Do clients care about our tech stack?
It’s often said that clients don’t care about your tech stack, but how true is this really? Maybe they care more than we give them credit for? Let’s talk about it.
Anyone who frequents X or LinkedIn has likely seen a post along the lines of this:
It’s often expressed in the context of developers defending their technical choices for a project or in response to debates about what the most appropriate technology is to use for a project. There is an amount of truth to this statement, but there is also a level of nuance that is often left out. Especially so on sites like X where most users are still conforming to a minimal character limit.
Agencies and Staff Augmentation
The most likely time that a client will actually care specifically about the atomic units of your tech stack is in an agency or staff augmentation project.
It’s not a golden rule. I’ve been on projects where the client hired an agency, in part, to dictate best practices and the best tech stack to solve their problem. In this case, the above (fictitious) tweet rings true.
In other cases, the client may have a specific tech stack in mind that they want to use. While they may not have the software engineering knowledge necessary to truly weigh the pros and cons of a choice, they are not any less affected by the programming hype cycle. This is most common with platforms like WordPress or Shopify. The client may have heard that WordPress is the best platform for their needs, but they don’t know why. They just “know” that it’s the best.
The client may also have existing codebase or even an in-house development team that is being augmented or taking over after your deliverables are completed.
The Client’s Perspective on Results
Now that we’ve addressed the elephant in the room, let’s talk about the more common scenario. The client has a problem that they need solved. They don’t care about the tech stack, they just want results.
What does results mean in this context? What is the goal of the software being created? To quote Sunil Pai:
Silver bullets are only useful for werewolf shaped problems
Let’s look at a few scenarios that I think are worth considering.
How readily available is talent?
There is always a hope in agency life or freelance life that, assuming the working relationship is good and mutually beneficial, that a client will come back with more work. But for any number of reasons, this is not always what actually happens.
To get personal for a moment, my favorite technology that I have ever worked with is Elixir. I enjoy the language, the philosophies, the tooling, and the community. If I could wave a magic wand and only use one language for the rest of my career, it would be Elixir. However, I would never use Elixir for a client project unless it was specifically requested.
Why? Because the talent pool is small. If the client needs to hire a new developer to maintain the codebase, they will have a hard time finding someone. They may also end up paying significantly more for talent due to the simple laws of supply and demand. In my opinion, this is not a good situation to put a client in.
How stable is the technology?
Stability in this regard can have several meanings.
Will the technology be around in 5 years? Market leaders like React, Vue, and Angular have either significant community backing, corporate backing, or both that give us a reasonable expectation that they will be around for the foreseeable future. This is not always the case with newer technologies.
How often does the technology experience breaking changes? Re-visiting the previous point, how difficult will it be to find a developer who is up to date on the latest changes? Does the technology do an adequate job of staying on top of security concerns and patching issues when discovered?
What does the deployment strategy look like?
What is involved both in the hosting and deployment side of a technology is a heavy consideration that we need to think about. Can the technology be effortlessly deployed to a cloud provider? Does it require a dedicated server? Does it require a specific operating system? Does it require a specific version of an operating system? Does it have specific requirements that will tether the client to a specific cloud provider? For example, you don’t have to deploy NextJS to Vercel, but if you don’t deploy to Vercel and you make use of server actions, what is the versioning strategy to ensure that your clients don’t hit a stale API endpoint or vic versa?
Even technologies that often seem “safe” in this regard like PHP, can become problematic very easily down the road. For example, I’ve been guilty of singing the praise of PHP because you can, “Find any random Linux server anywhere and your code will run”, but this is not strictly true. In my last agency, a large amount of the work I did was when clients who were deployed to providers like GoDaddy would get a surprise PHP update that would break their site.
What sort of technical skills are necessary to deploy changes? Can a non-technical user simply login to a CMS and make changes to content? Or will they need to at least be able to use the Github (or similar) web interface to commit a change to trigger a deployment?
What is the performance like?
Performance can be a bit of touchy subject and I contemplated leaving it out of this post entirely. I think there’s definitely potential in the performance sphere to “major in the minors” and waste time and financial resources on premature optimization, but it’s also a concern that can have real impacts.
For example, if you’re building a website for a client that is going to be deployed to a country with poor internet infrastructure, you want to consider the performance implications of a client-side rendered application and the network availability and device capabilities of the end user.
If you are building an eCommerce website, you need to be aware of the reality that latency and performance will have a very real impact on the bottom line and revenue potential. Financial systems are also very time critical and extra unnecessary latency cannot be afforded.
But maybe this is a travel booking website where performance is not as critical because things inherently take time and the user is used to this because all websites from Expedia to Kayak and etc have a very lengthy checkout process relative to say purchasing a product on Amazon. You don’t want to build a system that is significantly slower than it could be, but it’s not a situation where milliseconds matter.
In conclusion, it’s true that in most situations the client won’t literally care if you use Go vs C# for your backend, or React vs Vue for your frontend. But every technology choice we make is an exercise in trade-offs and it’s critical that we make our decisions with the client’s best interest in mind and always try to balance the trade-offs in their favor. Nothing will ever be perfect, but we should do our best.
Clients do care deeply about the cons and negative trade offs of our technology choices, even if they can’t explain the technical specifics of why a certain outcome occurred. As freelancers and consultants, they trust us to make the right decisions on their behalf and it’s critical we do our best to not betray that trust.