When AI Suggests a Plugin: An Evaluation Checklist

TL;DR AI is optimizing for what is visible. Professionals optimize for what is sustainable. Update reliability, conflict history, security posture, and ecosystem stability indicate survivability. For long-term website health – especially in small businesses with limited resources – reliability & security are the metrics that matter most. A plugin that works today is useful. A plugin that remains stable across years of updates, integrations, and platform shifts is an asset.

A client recently sent me a request to install a plugin that Claude had suggested to her. ChatGPT, Claude and other chatbots, and LLMs are increasingly valuable to me and to my clients. They research, evaluate and recommend solutions in an increasingly valuable way. So here’s how to evaluate if the suggestion from your AI of choice, is a good idea or not.

TL;DR: LLMs are great for suggestions, but like medical advice, if it matters, run it by a professional you trust before installing anything onto your website.

First let’s define a plugin: this is an add-on that you install (usually with one click) in the WordPress interface. They have all sorts of benefits but primarily they add function to your existing website. We add plugins because we need something that isn’t there. Or fix something that is there.

For the above, reason, plugins should be used with caution. It can become, much like a health issue, a prescribing cascade. You take a pill to solve problem A, but it creates problem B. So you take another pill to solve problem B. It’s important that you understand, anything you add can leave life-long ramifications in your website. We commonly clean out databases that have plugin remnants from decades ago. This is called ‘technical debt’ and it grows over time – causing more costly maintenance and repairs. And for each debt you add, it gets increasingly more difficult to troubleshoot issues – which pill was it that created the symptom or was it the original disease?

Understanding how chatbot recommendations are generated is the first step toward evaluating them responsibly.

AI Bots will Recommend Solutions Based on Specific Criteria

As we’ve discussed, LLM’s (Chat bots like ChatGPT and Claude) are at the fundamental level pattern-recognition software. But they also are synthesizers of information. They do not just copy/paste – that would be stealing. They take 1000’s or millions of datasets on that one word, and then come up with something that would make sense.

Knowing this we can see that ChatGPT’s will base their recommendations on the stuff that they have learned (right or wrongly). And the data they are trained on, is public data, documentation, tutorials, and widely discussed tools. This leads them to recommend more often the visible tools instead of the best tools. Chat bots will present plugins with the following signals.

Plugins in Feature Lists

Chatbots heavily weight stated features. If a plugin advertises many features clearly, it is more likely to be recommended.

This does not mean those features are:

• Stable
• Efficient
• Well maintained
• Compatible with real-world websites

A long feature list increases visibility, not necessarily reliability.

Popular Plugins

AI models see mentions in blogs, mentions in tutorials, the base of documentation, github stars and reviews, the install counts listed on WordPress.org. For an AI, popularity must mean that it is high quality. (In a lot of cases, they are right!)

However, popularity can ALSO mean that it:

• is an early market entry
• it has a great effective and aggressive marketing strategy
• or it simply is the legacy choice.

Popularity does not necessarily mean that it is technically sound.

Documentation Visibility

Plugins with strong documentation – the knowledgebases, forums and YouTube videos – appear more frequently in AI chats because they are easy for the AI to reference, they are structured in their behind-the-scenes coding, they are often cited in other tutorials and blogs.

This creates a feedback loop:

Well-documented plugins are recommended more → recommended plugins get more documentation → cycle reinforces the AI recommendations.

Documentation quality is valuable, but it is not the same as a technically sound plugin.

Marketing Footprint

Plugins that maintain other marketing assets – similar to documentation – will remain the most visible to AI. So those produced by bigger companies that are backed by huge websites, active blogs and YouTube influencers will be mentioned more often than not. These plugins will have a disproportionate presence in AI training databases.

This makes them more likely to be recommended, even when other plugins might be the better choice.

What AI Does NOT Optimize For

Experienced consultants and developers evaluate plugins using a rule of thumb gathered over time, from failure. This is entirely different than marketing footprint! And AI systems will rarely see these characteristics. When we are evaluating a plugin, these are the criteria that we look for.

Update Reliability

We are looking for updates that are on a regular schedule and not only reactive. We’re looking for wide testing and reliability in the deployment. A plugin that updates frequently is not automatically safer. A plugin that updates predictably and responsibly is definitely a better choice.

Long-Term Support Consistency

Look for the owner, and their participation in the plugin forms in WordPress. Look for their updates in response to PHP versions, that the same owner has had it for the last few years, at least. And how long the plugin has been in the market and has been handled through changes in the technology.

Historical durability is one of the strongest indicators of a good choice.

Conflict History with Other Plugins

Your website runs with other things going on – and it is important that the new feature or plugin can fit with the mix. It is important that it doesn’t conflict with the other important features of your website. Always look at how the plugin is deployed, where it is enqueued, if it is duplicating other feature and if it will overwrite essential functions you need. The likelihood of conflicts skyrockets the higher the number of plugins you have. This is why we recommend and keep our clients at around 15.

A plugin that works perfectly in a test site, may actually cause issues in a mature live environment.

Security Over Time

Of course security is a big concern for every website owner. But as our clients know – even the best maintenance doesn’t mean that it is “never having issues.” It does mean, that you have a plan and consistently and responsibly deploy fixes. So always look at the pattern of security releases. Check the security databases for issues – and most importantly – how quickly they were handled.

Website Stability

A plugin does not exist in isolation. It exists inside your full website environment that includes WordPress updates, web hosting environments, theme and style frameworks, and the other third-party plugins. When adding a potential plugin always look for what risks you are bringing into the environment. Will it work with existing caching systems, hosting, security and themes? Also important is to know what the vendor costs will be moving forward, is the plugin in the open source community should you ever need to adjust it. Is the plugin interfacing with another API or service – is there a cost to that? is there reliability and continuity of that service/ API.

To wrap it up – here is a quick checklist to ask yourself, after you’ve received recommendations from your favorite AI chat bot.

New Plugin Checklist

Maintenance Questions

• When was the last meaningful update?
• How stable are major version transitions?
• Are fixes incremental or disruptive?

Compatibility Questions

• Does it duplicate features already in my stack?
• Does it introduce global scripts or heavy assets?
• Has it historically conflicted with caching or SEO tools?

Support Questions

• Is the developer responsive over multiple years?
• Are support threads unresolved or consistently addressed?
• Is documentation updated alongside releases?

Risk Questions

• What happens if development slows or stops?
• Can the functionality be replaced easily?
• Does it create database lock-in or proprietary data structures?

---

Adding a AI-Recommended Plugin to WordPress

The most effective workflow is to use these quick and helpful AI suggestions to bring you the most popular suggestions, then evaluate like so.

1. Use AI to generate initial plugin options
2. Narrow to 2–3 candidates
3. Evaluate using checklist above
4. Test in a copy of your website environment
5. Monitor performance and conflicts post-installation

And if you’re nervous about any suggestions you see in AI – we are here to help. Start with a consultation with Cathy and bring us alongside your favorite AI tools to keep your site functioning well for many years to come.