Forum Moderators: phranque

Message Too Old, No Replies

Reviving a Mothballed Site

A 365 days project

         

lammert

1:07 pm on Apr 11, 2021 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member Top Contributors Of The Month



The Project

My first website which I launched in 1996 was--as many sites in that era--a static informational content site. First written in plain HTML, later enhanced with SSI and some PHP, I eventually converted it to a WordPress site. During that period the content stayed the same. The outlook changed somewhat with the move from hard-coded HTML to CSS and a responsive layout, but basically the site was in a frozen state for more than twenty years. In WordPress I even wrote my own theme, to mimic the previously hard-coded layout.

During that quarter of a century, it served me well. In the golden era of AdSense it was my flagship site and earned me quite some money. Rankings are still decent, but part of the evergreen content is outdated. The updates of Google in the previous ten years were effective in making a clean distinction between the real quality pages and the fluff pages which ranked good purely based on the internal linking structure, but not because of their individual content quality. But the site still has some pages on page 1 for competitive keywords.

The Goal

I started the endeavor on April 1st. 2021, and reserved a year to revive the site. My goal is to have a 10-fold increase of page-views by April 1st. 2022. An ambitious goal, I know, but the site is established, still has decent traffic and a solid back-link profile.

My plan is to report on the progress here on a weekly basis.

Deleting pages

The first step was to delete part of the content. Yes, that hurts. Some pages on the site are not on-topic, and don't have much interesting content. Other pages have lost their meaning, like a page with "last updates", which linked to updated pages in 2010, the last time I wrote some new content. And a handful of pages were in a foreign language in which I am not proficient enough to write them myself. They were translations of existing pages in English, translated by a translator I hired 15 years ago.

I removed all these pages last weekend. The URLs for the translated pages were 301-ed to their English version, the "last updates" page is 301-ed to the sitemap and the pages with little information were manually redirected to the best matching existing pages.

The page-view graphs dipped somewhat after this cleanup, but now after a week, traffic is back at previous levels. Google still has to remove some of the old URLs from the index, but I can accept that Googlebot is not that active on a site which has been stale for more than a decade.

Writing new content

Some of the evergreen content is still relevant in 2021, but as in almost all niches, interest of searchers has changed. I found a modern topic closely related to the original content of which I have a lot of knowledge and started writing new pages on that new topic. The current pace is about one page per day. These pages are not seen anywhere in the SERPs yet, but I don't care. My main focus in this phase of the project is the quality of the content, rather than the ranking.

E-A-T

15 years ago, expertise, authoritativeness and trustworthiness were mainly determined by the linking structure of sites. Much has changed and the individual content quality of pages is now an important way yo determine E-A-T. Many of the existing pages acquired their back-links naturally because of their high quality, so they fared well through all the Panda updates. But getting natural links to new pages requires more. I recently added a small biography box above all relevant content pages, and am now in the process of rewriting the About section of the site.

In twenty years I have gained a lot of expertise (and authoritativeness I hope :)). Translating that in a concise way in an about section is a challenge and may take another week or so complete.

Ditching AdSense

Yep, I removed almost all AdSense blocks on the site. The reduced revenue doesn't hurt me that much overall. Other sites are currently making much more, and it gives me the freedom to do more experiments with themes and layouts. In 25 years, the top menu and side menu are still in the same place with their menu-items in the same sequence. Part of that rigid layout was dictated by experiments in the past to determine optimal ad placement. Traffic and user-satisfaction are more important in this phase than ad revenue and it is open how the site will be monetized afterwards.

I didn't remove all AdSense code though. Under the new rules, sites need re-assessment by Google if they have been six months without ads. To prevent that from happening, I keep AdSense in some non-intrusive locations. Just to prevent the site to be removed from the accepted site list in AdSense.

What's next

The next week will be mainly used to revamp the about section, and dive in the schema.org documentation to see how I can properly mark all pages and content. I guess I'll write one or two new content pages, but that depends on available time.

not2easy

3:02 pm on Apr 11, 2021 (gmt 0)

WebmasterWorld Administrator 10+ Year Member Top Contributors Of The Month



Congratulations on starting the renewal! 365 days sounds about right for a time frame to see things happen and get further direction on interest. I hope it is as rewarding as it appears from your description here.

robzilla

3:42 pm on Apr 11, 2021 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member Top Contributors Of The Month



How fun, good luck! I'll be following. It sounds like a passion project, I'm sure that helps in staying motivated along the way.

Quick question about your goal: what's the 10-fold increase based on? It has a nice ring to it, but do you know that the demand is there?

(This will also depend on your starting numbers, of course. Going from 10 to 100 is quite different from 1,000 to 10,000.)

I had a similarly mothballed site that I revived about 6 years ago. At the time the main motivator was to crank up the usefulness of the now 15 year old site. I was unhappy with the quality of the content and user experience, and thought I could do better. Frankly, I was a little embarrassed by it. I also saw that competitors weren't providing an experience that was significantly better; it seemed they had all but abandoned the niche, even though the demand was still there. So there was some luck involved. Before I relaunched, I was at around ~200 users/day, after about half a year I was at 500, a year after that it was 2,500, and last January it was nearly 7,000 and it's been one of my main earners for the past couple of years. It has potential for further growth, not ten-fold because I don't think the demand is there, but two-fold might still be doable -- just a lot of work.

[edited by: robzilla at 4:15 pm (utc) on Apr 11, 2021]

NickMNS

3:57 pm on Apr 11, 2021 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member Top Contributors Of The Month



I keep AdSense in some non-intrusive locations.

You could use vignette ads on some pages. Like that there would be no ad-blocks on the pages, just the script in the head of the document.

lammert

4:11 pm on Apr 11, 2021 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member Top Contributors Of The Month



The 10-fold is based on the traffic to the site in the good days. It is a jump from high 5-digit page-views per month to high 6-digit page-views to a site with approximately 100 pages.

I have thought about vignette ads, but I don't like them on other sites. I feel they erode the trustworthiness of a site with their slap in the face approach. It makes me afraid it will hurt getting natural back-links.

lammert

4:28 pm on Apr 20, 2021 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member Top Contributors Of The Month



Two weeks in

Two full weeks in the process and time for an update.

Originally, several pages on the site served an About function. Some where for contact information, others for background information. All that content has now been condensed in one new page which can be reached directly from every page through the main menu. On pages with content where EAT matters, a small excerpt of the about info is placed above the content.

Getting that info properly formatted in a schema hierarchy caused some initial problems. The schema.org site was definitely written by nerds. When reading a resource about a technical system, I expect two questions to be answered, "why" and "how". The Apache server documentation is an excellent example of how that can be done. Instead, the creators of the schema website decided to focus on the "what". I'll explain this with a simple example of a property of a WebPage.

significantLink
One of the more significant URLs on the page. Typically, these are the non-navigation links that are clicked on the most. Supersedes significantLinks.

Every person on this planet reading the WebPage information on schema.org knows what a link is. Otherwise they wouldn't be on that site The adjective significant is known by most people from the age of three, four years and there is no ambiguity in the meaning of significantLink. Yet 23 words are used to explain just that. And the worst thing of all, there is no more information beyond that. No info about why I should mark significant links on a web page, and no information about how I should implement it.

If schema.org was not a brand heavily supported by a solid back-link profile, it wouldn't appear in the SERPs at all because it is adhering to almost none of Google's current quality standards. A lot of repetition and duplicate content, no unique insights and no in-depth information.

Because schema doesn't focus on the how, different standards have emerged around it which define the way it should be implemented. Microdata and RDFa focus on embedding the information inside HTML tags, while JSON-LD uses separate JSON blocks which can be embedded as scripts. I currently settled on the JSON implementation because it was best supported by the SEO plugin I already use on the site, but I am not fully happy with it yet. Implementing it this way gives mainly context-free information about a web page as a single identity, but it doesn't relate that with other pages or objects. Either Microdata or RDFa can be used in link elements to fill that gap, but that is for a later date.

Results

My comparison time-frame is the rolling 30 day period with the same period 52 weeks ago in Google Analytics. Absolute traffic figures are difficult to compare, not in the least because of the traffic anomalies due to the pandemic hitting one year ago. But some statistics are stable and are indicators for improvement.

My first goal is to (re)establish EAT with the website. Session length, bounce rate and sessions per user seem to be good indicators for that. The first two indicators tell the impression a landing page makes on users when they visit the site for the first time. The second is an indicator about the trustworthiness, and if visitors decided to come back later.

Session length is up by 11% compared to previous levels. Bounce rate and sessions per user didn't see a significant change yet.

What's next this week

In the early times, the most time consuming task when adding new content to the site was to format the pages and link them together properly with the existing content. Now that I use WordPress, a lot of that is taken care of behind the scenes. But the legacy pages still pose some problems displaying nicely using the built-in WordPress formatting because they use custom tables and description list elements. The tables support in WordPress is less versatile than I would like and description lists are not supported at all, not even in the latest version of the Gutenberg editor. I want to get rid of the custom HTML sections though. Custom HTML is difficult to manage in the multi-language plugin which I plan to use.

This will be a week with lot of late nights I'm afraid.

lammert

1:18 pm on Apr 25, 2021 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member Top Contributors Of The Month



Three weeks in

The first sites I made were built with hand-crafted HTML, simply because there were no other viable alternatives. When I wanted to expand my web presence in 2005, I decided to try CMS systems for new developments. A personal blog was created using one of the very first versions of WordPress (1.2 if I remember correctly) and for the larger sites I used Mambo. Due to a dispute, Mambo was soon after split in the original version and a community based fork called Joomla. Both forks increasingly started to differ in functionality. That was for me a sign that relying on open source CMS systems could be problematic for the long term. Except for my personal blog, I started to hand code all my new sites again.

It wasn't until a few years ago that I felt the need for a CMS again. Some projects I was involved in needed informational websites to be created in a very short time span. All the big names I knew from a decade before like Joomla, Drupal and Typo3 had faded away. Except for WordPress which had gained a lot of users in that same period.

At first I found it strange, because of all the systems I knew from the past, WordPress was the simplest and had the least built-in features. And looking at it again in a test environment, that was still the case. I looked at the database structure and it still had only a dozen tables. How on earth could a CMS so basic, take over the task to power almost half of the Internet sites?

While digging deeper, I found that the core strength of WordPress is that it can do nothing by itself. It only stores posts, comments, users, metadata and relations. That is all. WordPress does not make any assumptions about the real-world meaning of the data. No website can be generated from WordPress data without a theme. Instead of a solid website core with themes to customize the layout like all other CMS systems, the core engine in a WordPress site is the theme pulling all information from the WordPress back-end. The theme runs the site.

While these seems odd at first, the idea is brilliant. What started as a blog architecture could be used as a static site CMS by only changing the theme and the meaning of the posts and comments. If used as an eCommerce site, products are stored as posts, and order updates like payment process steps are stored as comments. The message loop to generate a blog archive page is the same message loop used to generate a product category page in an eCommerce site. Every aspect of the core functionality of Wordpress can be altered by adding hooks.

When I realized the core-stability and flexibility of WordPress, I finally understood why so many sites were running it. It was the only CMS system which didn't make any assumption about the functionality of a site. I just provided core data functionality. With that knowledge I decided to change all sites which didn't have much hand-coded unique functionality to WordPress. The site which I am now reviving was converted in the last quarter of 2019.

Finishing the conversion to WordPress

The pandemic had shifted my focus and the conversion to WordPress of the site hadn't finished yet. A lot of pages were still partly in custom HTML code because the HTML constructs were too fuzzy for the WordPress editor to determine what was going on. Tables couldn't be converted because I had used a lot of colspan and rowspan attributes, which are not supported by the native table types in WordPress. Description lists are not supported at all. It wasn't until about a month ago that I started thinking about finishing the conversion.

After more than ten years, I had now to make an important decision. Should I change the CMS in order to keep the content as original as possible, or should I change to content to fit in in a rigid CMS? To be honest, pride was more an inhibiting factor than anything else. Revenue of the site had slowly slipped away and from that point of view, even if I totally ruined the site, it wouldn't really hurt me financially. Maintenance would be much easier if I could use the full functionality of the WordPress block editor instead of having half of the pages still in custom HTML, and also using plugins for functionality, layout or SEO would be easier to use.

Therefore this week, I have converted almost all pages to be fully functional in the Gutenberg editor. It took about half an hour on average per page. Some pages were functional right away with only the need to change some image blocks, but pages with complex tables needed more time to find an alternative solution for the layout quirks I previously accomplished with the colspan/rowspan attributes.

Although the site is multi-lingual, I only converted the English pages. For the multi-lingual functionality I will use a multi-language plugin which I already deployed with success on another site. That plugin separates layout and text, making it much easier to maintain multiple language versions of a page, while keeping the layout identical. The coming week I will add this plugin and try to transfer the first pages. Because I want to keep all my original URLs, this may need some low-level configuration because this plugin normally generates their own preferred URL structure for translations.

One surprising result

No significant changes were visible in site usage last week. I saw one strange phenomena though while sifting through the site logs which defies all link-building theories I have seen here on WebmasterWorld.

About 15 years ago I started a page on the site about a specific widget from a high value brand. I didn't finish the page because it needed quite some research and my interest in the site was fading away. That specific page has never had any internal links to it. There are no external links I could find with Google, AHrefs and MOZ. The page has not been picked up by archive.org. The link to that page was accidentally included in the sitemap.xml file when I converted the site to WordPress in 2019. The SEO plugin I use decided to put in, because the state of the page was "public/published".

But now the real bummer. In my site logs I discovered significant human traffic to the page. Digging further, the URL is at spot 14 in Google USA, spot 6 in DuckDuckGo, has a special feature mentioning in Bing on the front page.

So, the three major players in the search world have found this URL through the sitemap.xml, and have given it a prominent place in the SERPs, despite not a single internal or external link to that page. The only contributing factors I can see is the trust of the domain name, and the quality of the page content itself.

And we are not talking about a low value search term. The brand of the widget is a 50 billion+ yearly revenue company with 250,000+ employees world-wide. The search term has millions of results in Google.

I previously thought link building was dead. But with this page I now start to believe that link building is instead killing, and may even prevent quality content from ranking by killing their life prematurely by adding links.

not2easy

2:03 pm on Apr 25, 2021 (gmt 0)

WebmasterWorld Administrator 10+ Year Member Top Contributors Of The Month



Interesting finding about the page status. I would have been using 'private' status setting for a page in process or unfinished. It is how I deal with avoiding a bunch of 404s on posts that have outlived their reason for being posted - or content not ready for public view. I use a generic title for seasonal announcements (as an example) and then change them to "public/published" as they are updated and the time/date approaches. For your widget content it did not have a time dependent feature and was informational - and apparently evergreen.

Also interesting is the success without link building. An idea I like personally because it is not something I have ever enjoyed putting effort toward. ;)

lammert

10:28 am on Apr 26, 2021 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member Top Contributors Of The Month



I have never done any link building in my life. The only links on other sites for which I did some effort were links in on-topic directories some 15+ years ago. I guess I am too shy to ask other webmasters for a link :) If a site offers value, links grow automatically is my experience. If links don't grow, something is wrong with the site, not with the link strategy.

I rework my sites until I get an effective traffic growth rate of 5 to 10% per month. From then on, a slow but steady viral growth occurs until the niche is saturated. And that growth is completely autonomous. Even if a Google update kicks in and reduces traffic, the natural growth rate from that new traffic level stays the same, because the growth rate is determined by a combination of site content and visitor expectations, not by Google's algorithms. While 5 to 10% may seem low, the pandemic this year has learned that an R0 of 1.05 to 1.10 can generate dramatic results after a handful of cycles. The same for a site that has been out in the wild for some years.

The targeted average traffic growth rate for the site in this topic is 21% per month, which is unusually high for my normal operation. That is why it is an experiment.

While I don't do active link building, I do passive link building. That means lurking in on-topic discussion forums, Facebook groups etc and listen carefully what visitors expect. Then, voilą, at some magic moment a page or feature pops up on one of my sites which matches what people wanted, without them actually asking me for it. As a result, people start to talk about it and link to it.

engine

10:43 am on Apr 26, 2021 (gmt 0)

WebmasterWorld Administrator 10+ Year Member Top Contributors Of The Month



This is very interesting project, Lammert, and thanks for keeping us updated.

I previously thought link building was dead. But with this page I now start to believe that link building is instead killing, and may even prevent quality content from ranking by killing their life prematurely by adding links.


I would agree 100% that poor quality linking is a potential negative signal, and should not be part of any linking program. Years ago it was a way to generate links, but that ship of value sailed a very long time back.

I would add that some positive linking activity ought to be undertaken, and it sounds as if you're taking the safe and cautious approach. Good.

lammert

5:54 am on May 5, 2021 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member Top Contributors Of The Month



Four weeks in

This week I mainly focussed on the technical side bringing the multi-lingual content properly under plugin control in Wordpress. That undertaking came with their own challenges.

I prefer to only use plugins which are either well maintained by a sizable organization or community, or plugins which are so small that I can easily maintain them in case the developer stops updating them. With only four plugins active, I ran into problems with two of them. One plugin--a paid one with an annual maintenance fee--had a clear bug, which was quickly acknowledged by the support team and scheduled to be fixed by their development team. In the mean time they provided a work-around.

The other plugin--a free plugin for some minor functionality--was so small that I could fix the problem myself and uploaded the fix with a pull request to their GitHub repository.

With two bugs found in a few weeks, one might wonder how many sites with buggy plugins are around on the internet. Most website owners do not have the skills to locate those bugs. Plugins are abandoned by their developers at some moment and may get buggy with future upgrades of WordPress. With so many sites running WordPress (more than 40% according to their developers recently) this is a reason for concern.

Results so far

Bringing the multi-lingual content under plugin control had one major drawback. I had to redefine the URL structure to match the available options in the plugin. Where the language code was previously encoded in the last part of the URL, i.e. www.example.com/sub1/sub2/XX-topic.html, I had to go with a sub-directory per language structure like www.example.com/XX/sub1/sub2/topic. With a lot of custom .htaccess work I would have been able to keep the old structure, but that would have been a maintenance PITA in the future.

I therefore decided to bite the bullet, change to the new supported URL structure and put a bunch of 301 redirects in place from the decades old URLs to the new ones.

SERPs reacted immediately and traffic is now down to 70%. I have done this before on another site and expect Google to fully return to the normal SERPs in two to three weeks. No reason to panic (yet).

The changes a few weeks ago to show more on-page authority of the topics seems to be effective. I had a steady decline in the back-links profile in the past. The trend seems to be reversed and there is an increase visible in incoming links from quality domains. Not enough yet to warrant the 20%+ monthly traffic increase I am aiming at, but it is a positive sign.

The next steps and broader plan

The first month was focused on creating a foundation for further development of the site. Except for the URL restructuring there were no significant changes in SERPs or traffic. The remaining 11 months will be focused on that.

I already started adding content about a new 2021 type popular evergreen topic. As this topic is only partly related to the existing content, traffic to the new content is still minimal. But starting this week, I will add glue content between the old and new topics.

This glue content is a major part of the site link building strategy. It is glue content for which I can create unique insights and perspective. It is a glue topic which is both interesting for visitors interested in the old topics on the site, and visitors of the new topic. The glue topic is a niche for a relative small but dedicated set of users. These users are web-savvy, are involved in working with high authority sites on the net and are often active in communities where the topics are discussed.

I will target this specific group--both on-line and off-line--with the goal that they link to the glue content in an organic way.

My hypothesis is that the matured back-link profile of my original site content, together with the fresh on-topic links to the glue content, will help to convince the search engine algorithms that the site is also authoritative for the new topic.

londrum

6:50 pm on May 10, 2021 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member Top Contributors Of The Month



i've done pretty much the same as you lammert, i revived my site on 2nd may last year so i've just passed 365 days this week. i totally re-made a huge old site and put up 50,000 pages from day one. my only link building was adding original content to some brand-new instagram, pinterest, tumblr and twitter accounts. I uploaded about 50 original videos to youtube as well.

according to google analytics the first time i reached 100 sessions in a week was mid-september, so that was after 3 and a half months.
the first time i reached 500 sessions in a week was after seven months.
the first time i reached 1,000 sessions in a week was after nine months.
from there it has noticeably sped up.
in the first week of may (so exactly one year since launch) i'm over 5,000 sessions and 10,000 page views a week

so in my experience you need to give it at least six months without worrying about getting much traffic from google. and things will really pick up once you start getting close to your one year milestone.

i wouldn't go changing the structure of your site too much in your first year either because google seems to look at it anew every time you do a big redesign, and if you do it too often you can't come back from it (that's what killed the traffic to my old site). so now i just start with a good structure and just add new, original content to it, and i leave the structure and layout well alone.

Brett_Tabke

3:33 am on Jun 8, 2022 (gmt 0)

WebmasterWorld Administrator 10+ Year Member Top Contributors Of The Month



How about an update a year later?

phranque

6:41 am on Jun 8, 2022 (gmt 0)

WebmasterWorld Administrator 10+ Year Member Top Contributors Of The Month



as of a year ago yesterday, we may never know:
[webmasterworld.com...]