SEO Helper: Automation Tools for Online Marketing

The Problems

WordPress is an excellent and powerful PHP based framework for building websites. It is well maintained and very polished, but it is also complex. The reality of onboarding clients who are already working in WordPress is that they are not doing all of the work that they need to in order to get the best SEO results.

In our experience, these are the friction points for clients who have been doing their own SEO:

1. Page bloat due to pictures that are too large
2. Lack of page retention due to lack of pictures
3. Lack of content length, something that search engines use to determine your page authority
4. Unset SEO titles, descriptions, and keyphrases in page meta-data
5. Inappropriately used alt and title tags or alt and title tags that are completely omitted
6. Broken links and mixed content when updating URLs or to an SSL protected website.

While these are very different problems with very different solutions, the root cause is the same for all of them. They are small but numerous tasks that the WordPress developers expect users to do manually on a page by page and case by case basis, which is not what users of a framework like WordPress expect when making pages is a relatively streamlined process.

But more than that while we knew the problems and how to solve them manually, asking a client to spend hundreds of dollars just to get their website to a baseline point where we could start to build their SEO is unrealistic and uncompetitive. What we needed was a platform that would enable us to insert shims as needed. [A shim in this context is a piece of programming meant to expand or patch existing code in a small way. For example, many websites utilize JavaScript shims to ensure they render correctly on older browsers.]

Problems 1 – 3

The Solution

These are content problems. While we can fix the content once it is on the website, the vast majority of clients are generating some or all of their content themselves. The best way to fix those problems is to help train the clients to recognize good SEO friendly content writing habits. Existing plugins like Yoast do most of this work for us, but Yoast does not yet include evaluation of the number of pictures included or the size of those pictures. And while Yoast does evaluate the length of the content, it’s not a clear goal as Yoast is far more nuanced than a lot of clients are prepared for.

The solution we came up with was was a simple Red [Bad], Yellow [Okay], and Green [Good] rating system for these three issues. We have currently only implemented this on blog posts, as most client generated content after website launch is confined to blog posts and we are usually directly involved in page redesigns.

The Result

There was an immediate improvement in client generated content after implementing this tests. While simple tests like these cannot replace a developer going in and further optimizing content, by providing immediate feedback to users generating the content we can minimize the amount of time a developer has to spend on any given content, maximizing the value of the dollars that businesses give us, and maximizing the value of their own time that they spend working on their website.

Problem 4

The Solution

This is something that only effects development, and thus requires a developer-centric solution. While fine-tuning SEO should absolutely be done on a page by page basis, in the absence of existing SEO or a situation where the SEO has to be changed site-wide we want to facilitate a developer being able to accomplish this task as quickly as possible.

We ultimately settled on a CSV import/export that allows a developer to export a document of all of the relevant SEO meta data for every page and post at once. They can then open this document in a spreadsheet editor like Excel and update every page’s title, description, and keyphrase without having to open individual pages to edit or waiting for subsequent pageloads. When this is done, the developer can then reupload the changed CSV file and the database will be updated to reflect the changes.

The Result

When updating an entire website’s worth of SEO data, a developer could reliably expect to spend an entire eight hour day on that single task. Between waiting for each page to load, trying to see what pages were done or not done, and checking to see if the SEO was inline with data on other pages there was a lot of time that was spent with very little gain to our business or our client’s.

By implementing this automation, a sitewide SEO update went from a task that would take hours to one that took an hour. A good developer will still spend some time reviewing content and crafting a good SEO strategy, but by removing page loads entirely and moving the data to a local spreadsheet we’ve removed essentially all of the time a developer is just waiting or trying to organize themselves. Now start to finish they can concentrate on the task they are working on meaning you get it faster and at a higher quality.

Problem 5

The Solution

Alt and Title tags are some of the most misunderstood HTML properties on the web today. Many clients have no real idea what they are used for, and many marketing companies will abuse them to pursue a gray hat SEO strategy that hasn’t worked in at least a decade. Gray hat is an internet term for someone who is skirting around the rules. They don’t break the rules outright, but they are not following guiding principles of those rules.

The only thing an Alt tag should be used for is to describe the content of a non-decorative visual element like an illustrative image. While Alt tags are indexed by image searches like Google Image, by keyword stuffing them to oblivion you are making an objectively worse experience for anyone experiencing your website with image loading turned off or through a screen-reader.

Title tags however are also read and indexed by search engines. While keyword stuffing still doesn’t work, title tags also have the benefit of appearing should a user hover over an element so it’s a useful way for adding additional optional content like jokes. If they are appropriate to your branding, jokes can create word of mouth marketing and raise engagement with your content, a trend we can observe in the wild with social media accounts like Wendy’s Twitter.

Knowing all this, there’s no good reason to not use these tags. The developers of WordPress understand this and provided a centralized way to edit them for all of the images on your website using the Media Library. What the WordPress developers left out is a way to update that content on all the existing pages of your website. Instead they again favor the developer going in an updating each instance of a photo manually. This is useful in a healthy SEO situation as repeated title tags can work against your page rank, but in a situation where non of the image SEO is done it’s a large hurdle to overcome.

Our solution to this complex problem is just a shim that takes the most recent data from the Media Library and updates all instances of that image in page and post content to include the most recent Alt and Title data. Once a developer has updated the data in the media library, pushing it out across the website is as easy as pressing a button.

The Result

Much like the SEO data from Problem 4, this was a task that seemed to stretch out for developers, but this was compounded by the fact that most pages and posts should have at least 3 if not more images on them meaning that this task takes exponentially longer than updating the SEO data.

By removing the need to individually update the images on a per-page and per-post basis we’ve completely eliminated about 2/3rds of the length of the task. Not only does this mean we can update image SEO faster than our competitors, but it also minimizes situations where your SEO is working against you which is an enormous value add for us and our clients.

Problem 6

The Solution

WordPress does not natively support changing your URL after install. There is a URL update tool, but it seems to be incomplete as it does not address links in existing content, links in plugins, and a number of other edge cases. This causes a lot of problems with changing a domain from a development URL to a live one, or even in just installing an SSL certificate. Again, it seems like the Core WordPress developers are under the impression that this task is too complex to automate, and should instead require a developer to go through page by page, plugin by plugin, to completely update and resolve.

While this makes a certain sort of sense, it is a completely unrealistic expectation of businesses using the platform. Which is why we wrote a shim that updates all instances of an old URL and replaces them with a new one in all existing content.

The Result

Launching a website was a process that used to take two or so hours. The first hour was the launch, the second hour was finding and resolving any problems the launch created in the content. This could range from broken links and mixed content, to completely non-functional parts of a website because a resource was trying to load from a website that no longer existed at that location.

By automating all of the solutions we regularly had to do manually, that hour long launch process is just an hour long process. While we still manually check for problems, we can reasonably expect that any traffic to the website during that check is going to have a reasonably safe and functional experience which reflects far better on us and our clients.

Conclusions

Automation is one of those technologies that can be extremely useful is applied with a steady hand and a solid plan. Despite what a lot of Neural Net and AI proponents argue, there are a lot of risks in automating a task. Depending on the approach you are almost always sacrificing quality and reliability to achieve speed.

The best way to remove that sacrifice is to evaluate what is better addressed by automation or human intuition. A computer is great at propagating data, but a human will always have an easier time coming up with engaging content for other humans to consume. Our approach to automation is to take these things into account. In situations where we do not want to remove the human element we will do our best to ensure they have the information they need to do the best job they can, and in situations where a human isn’t needed at all we will make sure there is sufficient information recorded to allow a human to review the work as needed.

While this may seem overly cautious considering that we have seen a universal positive gain for every element of automation we have deployed in this case-study, each and every one of these could have just as easily destroyed the websites they were deployed on, and in our development of these processes the test-bed websites were destroyed several times and had to be restored to a pre-test state. Automation is a powerful tool, and like all powerful tools it can be incredibly destructive if it is not deployed in a safe and reasonable manner.

With that possibility in mind, automation is a growing part of every industry, and ignoring it puts businesses at an insurmountable disadvantage. When applied with sufficient development and caution, as it was in these cases, it will mean more work gets done with less time which means your dollars stretch that much farther.