Welcome to the Q&A Forum

Some developers always implement the hreflang for German (which should be "de") as "de-de", so language German and country Germany. There is usually no other German version targeting the other German-speaking countries (mostly ch, at). So obviously the recommendation is to make it "de" and that's the end.

But I kept wondering and not finding anything: IF there is a more specialised hreflang, will google take that if there is no default?

Example:

Search in: de-at (or de-ch)

Search result has the following hreflang versions: de-de; x-default (==en), en

=> Will Google give the result for x-default or de-de?

You cannot set en-be in any way? This would be correct for targeting "people with English settings in Belgium" or "English content for users in Belgium", regardless whether it is in the ISO standard or not. Cf. Google's help document: https://support.google.com/webmasters/answer/189077

Depends ... do you WANT the brands to compete with each other? That might sound strange but it really isn't. Mabye the three brands have some feature that separates them from each other and makes them uniquely suitable for a different target group. Then it might not be bad to have all three rank for slightly different (long tail) terms AND each for the generic overall term. The brands itself also may carry different values; answering in a vacuum on what I would do is hard to impossible.

Nico

I'd have to disagree with David, strongly so.

But then it is probably a matter of what "tag page" means for you and what you do with your tag pages!

Is a tag just something you slap on articles and stuff to simply have all a page that merely displays them all? Well yeah, then I'd tend towards noindex. I'd also wonder how many people really use a tagging like that - and have a use for it.

Personally I favour the other approach where tag pages are made into a page that can stand alone, for example giving a general overview over a topic and then listing the tagged articles for more specialised content. (Yes, you need a certain minimum of articles for tags to be useful at all ... and too many of one tag is probably also a warning sign of over-specialisation.) Of course this is not possible with every website and organisation. I have found, though, that such "hub pages" work exceptionally well for more generic than long term searches.

TL;DR: The importance and treatment of tags and tag pages largely depends on how you define their role.

Nico

"100 link rule": At some point in early 2016, John Müller or sb. else at Google said "a reasonable number / a few thousand links at most". Unfortunately I did only save the statement, not the exact source/date.

Nico

I would chose variant 2. Less for an omnious Ranking Boost and more for segmentation of the site. With a subdirectory like /business/ you can analyse the behaviour in that content/business section much better than if you just put everything in a "no-subdirectory url".

No subdirectories are, imo, only useful if you have no clear sections or topics - or a single one defined by the domain.

As Marcus Miller mentioned, this has the added benefit of making sense in a vacuum. At least in my opinion.

Nico

I am wondering what exactly you mean by "not considered for SEO"?

For many purposes, subdomains are considered as independent domains. If they are excellent landing pages for a specific search, I see no reason to exclude them from appearing in organic search; else a noindex might be wise. But I cannot really imagine a good page deserving that, assuming linking to your regular page and no cannibalisation.

In general I would slap on the basics - proper description, title, maybe social media data and be done with it unless I have a really good reason to assume they are actively harmful somehow.

Regards

Nico

Ahoy!

Variously I have heard the opinion that content which does not generate regular search traffic (let's ballpark it at >10 views in any given month) should be noindexed or even removed. Allegedly this would improve the overall quality of the site, rankings and traffic.

I remain doubtful. What would you do if the interest in a given matter goes down over time for any (most) given topics of your content and is replaced by "newer" specific interest?

Concrete example:

I have a website about (book) reviews. Naturally, there will always be new books; old books are not in the media as much and "forgotten". Nevertheless, the reviews (all unique, based on really having read the books, no trace of the standard back cover copy) are obviously still there. Personally I feel that they do not really lose any value - they are still reviews of that one book, even though it is not the most recent one.

So, what would you do: Deindex "older" book reviews after a certain time? Even remove them completely? Just let them run?

I am looking forward to your opinions - and even your experience if you have done something like this!

Nico

In my opinion: No, you should not. The result will be that those URLs are indexed but with a standard description to the effect that google knows the URL exists but not what is behind it; titles may be taken from links or past data.

Canonical is the way to go.

Regards

Nico

I don' think Martijn's statement is quite correct as I have made different experiences in an accidental experiment. Crawling is not the same as indexing. Google will put pages it cannot crawl into the index ... and they will stay there unless removed somehow. They will probably only show up for specific searches, though

In September 2015 I catapulted a website from ~3.000 to 130.000 indexed pages (roughly). 127.000 were essentially canonicalised duplicates (yes, it did make sense) but also blocked by robots.txt - but put into the index nonetheless. The problem was a dynamically generated parameter, always different, always blocked by robots.

The title was equal to the link text; the description became "A description for this result is not available because of this site's robots.txt – learn more." (If Google cannot crawl a URL  Google will usually take titles from links pointing to that URL). No sign of disappearing. In fact, Google was happy to add more and more to its index ...

At the start of December 2015 I removed the robots.txt block - Google could now read the canonicals or noindex on the URLs ... the pages only began dropping out, slowly and in bunches of a few thousand in March 2016 - probably due to the very low relevancy and crawl budget assigned to them. Right now there are still about 24.000 pages in the index.

So my answer would be: No - disabling crawling in the robots.txt will NOT remove a page from the index. For that you need to noindex them (which sometimes also works if done in robots.txt, I've heard). Disallowing URLs in the robots.txt will very likely drop pages to the end of useful results, though, as Andy described. (I don't know if this has any influence on the general evaluation of the site as a whole; I'd guess not.)

Regards

Nico

In my opinion, people have been way too quick to slap nofollow on just about everything - out of envy ("do not give them juice!"), out of fear ("I cannot influence that great content on that page - it might change and even be illegal!") or various other reasons.

Would you say you have to nofollow a link given from a football fan who was invited to a VIP lounge for one game (after some raffle or whatever)? Would you say, you have to nofollow links from fair visitors who were given some goodie, enjoyed it and mentioned you with link?

I feel rather strongly about the - pardon my directness - idiocy to slap nofollow on just about anything. In my opinion they belong on exactly two links: those you directly paid for; and those you want to actively distance yourself from.

Then again, the real question might be how google sees those links, their context etc.. If you get punished for those links, I'd think that was a misjudgement - which does not help you in that moment; so if those links look very much paid for ... But as I said: you did not pay or ask for them, they are freely given. Stated generally, I see no problem from those links being just normal links. Let us not forget that there used to be just "links" - and that few people overall have ever heard of "nofollow" and "follow".

Regards

Nico

Provided you mean Google PageRank: Then, because it is discontinued (or more exactly: the displaying of page rank is discontinued), cf. https://www.seroundtable.com/google-fully-turns-off-toolbar-pagerank-feed-21946.html . So I would not expect Google PageRank to show up anywhere any more.

Regards

Nico

Hiya,

As you know, you can specify the date of the last change of a document in various places, for example the sitemap, the http-header, ETag and also add an "expected" change, for example Cache-Duration via header/htaccess (or even the changefreq in the sitemap).

Is it advisable or rather detrimental to use multiple variants that essentially tell browser/search engines the same thing? I.e. should I send a lastmod header AND ETag AND maybe something else? Should I send a cache duration at all if I send a lastmod? (Assume that I can keep them correct and consistent as the data for each will come from the very same place.)

Also: Are there any clear recommendations on what change-indicating method should be used?

Thanks for your answers!

Nico

I would not use a canonical here. I'd implement a logic that sets empty categories on noindex as long as they are empty IF they are really, really useful and needed there, which I kinda doubt. Might be if you display a "no entry yet - place your location here" registration form or something like that. Nearly as likely I'd just kill them and recreate them when/if there is actually useful content for the page. Core question I'd ask is: What exact use DO such pages have for users/search engines? (and NOT: What future use might they eventually have some day for you/others?)

Nico

You can also very likely change this directly in the wordpress settings without fiddling around in plugins. Normally this Suffix comes from the "website name" you enter in "Settings" > "Tagline".

Many people tend to enter the whole, long name of their business here, which is quite commonly very, very long and thus exceeds the displayed title limit.

Nico

Personally I'd use Screaming Frog for this. The URLs for prev/next elements can be found in the "Directives" tab and can be exported to excel for easy comparison.

Regards,

Nico

... ah, too slow and essentially same answer

Actually, you have just one option to not index them - the second one. The first will, still keep them in index if google can find them. I currently have roughly 27k URLs indexed that were blocked via robots.txt from the start (generated with a time-based parameter; yeah: ouch.).

Those results do not usually appear in "normal" search but can be forced (currently you may try site:grimoires.de inurl:fakechecknr and showing skipped results to see the effect of that). So basically I'd advise against using robots.txt - it does not prevent indexing, only the visiting/reading of that page.

Regards

Nico

There has even been a Google Webmaster Guidelines Update in February 2016 which states

"Limit the number of links on a page to a reasonable number (a few thousand at most)."  (Source: https://support.google.com/webmasters/answer/35769?hl=en)

So I really would not bother too much, especially not in a navigation - it often makes lots of sense to have lots of links there. (For example I have several alphabetical selections available on hover from main categories. It would not make sense to remove them just to have fewer links.)

More links are of course not always better - consider what the user is likely to expect/need in navigation etc. Of course, more links mean that the relative importance of each link decreases; but google is able to identify navigation and repeating elements that appear on every page. I'd assume that they treat them different to main-content links. Because, well, they feel a lot different.

Regards

Nico

It should not make any difference and is a question of style/syntax. The forward slash at the end is XHTML-Standard-Syntax. Sometimes, some validators may complain about one way of writing it or the other. Compare other Elements:

works if you write it just once at the start of a new parapgraph. Properly, a paragraph should be encased by

TextOfTheParagraph

, a DIV goesDivContentetc. Some elements ("void elements") do not have a beginning and end as such, for example a line break (
) or an image () or meta tags. Those can be written just like before or in XHTML-Syntax
indicating that the element is opened and then closed/finished by itself - there will not be a closing tag for it later on.

Main point speaking for XHTML (HTML+XML) is that it is really well structured. Cf. http://www.w3schools.com/html/html_xhtml.asp

Regards

Nico

In the revised webmaster guidelines, google says "[...] Provide a sitemap file with links that point to the important pages on your site. Also provide a page with a human-readable list of links to these pages (sometimes called a site index or site map page)." (Source: https://support.google.com/webmasters/answer/35769?hl=en)

I guess what they mean by this is something like this: http://www.ziolko.de/sitemap.html

Still, I wonder why they say that. Just to ensure that every page on a site is linked and consequently findable by humans (and crawlers - but isn't the XML sitemap for those and gives even better information)? Should not a good navigation already lead to every page? What is the benefit of a link-list-page, assuming you have an XML sitemap? For a big site, a link-list is bound to look somewhat cluttered and its usefulness is outclassed by a good navigation, which I assume as a given. Or isn't it?

TL;DR: Can anybody tell me what exactly is the benefit of a human-readable list of all links?

Regards,

Nico

Let me ask a counter-question: If a website is bad quality, why would you link it at all in the first place? (Barring that you want to give a negative example ...)

Imo the use of "nofollow" should be clearly restricted to paid links and ok, maybe forum/comment links. It is somewhat idiotic to slap a "do not follow or count this link" on a link you set yourself. A link you set yourself with a purpose, because it is relevant to the topic, gives additional information etc. If I was a search engine, I'd really consider NOT honoring the nofollow tag when somebody puts it on every link. To me it also feels somewhat dishonest: You point to a good ressource and also say "but do not give them any credit for me linking and recommending it!". Never mind the human visitor does not see or feel the nofollow in any way at all... and when user signals get more and more important, why exactly should search engines (continue to) honor something that has no bearing on human users at all?

In short: IMO nofollow belongs on paid links and on links the website owner has no way to endorse/influence (comments &c.) if necessary. If you set links yourself, it should be a normal follow link. If you want to put nofollow on it, question why you want to set this link in the first place.

Nico

My take: Do not use bold/italics/underline to mark keywords; use them to mark up important parts in your text in a consistent way. That can be the same but needn't be. For example, I tend to italicise proper Names and quotes; I tend to bold the most important points of a text so somebody can skim the page, immediately see the most important points and decide to read the rest or not (in addition to subheadings); I outright avoid underlines. Those are for links.

Nothing of that has to do with search engines. Its aim is improved legibility/"scannability". I doubt google would directly penalise for marking keywords up like this - it might worsen the user experience, though.

Nico

I did a migration in December 2014. The old domain was <old_brand>.TLD . It moved completely to magazine.<new_brand>.TLD which was previously empty - for non-SEO reasons mostly. (So no moving via WMT).</new_brand></old_brand>

Results was a somewhat reduced traffic for the first ~1 month or so. Afterwards the rankings transferred well to the new page. So, yeah - it definitely does help rank up and relatively fast. Note: I redirected each article to its equivalent - the structure was not changed in any noteworthy way, just the (sub-)domain. The old domain was ~11 months old, had close to 4-digit traffic per day and had hardly any external links at that point - those it had were set without request.

What I find funny: the URLs of the old domain are STILL within google's index, despite 301-redirecting to the new domain.

Nico

It is no secret that good meta descriptions should be written to incite the searcher to click on the result without misleading them. Time and again I read that there are measurable effects by including "strong" CTAs (calls to action).

What constitutes a call to action seems by some to be taken really narrow (i.e. "Click here to learn more!" - a very specific action that is spelled out) and by others rather broadly ("... Offer available till December 31" - only implicit, the action [buying/securing] not even mentioned).

I now wondered: Many "guides" still recommend rather blunt calls like "Click here", "Read more", "Discover how". Personally I find those really unattractive and often a waste of space. However, I am not the benchmark and favour the informational side perhaps a little too strongly.

Do those direct but general CTAs really work well in every case* or should one be more elaborate/indirect? I am looking forward of hearing your experience/opinion!

Nico

• Yes, of course it is "test, test, test" and to some degree each case is different; looking for general patterns, though

Thanks - so I have to continue the search for where a tenfold increase in indexed pages (according to Search Console) might possibly come from. Sadly, the rest of your reply misses my problem; probably I have been unclear.

The reason I was asking for a method to know what pages ARE indexed is: I seem to have no problem getting stuff indexed (crystal-clear sitemap with dates; clear link structure &c.) but google seems over-eager and indexes more than there really is. If it is some technical problem, I'd like to fix that - but Google does not show anywhere what pages are actually indexed. There are lots of methods around - but none that I found do work as of now.

I have been well aware of JumpTo-Links, as I stated, and it works nicely. No problem at all with "not enough" indexed pages - really rather the opposite with no idea what causes it.

Regards

Nico

Hi,

Back in September, I added a function that sets an anchor on each subheading (h[2-6]) and creates a Table of content that links to each of those anchors. These anchors did show up in the SERPs as JumpTo Links. Fine.

Back then I also changed the canonicals to a slightly different structur and meanwhile there was some massive increase in the number of indexed pages - WAY over the top - which has since been fixed by removing (410) a complete section of the site. However ... there are still ~34.000 pages indexed to what really are more like 4.000 plus (all properly canonicalised). Naturally I am wondering, what google thinks it is indexing. The number is just way of and quite inexplainable.

So I was wondering:

Does Google save JumpTo links as unique pages?

Also, does anybody know any method of actually getting all the pages in the google index? (Not actually existing sites via Screaming Frog etc, but actual pages in the index - all methods I found sadly do not work.)

Finally: Does somebody have any other explanation for the incongruency in indexed vs. actual pages?

Thanks for your replies!

Nico

I think that page anchor's importance is increasing. Moz has pointed out that long content tends to rank better (for in-depth articles) than short content. The (shortened) logic being that it is better to present one articles that gives readers everything (and maybe a bit extra) rather than forcing them to do another search.

Personally I am a fan of a Table of Contents that links to the sub-headings in an article. (For Wordpress there is the rather good plugin Table of Contents Plus. It ads an anchor to all <h[1-6]>with an id that equates to the heading, replacing certain invalid signs.)  Google also likes this and USES the anchors: When you Google for https://www.google.com/search?q=wikipedia+en+SEO+Methods&ie=utf-8&oe=utf-8 , for example, you will get an extra "Jump to"-link that gets you directly to the relevant paragraph (or what google thinks is the most relevant paragraph for your search query anyway). This, of course, needs anchors.</h[1-6]>

I do not know how many people actually use these anchor-links and would be very much interested in that data. Personally I find this very nifty and would recommend it. It is not that much extra effort in most cases and it adds a very useful feature - also for linkbuildling and sharing.

You should of course consider possible implications for tracked scroll-depth and such stuff!

And for exactly that reason I would here chose different pages for tables, chairs etc - how would you know what people were searching for exactly? (So far I mostly used anchors for articles, not for different product sections). As I see it, that starting page is more a "furniture" landing page than a dedicated "tables" or "chairs" landing page.

Hi,

Thanks for your valuable advice! I fear, due to factors outside my influence, it is a little more complicated.

• AFAIK the group practice still exists ... at least technically.
• It does have a website with a (very) good and fitting domain name that is slightly more than a placeholder.
• Its G+ Page is basically empty.
• There is a Facebook Page for the group practice.
• There is legal dispute(!) about the ownership of the domain and in connection with the group practice. Unfortunately, this limits my actions quite strongly.
• Both their practices are still part of the same Clinic. (Before, they shared a practice at a different location.) The clinic is not important for me.
• The telephone number listed with the group practice goes to the other doctor. I assume that this is a major point of trouble - as above, I probably have currently no way to influence this.
• So: Is there anything I can do if telephone numbers ARE wrong but lead to somebody who has no interest in surrendering those phone numbers or having the entries changed?
• (a former duplicate GMB entry for the client has already been resolved)
• sameAs: thanks for the reminder! I indeed forgot this tiny bit when doing the complete schema.org suite on the website (to good effect, btw.)
• Both doctors have different domains now ... and the one group practice domain; the GMB entry is connected to the Website of Doctor A, though.

So again, thanks for your hints!

Nico

Hi,

One client has a website, Google My Business etc of his own. He ranks ok to good locally for search terms.

However, his entry simply won't show up within the local n-pack (where it objectively should) and also does not appear in the map. It seems to me that instead a group practice with a colleague that has both their names in its name/title. (Moreover, it is in the same spot - they decided to go with different websites and entries of their own, though.) For some reason, this practice is also connected to the ranking website of our client.

I suppose (NAP problems and previously used phone tracking numbers aside) that this group practice essentially blocks the real client-entry from appearing. Has anybody made such experiences?

(My provisional ToDo would look like: Disconnect the group practice from the client's website; erase/merge it if possible; do proper LocalSEO otherwise.)

Regards

Nico

Hi,

First of all: I think apart from QDF results, effects of this are rather small and trumped by such things as the actual content and value a page offered. Nevertheless I got to wondering how the publication date and modification date are used ... effectively and correctly.

Fact: Google displays the publication date on SERPS (if it is given via schema or through the CMS or in any other form). This also applies if you have a date of last modification, for example via schema.org/dateModified - regardless of the extent of changes.
Google only considers the publication date. Google also uses it as an indicator for "freshness". There are quite a few articles on that out there, ex: http://www.kevinmuldoon.com/change-date-article-boost-seo/ and http://www.viperchill.com/new-seo/

Q1: In my opinion, faking the publication date is at the very least a darkish grey area which nonetheless seems to still work. Would you agree?

Q2: Would you see it as legitimate to (at some point after thoroughly reworking one page) update the publication date to the date of republication?

Case in point: I have a page with book reviews. These reviews do not really go stale - much like recipes; tastes may change a bit, but essentially it stays the same. I find it somewhat irking to see a 10 year old date there - even if I maybe have restructured and rewritten, maybe even completely redone a review...

But apart from the question of whether to ever "update" your publication date. I started pondering when it was proper to change the modification date (especially as it seems to have little effect apart from serving as date for last changes in headers, caches etc.)?

For example, content changes when

• Manually changing text
• a visitor leaves a comment
• a visitor gives a book/article/page a rating
• a visitor gives a book a rating and this rating is part of another entity's aggregate rating

Q: Which of these events would warrant an update of the last modification? ratings and aggregate ratings typically only change single numbers (vote count and sums/averages); yet there is [legitimate] change and it is utilised in SERPS (review stars).

I am still hesitant.

My answers would be: Changing the publication date might be valid in case of a MAJOR overhaul with new or lots of extra content - when, for example you could publish the same article again in another issue of the same print magazine the article has been published in before; and all of those changes warrant an update of the last modification, at least as it is currently used, i.e. only to show when change has happened with any real influence.

Personally I'd wish for lastModified carrying more weight compared to pubDate AND especially for more google-side checks if actual change has happened. (To be ignored in case of small things like legitimately switching a sentence or correcting a typo; to be penalised if changed when nothing really changes; to honour when real change happens)

Looking forward on your opinions for dating content - and of course on your hints what I am forgetting.

Nico

Thanks for the input, Dmitrii and sorry for not replying sooner.

My experience is a bit different - maybe because it mainly stems from well-structured pages that was daily reindexed. Breadcrumbs generally showed up "almost instantly" (i.e. next index round) or "not at all". Same with some other cases of schema - even being deindexed and reindexed (be it after a silly mistake or the realisation that google for some unclear reason dislikes some markup within some others) within the same week has happened to me and is far from extraordinary.

My prior experience with (old) schema.org breadcrumb markup was frustrating to the point of avoiding it; but next time I shall give it a try again. Always better to not mix syntaxes.

Hi,

TL;DR: Does https://schema.org/BreadcrumbList work?

It's been some time since I last implemented schema.org markup for breadcrumbs. Back then the situation was that google explicitly discouraged the use of the schema.org markup for breadcrumbs. In my experience it had been pretty hit or miss - sometimes it worked without issues; sometimes it did not work without obvious reason. Consequently, I ditched it for the data-vocabulary.org markup which did not give me any issues.

However, I prefer using schema.org and currently a new site is being designed for a client. Thus, I'd like to use schema.org markup for the breadcrumb - but of course only if it works now. Google has dropped the previous warning/discouragements and by now lists a schema.org code https://developers.google.com/structured-data/breadcrumbs based on the new-ish https://schema.org/BreadcrumbList.

Has anybody here used this markup on a site (preferably more than one) and can confirm whether or not it is reliably working and showing the breadcrumb trail / site hierarchy in the SERP?

Thanks for your answers!

Nico

Thanks for your thoughts! That surely is some stuff worth thinking about - and reminding me that many people are a lot less familiar with online vocabulary than us her.

In the current case it does not really solve the problem due to it being a bit of a mixture of things like /motifs/ /themes/ /objects/ /characters/ /topoi/ etc. as they appear in a certain kind of literature.... but I really do not want to use multiple URLs for that. I might end up settling for /info/ as catch-all - and give it a more meaningful breadcrumb probably.

Regards

Nico

TL;DR: Do parts of URLs that are used throughout the web quite consistently have any influence on robots (or users)? Are there any studies? What would you use for pages that are something between a tag-page and a wiki-like article?

Long version:

On a site with a lot of content, I decided to go for tags to present articles on that topic together. My first thought was to simply list those under the URL /tag/{Tag_Name}. Short. Simple. Grabs the core meaning - on this page you'll find stuff about the tag.

But: those tag-pages will be more than just lists of the tagged pages (let's say they are articles on various topics and products with certain attributes and the same tag can apply to a product and an article). The tag pages themselves will often talk a lot about the use of said tag - extensively, without blabbering. It is aimed at being a landing page and hub for the tag/keyword. Having this in mind, I pondered using /wiki/. It does fit in some respects, but it really is not a wiki.  /info/, /lexicon/, /knowledge/ and other ideas came to mind but the more I thought the weirder I did find most ideas.

What I am now wondering:

1. Do these parts of URLs (/tag/, or /product/, or /wiki/) that are not really keywords in most cases have any influence on search engines? They are used quite consistently across the web and therefore could be used as signals. I suspect, though, that they might have more influence on shaping user expectation. (If I see /wiki/ in an URL or site hierarchy display (breadcrumb), I expect ... well, a wiki-style page; if I see /tag/ I expect a collection of stuff with that tag.)

2. What would you chose if it is not quite a tag, nor quite a wiki but something in-between? Or do you think it does not matter at all? (Breadcrumbs will be used and google has used them for display in just about all SERPs.)

3. Are there perchance any studies concerning these parts of URLS?

Regards

Nico

Thanks, guys, for confirming my thoughts on this! I shall surely change those links and make something more useful out of them.

Regards

Nico

More than "minimal", "working" html, css etc is probably a ranking factor - if not directly, then via good UX.

"Minimal" is a somewhat  problematic attribute in my opinion. In some context, you can have really minimalised code; in many contexts, best practices should be applied - if only to better maintain code, markup etc. Sometimes you do face technical or ressource restrictions and sometimes, non-minimal practices are just better. (All-time favourite: A certain wordpress plugin that minifies code, in theory speeding up loading time; in practice it oftentimes quadruples the loading time due to its own execution. Nothing won there.)

Moz has now taken to see  small markup as a spam signal (whyever) - one of my pages gets the warning that  "Site Mark-up is Abnormally Small There's a high ratio of visible text compared to HTML, JavaScript, etc." Well yes, it is self-coded and has pretty much nothing that is not needed, making concessions only for having source code that is readable by humans as well. This should for sure not be seen as a call to make bigger source code when it is not necessary.

TL;DR: Clean code without unnecessary stuff helps in maintenance and with finding errors should they appear; more than minifying code, I'd worry about page speed and functionality.

I recently discovered, that previous SEO work on a client's website apparently included setting links from subpages to the homepage using keywords as link text that the whole website should rank for.

i.e. (fictional example) a subpage about chocolate would link to the homepage via "Visit the best sweet shop in Dallas and get a free sample."

I am dubious about the influence this might have - anybody with any tests? I also think that it is quite weird when considering user friendliness - at least I would not expect such a link to take me to the homepage of the very site I was just on, probably browsing in a relevant page.

So, what about such links: actually helpful, mostly don't matter or even potentially harmful?

Looking forward to your opinions!

Nico

From what I observed, it seems to be the datePublished. Example is this page http://www.grimoires.de/buch/1/. You see "Diese Rezension wurde veröffentlicht am 31.12.2002 und zuletzt geändert am 16.05.2015." (with appropriate schema.org data) below the writer info in the main part. This info is not cheated or manipulated in any way, also used in the sitemap and accurate (it takes the unix time stamp of the time of creation/modification). In a SERP, 31.12.2002 appears. Same for another page that had the content (text) extended.

What might influence it: changes in 2015 were rather minimal - the meta description only, if I remember correctly. I am not sure, if the extent of a change might push google towards displaying the dateModified.  It would seem reasonable to me to expect a "change threshold" but in the end I simply don't know. Anybody with hard empiric data?

Regards,

Nico

You can even have it appear, be totally valid, then suddenly disappear at some point with no visible reason. By now I think of schema.org as offering google extra data for inclusion - after that it is hope and pray...

This happens quite often to my own review site which aggregates reader reviews - right now, no single aggregate review is displayed. Breadcrumbs did prove rather weird, too: sometimes schema.org works fine; at other times it does not work at all and you have to include different markup - then it works almost instantly. (Google admitted problems there at some point; does not make it less weird, though...)

Also, the "common standard" of schema.org time and again seems less common and less standard than it is oftentimes presented. (e.g.: 'till shortly google's testing tool declared isPartOf and hasPart as invalid for CreativeWork/Book ... which is, frankly, stupid. The new tool accepts it.)

Nico

I recently posed this question the other way around, i.e. a site had different pages (technically totally unconnected) for each product variant and it caused certain problems: https://moz.com/community/q/duplicate-content-through-product-variants

I am not sure if this 100% applies to you: I would not use "different model" and "variant" as synonyms. If they are different models, they are different, unique products and have distinct features, advantages, maybe extras etc. Variants, to me, are for example different colour, maybe the same model with different extras or similar. There might be some grey zone where different models might actually be quite like variants, differing little from each other - then I'd ask if they differ enough to say something about each and/or contrast them.

For different models I would in fact chose different pages that go deeper into details. (Think of different books from the same publisher!) For variants, having more than one page is problematic. If I had the free choice, I'd bundle all variants on a single page with dropdowns to select the variant. There is also the question if variants are sufficiently different to write unique content for each - if that is the case, it could justify separate pages. A definite benefit of those is that they can be linked directly.

Regards,

Nico

Thanks for the hint!

Personally, I am a big fan of schema.org and marking up all the products has been on my further ToDo list.

Hi Martijn,

Thanks for your reply. I'll have to check with the responsible developer - but I fear that this option is not on the table. Then again, I have been hinted at that a complete redesign might eventually be. As I said below: Nobody who does SEO seems to have been around when the site was created. And we all know what happens in such a case, don't we?

Hi Matt,

If it were only that easy... I have since learnt that way back when the client had that website developed he specifically asked to NOT have an ecommerce website. (I, nor anybody advising on SEO, was not around back then AFAIK.)

The products are not connected. They are litereally independently created pages with the same template. The URLs are not parameter based but look like

http://www.example.de/category/subcategory1/subcategory2/product_name-further_description_1

http://www.example.de/category/subcategory1/subcategory2/product_name-further_descripittion_2

So, identical apart from the last bit that is NOT a parameter. And the last bit might be "750-kg" or "Alu" or "with-brakes". Thanks for the advice; I agree that it is generally a good starting point but sadly not possible in this case.

Hi,

Before you shout at me for not searching - I did and there are indeed lots of threads and articles on this problem. I therefore realise that this problem is not exactly new or unique.

The situation: I am dealing with a website that has 1 to N (n being between 1 and 6 so far) variants of a product. There are no dropdown for variants. This is not technically possible short of a complete redesign which is not on the table right now. The product variants are also not linked to each other but share about 99% of content (obvious problem here). In the "search all" they show up individually. Each product-variant is a different page, unconnected in backend as well as frontend. The system is quite limited in what can be added and entered - I may have some opportunity to influence on smaller things such as enabling canonicals.

In my opinion, the optimal choice would be to retain one page for each product, the base variant, and then add dropdowns to select extras/other variants.

As that is not possible, I feel that the best solution is to canonicalise all versions to one version (either base variant or best-selling product?) and to offer customers a list at each product giving him a direct path to the other variants of the product.

I'd be thankful for opinions, advice or showing completely new approaches I have not even thought of!

Kind Regards,

Nico

Hi CleverPhD,

Thanks for your answer! The website is indeed a little dated and did not consider SEO - or so I have been informed.

http://www.g1.de/projekte.aspx is the URL with the most clearest problems, although similar tiles also exist on other pages. As you can see by checking the code, the URL is changed, albeit in a non-ideal way (parameter) and the page basically stays the same with only a tiny fraction of its content changed.

The USAtoday approach is interesting and I will look into it. I have a slight feeling, though, that the approach is quite a bit different(?).

Hi,

On one website, the "real" content is loaded via AJAX when the visitor clicks on a tile (I'll call a page with some such tiles a tile-page here). A parameter is added to the URL at the that point and the content of that tile is displayed. That content is available via an URL of its own ... which is  actually never called.

What I want to achieve is a canonicalised tile-page that gets all of the tiles' content and is indexed by google - if possible with also recognising that the single-URLs of a tile are only fallback-solutions and the "tile-page" should be displayed instead.

The current tile-page leads to duplicate meta-tags, titles etc and minimal differences between what google considers a page of its own (i.e. the same page with different tiles' contents).

Does anybody have an idea on what one can do here?

yeah, thanks. That helps.

Hi there,

I am aware that it is not good practice to have a redirect chain but I am not really sure hoe to do it (on Apache).

I have multiple redirects in a chain because on the one hand I had to redirect because the content of the site got a new URL and because on the other hand I changed from http to https. Thus I have a chain like

http://example.com via 301 to http://the-best-example.com via 301 to https://the-best-example.com via 301 to https://greatest-example.com

Obviously I want to clean this up without loosing any link juice or visitors who had bookmarked my site. So, I could make three separate redirects:

http://example.com via 301 to https://greatest-example.com
http://the-best-example.com via 301 to https://greatest-example.com
https://the-best-example.com via 301 to https://greatest-example.com

But is there a way to combine it? Can I use an "OR" operator to link the 3 conditions to this one rule?

Any other suggestions?

Thanks a lot!!!