly when we were testing the site from the United States or from Canada, we were getting regular timeouts and outages, so there's quite a lot of complexity when it comes to measuring site speed.
Now inside of Search Console, there's a post on the Google Webmaster Central blog at the start of the month, it was actually the 4th of November, not the 6th, where they have released some details and it's in what they call ‘experimental mode’ at the moment but they will give you a speed report now that groups pages into very simply, like a traffic light system of red, amber, green of slow, moderate and fast pages. What I like about this data is that the source of the data is the chrome UX report; for those that don't know, the chrome UX report is real in the field user data from chrome browsers, so Chrome browser can opt in to sync and send usage data to Google and kind of phone home about the speed of the sites that it's visiting. So at least as a, lumping together generally what is the experience of our users, if you're getting 90 percent of your pages coming up as slow from the chrome user experience report from within Google search console, there is a fair bet that you have an issue, regardless of what any other tool is telling you, you've got actual user data from the field that is comparatively for other websites slow, so that I think is a really good place to start investigating if you have any performance issues, so you don't really need to necessarily start with some, very carefully set up test over a certain amount of time, you've just got the data there as long as you have enough traffic to get data back from the chrome UX report.
The only downside is that very small sites that don't get much traffic, probably won't have enough data to actually show any information in a report, but for the majority of sites that will be investing in search and SEO, you should have data in here. The other thing I really like about this is, Google has realised that when it comes to fixing performance issues on pages, they normally manifest themselves on a template basis, which means if you have a slow page, generally as long as it's not a server issue that's making all the site respond slowly, generally the performance issues will come grouped in templates, in types of pages, so for instance all of your category pages are slow because of this reason or all of your product pages are slow or all of this type of article page is slow and what this report within Google Search Console does is, it will buck it together and group together what it considers to be groups of similar pages and that's really really helpful. So, we've got a tool here that I don't think is particularly useful for real in-depth analysis of the performance, but we've got a really good place to start to say, okay we've definitely got a bunch of slow pages and they're definitely caused by this type of template and at that point you can start using your tool of choice to investigate that.
Interestingly I saw some discussion even about Google Lighthouse recently and someone saying, it was a useless tool because all of the speed tests there are based on 2G and 3G connections and it's not realistic and so on and so forth and it's actually quite possible, you can run the lighthouse performance audit through chrome developer tools, so if you open up developer tools and you've got the audits tab and you can check performance and you can actually uncheck throttling, so if you did want to run these speed tools locally with your maximum connection speed and see what speeds you got, apart from doing a report that caters to the lowest common denominator, that's completely possible as well.
So I think it's important to make sure you understand the tools you're using, of course, you're not taking spot checks, that you agree on what tool you're using and how you're going to measure and of course that you know how to use it, but as I said that's a really good place to start that's now should be in everyone's Google Search Console, it's in experimental so I expect they will be adding more features to it soon.
This week Google posted about an update on ‘our political ads policy’ and I thought this was really interesting timing, what with the UK coming up to a general election. I won't read the whole post out to you because it's fairly long, I just wanted to cover off a couple of bits, so Google says about their app platform today, it says ‘Google's ad platforms are distinctive in a number of important ways’ and this is in context to political ads they say ‘the main formats we offer political advertisers are search ads, which appear in Google in response to a search for a particular topic or candidate. YouTube ads, which appear on YouTube videos and generate revenue for those creators and display ads which appear on websites and generate revenue for our publishing partners. We provide a publicly accessible, searchable and downloadable transparency report of election add content and spending on our platforms, going beyond was offered by most advertising media. We've never allowed granular micro-targeting of political ads on our platforms. In many countries the targeting of political advertising is regulated and we comply with those laws. In the U.S. we have offered basic political targeting capabilities to verified advertisers such as serving ads based on public voter records and general political affiliations, left-leaning, right-leaning, independent’ and this is where they come on to some of the changes and clarifications.
So Google says ‘taking a new approach to targeting election ads: while we've never offered granular micro targeting of election ads we believe there's more we can do to further promote increased visibility of election ads, that's why we're limiting election ads, audience targeting to the following general categories; age, gender, and general location.’ By general location they mean on a postcode level. ‘Political advertisers can of course continue to do contextual targeting, such as serving ads to people reading or watching a story about say the economy. this will align our approach to election ads with long-established practices in media such as TV, radio and print and result in election ads being more widely seen and available for public discussion. it will take some time to implement these changes and we will begin enforcing the new approach in the UK within a week, ahead of the general election, in the EU by the end of year and the rest of the world starting on the 6th of January 2020.’ And this is where I find it gets interesting so they talk about clarifying ad policies and they say ‘whether you're running for office or selling office furniture, we apply the same ads policies to everyone. There are no carve outs, it's against our policies for any advertiser to make a false claim, whether it's a claim about the price of a chair or a claim that you can vote by text message, that election day is postponed or that a candidate has died. To make this more explicit we're clarifying our ad policies and adding examples to show how our policies prohibit things like, deep fakes which are doctored and manipulated media, misleading claims about the census process and ads or destinations making demonstrably false claims that could significantly undermine participation or trust in an electoral democratic process. Of course we recognise that robust political dialogue is an important part of democracy and no one can sensibly adjudicate every political claim, counterclaim and insinuation. So we expect that the number of political ads on which we take action will be very limited but we will continue to do so for clear violations.’
I think this is really interesting because there's been now, Twitter recently released that they're not doing any political ads on their platform, Google seems to be tightening the net here and in the last few weeks the examples that stick out for me was, there was a case in the news about the UK Conservative Party a few ago were bidding on kind of register to vote and voting terms and having their ad at the top, above the government site which allowed people to register to vote. More recently, so the Conservative’s and this is for listeners maybe not in the UK, so the Conservative Party's main opposition, the Labour Party and this week the Conservative Party had made a website called, I think it was thelabourmanifesto.co.uk and they were using Google ads to bid on that term ‘Labour manifesto’, so they were essentially bidding on their competitors political party name and manifesto to get there - I don't know if you’d go as far as to call it fake, but I think I would feel it would come under what Google here stated as may be misleading, misleading claims or could undermine the democratic process and it was interesting it doesn't appear that any action was taken. so anyone that's doing paid work in the political sphere I think is gonna have an interesting next 24 months because lots of these platforms are clamping down and generally I think we're seeing the consciousness of what is and what is not an ad in in the public domain rising, so there has been a backlash to these ads. so I just wanted to mention it as a point that Google seems to be now going along the same line as Twitter and it looks like if you're working in that sphere of political advertising, that your options have become a little bit more limited and it's going to be an interesting couple of years coming up.
On the 7th of November I published a tool that we have been working on not actually that long, for a week I guess called, AlsoAsked.com and I wanted to take a moment to explain a little bit about how it works and why we built it and introduce you to it if you haven't seen it. so it went down better than I expected, there was more interest
with it. So on the 7th of November, before Search Norwich, I tweeted that we put an alpha version up, alpha meaning that it's not finished, there were features that we would still like to add, that we haven't added and the caveat that everything might be broken, bits might be broken and things might not work quite right and what it is AlsoAsked.com is a tool for you to get information from Google, from the people Also Asked boxes. So you've probably seen, especially if you're in the UK or in the u.s., if you do a search in English for most search terms you'll now get a box in the search results of three or four different questions and it will say ‘people also asked’ and if you click on one of those questions, you'll get the answer to it and it will open up then another three or four questions that are related to that question. and this data I think is incredibly interesting and useful because it's a direct feedback loop between what people are typing into Google, how Google understands topics and questions and a real good insight into if someone types this topic into Google or this question, what is the intent or what questions should we be answering.
Now I was originally inspired by a dozen or so episodes ago I talked about a really great Python command line tool that was released, where from your desktop you could programmatically query the people also asked boxes, so you'd put in a search, it would do that searching google, it would find the questions and it would kind of automatically then expand those questions and draw you a really cool map of all the questions and how they're related to each other. it was really really good to get ideas for content, to instantly get feedback about what our content should be covering, how we should group, how it should organise it, what different entities Google think is related or people think is related to all of these topics and things and questions, it was really really useful. One of the things that occurred to me was most people probably are not comfortable running command-line Python tools, installing packages, making sure they've got the right version of Python installed; if you haven't done it before it's quite complicated, soo we wanted to make a web-based version of this which we have done.
Since we released it on the 7th, we've added some extra options, so now you can change the language, you can change the region, so you can do searches in Google.fr, in French or .de in German. Interestingly, there are a lot less people also asked results in other languages, apart from English, it seems to be predominantly English where Google is using these boxes so much. but as those results grow with other languages, the tool will adapt to this and it will provide those answers. So if you're doing a search, for instance in Arabic there aren't many also ask boxes triggered in Arabic, so it will just come up and say we can't find any people also ask results for this search term in Arabic and we've seen this actually when we spoke about the Google BERT update recently, I noted that they said the BERT algorithm is actually only in use currently in the US and in English and they were planning on rolling it out to other languages because they need to have a better understanding of the language.
So it's on AlsoAsked.com if you want to go look at it. I will put it in this podcast because I have answered this question quite a few times already which is, I’ve had feedback or seeing people saying ‘well we just use answer the public for that’ - if you haven't seen answer the public, I don't know where you've been, but it's a really great tool at answerthepublic.com, it's quite well known and if you put a topic into it, it will use Google Suggest data which is like the autocomplete data and suggested searches, to populate all other kinds of searches that are happening around that search term and it's really great for getting a really bird's eye view of a topic. It does not use what AlsoAsked.com is using, which is the people also ask data and they're two tools for two different things. So the biggest problem I see people having with Answer The Public is, they will put in a search term or a question that's too niche and the tool therefore, because it's using Google Suggest, will just return no data or one or two questions. Whereas people also ask being powered by those question boxes - even when you start to put in really long tail, specific, niche questions, you will almost always get a good set of results. So they're, well they are two different tools with two different purposes. So I like using out to the public we use it a lot for broader topic research and then when it comes down to ‘okay we've decided we're gonna write these ten articles’ that's what I would use a tool like also asked to find out what we need to be writing about in those articles. so it's free is that also ascom, do check it out and give us any feedback if you do find bugs or anything. you can now export results into CSV format or download an actual image of the taxonomy, the hierarchy, the tree of the questions so you can use it directly in your documents.
Last thing, I want to wrap up with some Q&A; so this question was actually submitted a few weeks ago and I just didn't get round to it. It's from a chap called Andy Woolley and he has asked a really interesting question around canonical tags and I was trying to dig up the question. I spoke to him today, just to confirm exactly what it is he was asking and essentially what Andy was asking was, he is working on an e-commerce site and he's asking whether he should be using canonical tags from things like product categories and using canonical tags to point that category page to the high-performing product because the high-performing product pages that are viewed predominantly through mobile, where space is a big issue, real estate is a big issue on mobile displays, they don't want to put loads of content, they want to just you know make the sale and keep it pragmatic and functional for people. would there be a benefit in, if you've got the extra content on the category page to use that as a canonical version or use the product version as the canonical version the category page, therefore the result is aiming for is, can we get the product page maybe to rank better for specific key terms or for a wider range of key terms that we might be targeting on the category page, that we don't want to put on the product page.
So I think it's the place to start there is to consider what the canonical tag is and what it's used for. So the canonical tag is a hint that we can give Google, that we have two identical or very similar pages, that we only want one appearing in search. So the most common use of the canonical tag is for things like, if we have a URL with marketing parameters on the end, because the URL is different, a search engine will count those as two different URLs and could index them both and they will then be competing with each other and that would be bad for you in terms of search visibility, having internal competing pages. You can't use a 301 redirect really because you want to have that tracking query in the URL to see what's happening,s o in these situations, in a/b test situations where you've got these query strings in the URL.
The other example is with like filtered pages, so if you're viewing a range of products and you want to sort them by price, while the order of the products and the pages is changing,
the page content is basically the same, it's just been reordered. so again that that applies, you could have that URL index but really you only want the one page. so this is what the canonical tag is for, so you take all of those variations of the page and you say to Google, look this is actually the root page that we want to rank and search, so it's a really good way to to to nip that problem in the bud.
Now I said at the beginning that the canonical tag is treated as a hint, this means that Google does not necessarily obey it. It’s been a bit cagey as to exactly what other factors they use, but they have been quite clear in their feedback multiple times saying, we use many signals apart from what do you put on a page to decide if the page should be canonicalized or not. This is something I actually tested a few months ago because at the beginning of the year where we made a page about two different brands of soda - I just made up brand names that weren't in Google, so if you Googled those brand names they were the only result. the two pages of content were written really similarly, so they're about a very similar product and they're talking about drink and quenching thirst and giving energy and it's soda the main difference was just the two different brand names of that soda. Both of these pages were indexed and they both ranked number one, as you would expect because there's no competition for those brand names.
What we then did as an experiment was we used the canonical tag to point one of the pages to the other, so the experiment was to see could we get one page to rank for both brand names so we could say, okay if someone typed in this brand name they'd just go to page a and if they type in the other one they still go to page a, rather than having page A and B - and this is quite similar to what Andy's looking to achieve here, which is he has two pages that are similar but still significantly different in some ways and trying to get them both to rank - we concluded, basically that Google completely ignored the canonical tag, so even after weeks and months of testing, Google will still 100% at the time just return that page for that search query, for that brand, it won't ever show the wrong - what it considers to be the wrong page - and this is I think, something Google's tightened up on because when the canonical tag was first introduced it was definitely being abused. We saw things happening like when people accidentally use the canonical tag on every page on their site to the home page, but then we saw the home page rankings shoot up and we definitely saw some dodgy stuff going on with cross-domain canonicals, whereby people were injecting or hackers were injecting canonical tags on some sites and pointing them towards their own sites. I think they're a lot more careful now in how they respect these canonical tags and it certainly had an impact on how equivalent the content is. so from my experience, from this experiment, if you want the canonical tag to do its job the content does need to be almost identical.
So in response to Andy; firstly, even if you tried that because the category page will list multiple products and it won't necessarily have the depth of product description that the actual product page has, my guess would be that Google will actually just straight-up ignore the canonical tag and even if it didn't ignore the canonical tag, I'm not sure that's going to be the optimal thing to do because your category pages are important. If they're high-level category pages, they're the things that are driving links to all of the individual products, they're giving context to it and they're useful for - the category page should have served a different search intent to the product page, so if someone is generically searching for a type of product and not a specific product that's when I believe and obviously there will be some edge cases to this but generally that's when you want to be serving them a category page. so it's like you're looking for this type of product you're not sure which one you would like but here is the range that we offer, whereas only if they are searching for you know a specific two inch diameter blue widget that does this, then that's when you want to show in the product page.
So the answer is, I don't think using canonical tags in that way would probably do what you're looking for it to do but certainly I'm all up for experimentation, I loved… that’s a lie, I hate being proven wrong, but I'm always interested when I am and it does happen so if you do want to test it Andy, please do and let me know the result and we can talk about it in a future episode.
That’s everything we've got time for now, we've run over half an hour already. I hope you've enjoyed this episode. I'm going to be back again in one week's time, next Monday which will be Black Friday territory, so we'll be talking about most likely what we're seeing in terms of Search, PPC and SEO around that. Hope you all have a great week, please do subscribe and all that jazz, if you're enjoying the podcast. As usual you can find all the links to the resources, at the notes at search.withcandour.co.uk.
Following on from our interview with Andrew Smith, this week we bring you his...
In this week's episode Mark Williams-Cook talks with special guest and SEO...
Get in touch