We have at least 20 projects with direct clients and agencies in various stages of setup in our status tracking. Our oldest project is 19 months, and another is 15 months along, with an average from start to first phase launch taking 93 days.
I took some time this week to understand why these were taking so long to get going and how we can improve our process to accelerate them. The most common reason for the delay is what I now call “Perfection Paralysis.
Perfection Paralysis is caused by the need for the HREFLang XML output to be perfect before it goes live. While we all strive for perfection, HREFLang XML is, by nature, a work in progress that can and should be deployed in phases.
The most common case is the lack of perfection due to complications uncovered during the process. Many projects can be built and deployed in a few minutes or hours. Unfortunately, the more complex your web infrastructure, DevOps, and SEO programs are, the more challenges you encounter.
Another key driver of these delays is most people treat HREFLang as a checklist item that cannot be completed until perfect. Also, many of our outstanding projects are agency led with a specific scope of work calling for them to implement a hreflang XML for the site, and they cannot check the box for this until it is complete.
The unfortunate reality of all of these projects is that none of them are reaping the benefits of hreflang currently and, in some cases losing $1 to 3 million dollars a month. The industry needs to try to educate clients that they should not treat this as a one-off project but as part of their retainer work that is more iterative.
To overcome these challenges, we often suggest that clients focus on getting the work done in steps and iterate until we get to a level of perfection they can be happy with and achieve greater success. We run into these perfection speed bumps at three phases of deployment.
The first problem starts with challenges of getting source URL’s. Many of the problems detailed in that post illustrate the challenges people have. You can easily get a working version by using existing XML or quick crawls of key sites.
Our second-speed bump is the mapping of URLs. In a perfect world, sites would have logical organization and uniform URL naming structures. When they are not uniform, and we can use one of our mapping techniques, we can get over this hurdle very quickly. Unfortunately, when we cannot, we have to revert to manual mapping, and things halt since no one budgeted for the time it takes to map the alternate URLs.
The third speed bump is the challenge of getting the final XML files on the web. We had one client where the IT team only made deployments/uploads twice a year and another required a ticket and a wait time of six weeks. We have developed several ways to accelerate this, including the use of cross-domain hosting, and even that has challenges we never expected.
The best way to break down Perfection Paralysis is to educate clients that they should not treat this as a one off project but part of their retainer work that is more iterative.
The more experienced agencies make this a multi-phase project. The goal of phase 1 is to get a reasonable version of the HREFLang live so we can get immediate benefits. The second phase is to work with IT and DevOps to fix any major issues identified during the setup and initial implementation. Phase 3 is the finalization of the automation that lets it work in the background with periodic checks and updates.
Start with the markets where you have the greatest content duplication, incorrect page ranking, or, more importantly, where cannibalization is losing the most money. Use the same methods you used to determine why you need to use HREFLang it should have shown you where you have the greatest problem.
You can also focus on the markets that get the most executive attention or are classified as priority or tier 1. One project the client had 147 sites but only 11 were in Tier 1 markets and all of. them had uniform URL structures allowing them to be mapped almost instantly.
For this project, more than half the sites were classified as tier 3 and had less than 100 pages making it easier to justify working on them later. By focusing on those 11 priority markets, they were able to show improvement that justified resources for the tier 2 markets which required more manual mapping.
Start with the markets where you have the greatest duplication or an incorrect page ranking. Use the same methods you used to determine why you need to use HREFLang it should have shown you where you have the greatest problem. Focus on markets that get the most executive attention
When we start a project, we pull in all the URLs we can get from any source, use the mapping strategies and build the master XML and then see what we get. Just loading that version of the HREFLang XML will start the process of Google understanding the relationships and help you to find any other speed bumps or friction in the process.
The reality is you just need to get started. It is very frustrating when you get the URLs and map them, only to find a complex process to get the files loaded. Had you tried to load an initial build of the files, you could have run a parallel project to improve that process. Waiting until everything else is ready only delays things longer. A side benefit of using HREFLang Builder is that you ultimately fix many other problems related to XML quality, crawl budgets, and other problems with getting indexed.