Beyond the Score: The Real Role of PageSpeed Insights in Website Optimization

PageSpeed Insights and Website Optimization: A Great Tool, But Not the Whole Story

When it comes to website optimization, speed matters. A fast website not only creates a better user experience but can also improve search rankings and conversions. Google’s PageSpeed Insights (PSI) has become one of the most popular tools for measuring site performance, and for good reason. It gives developers and business owners a simple score and actionable feedback for improvement.

However, like any tool, it has its limits. Understanding where PageSpeed Insights shines and where it falls short can help you use it more effectively as part of a broader strategy for building high-performing and effective websites.

The Pros: Why PageSpeed Insights Is Valuable

1. Free, Easy to Use, and Widely Recognized

   PageSpeed Insights is completely free to use, and is as simple as entering a URL and waiting a couple of seconds for the results. Furthermore, nearly everyone in the web development industry recognizes PageSpeed Insights scores and has at least some understanding of what they mean.

2. Clear, Actionable Feedback

   In addition to scoring your page's performance, PageSpeed Insights also provides insights and detailed suggestions for improvement. Recommendations are further categorized for easy reference when working on particular aspects of performance. Some examples might be optimizing images, reducing unused JavaScript, enabling text compression, and prefetching origins.

3. Consistency and Benchmarking

   Because PageSpeed Insights uses the same underlying technology (Google’s Lighthouse) for every site, it provides a consistent basis for comparing performance across pages or tracking progress over time. This makes it an excellent benchmarking tool during development or after site updates. With its clear and consistent scoring, PageSpeed Insights makes it easy to set and work towards specific targets before, during, and after the website is created.

4. Mobile vs Desktop Insights

   With mobile-first indexing now standard, the ability to see mobile-specific metrics is essential for modern websites. PageSpeed Insights provides separate scores for desktop and mobile devices, including sub-scores and recommendations for each of the components that make up the overall score for both desktop and mobile. This makes it easier to optimize for your audience and goals while promoting better search engine rankings related to mobile-first indexing.

5. Connection to Core Web Vitals

   PageSpeed Insights ties directly into Google’s Core Web Vitals, a set of metrics that measure user experience factors like loading speed, interactivity, and visual stability. These metrics have a real impact on search visibility and user satisfaction, so understanding them is critical for SEO and usability.

The Cons: The Hidden Challenges Behind the Scores

Despite its usefulness, PageSpeed Insights isn’t a perfect reflection of real-world website performance or effectiveness.

1. What’s a “Good” Score, Really?

   A PageSpeed score of 80 might sound mediocre, but in PageSpeed Insights terms that’s often considered very good. Many sites that score between 50–70 still perform beautifully for users. In fact, some of the biggest names in tech have scores in ranges that are often considered not good. The key is to interpret the results within context rather than chasing a perfect 100.

   For some context, while writing this blog post I tested several well-known websites that I assumed would be highly optimized to see how they performed on PageSpeed Insights. The results?

   • https://pagespeed.web.dev/ - Mobile: 82, Desktop: 95
   • https://www.amazon.com/ - Mobile: 60, Desktop: 93
   • https://www.apple.com/ - Mobile: 56, Desktop: 83
   • https://cloud.google.com/docs/ - Mobile: 68, Desktop: 62
   • https://www.microsoft.com/en-us/ - Mobile: 32, Desktop: 23

   Of course, these scores may change at any moment and will be different from one test to another (as discussed later), so feel free to run your own tests and see what kind of results you get.

2. “Lab” Data vs. “Real-World” Data

   PageSpeed Insights reports two sets of metrics: Lab data (simulated tests) and Field data (from real users via Chrome UX Report). Lab data is useful for diagnosing issues and for benchmarking a new site, but it doesn’t always match how real visitors experience your site.

   On the other hand, field data accurately measures and reports on how real visitors have experienced your site over the past few months, but also has a few drawbacks:

   • There is a large gap between when you make a change to your site and when you can measure its impact on performance, which makes it practically impossible to use during development.
   • It only measures a small portion of your site’s visitors and in many cases, this is not even enough for PageSpeed Insights to display any field data for your site at all.
   • The field data will be skewed towards your user base, and while that makes it tremendously useful for many purposes, it also makes it very difficult to set benchmarks and maintain consistency between different types of sites with different audiences (e.g., global vs local audiences, users on different networks and with differing bandwidths, network reliabilities, devices, etc.).

3. Inconsistent Results

   Even when you do not change anything on your site or your server, if you run PageSpeed Insights multiple times on the same page you may notice dramatically different scores. It is not uncommon to see a 10, 20, or even 30 point difference in scores simply by waiting 5 minutes and running the same test again. There is usually no indication as to why it varies so wildly, but some factors may include variations in server load for the webserver, variation in server load for the testing server, different networking conditions, backend routing in Google’s infrastructure, disproportionate impact of third-party services such as CDNs and APIs, request caching or queuing, or any number of other determinants. This is such a common issue that at Marketpath, we typically run each test multiple times and aggregate the scores when benchmarking performance.

4. Impractical (or Impossible) Recommendations

   Some PageSpeed Insights recommendations are difficult to implement, particularly for dynamic, media-rich, or e-commerce sites. Compressing images, minimizing fonts, and reducing JavaScript might improve the score—but may also reduce visual appeal, functionality, or branding consistency. In some cases, over-optimizing can actually hurt user experience and, as a result, reduce the effectiveness of your site.

5. Limited Context

   PageSpeed Insights evaluates the technical performance of one page under specific conditions. It doesn’t consider broader aspects of site effectiveness, such as:

   • Sites with high-quality images and videos that support branding.
   • Visuals that intentionally load slower for impact.
   • Performance differences between slow and fast internet environments.
   • Site-wide optimizations like caching or lazy loading that benefit subsequent pages.
   • Shared content delivery (CDN assets used across multiple sites).

   A perfect PageSpeed Insights score doesn’t necessarily mean your website performs well for your audience or meets your business goals.

Why You Shouldn’t Chase a Perfect Score

Trying to hit 100 on PageSpeed Insights is unrealistic and may actually hinder the effectiveness of your site. You might:

• Unintentionally reduce image quality or usability to the point of hurting your brand.
• Remove useful features because they increase load time.
• Overcomplicate your site and make it difficult to maintain.
• Spend hours fixing issues that shave off milliseconds with no real user benefit.

Instead, focus on specific target metrics, such as these ones recommended by Google:

• Time to First Byte (TTFB) under 0.8s.
• First Contentful Paint (FCP) under 1.8s.
• Largest Contentful Paint (LCP) under 2.5s.
• Cumulative Layout Shift (CLS) under 0.1.
• Interaction to Next Paint (INP) under 200ms.

Keep in mind that while these target performance metrics sound good, they still may not be practical for your website. They also do not fully represent the effectiveness of your website compared to other key business metrics, such as:

• Number of first-time visitors.
• Number of repeat visitors.
• Average time interacting with the site.
• Average number of pages visited.
• Percent of visitors who "convert" into a qualified lead.
• Percent of leads who "convert" into customers.
• Total number of form submissions or "conversion events".
• New business that is attributable to your website.

These are the metrics that truly matter and that deserve your focused time and attention.

Beyond PageSpeed Insights: The Bigger Picture of Website Effectiveness

Performance is just one piece of the puzzle. An effective website also depends on:

• SEO & Discoverability: Can people find your site easily?
• GEO (Generative Engine Optimization): Is your content discoverable and usable by machines to improve your visibilty in AI-driven search experiences?
• Accessibility: Can everyone use your site, regardless of ability?
• User Experience: Is the journey intuitive and enjoyable?
• Content Strategy: Does the content clearly communicate your message and convert visitors?
• Integrations and Compliance: Does your website meet all industry, legal, and technical requirements, including integration with other technologies?

Ultimately, the goal isn’t just to build a fast site. It’s to build one that connects with your audience and delivers business results.

A Smarter Approach to Website Optimization

No single tool can tell the whole story. In fact, even a perfectly optimized site may fail to achieve its business objectives if it is optimized for the wrong goals. To get a complete picture of your performance, combine PageSpeed Insights with other free tools such as:

• WebPageTest for detailed performance breakdowns.
• GTmetrix for comparative insights.
• Lighthouse audits for accessibility and SEO checks.
• Real-user monitoring tools (see below for a sampling) for live experience data.

Additionally, consider the following free tools to better understand your website’s effectiveness that goes beyond technical performance:

• Google Search Console and Bing Webmaster Tools for measuring your site’s actual performance and visibility in search engines.
• Google Analytics for visitor metrics, including tracking user journeys across your site and measuring specific events such as conversions. There are a number of free and paid alternatives, including Matomo, Plausible, and Fathom Analytics to name a few.
• Google Keyword Planner, WordStream, Ubersuggest, and AnswerThePublic for keyword analysis and content ideas.
• LogRocket, Hotjar, Crazy Egg, Microsoft Clarity, Mouseflow, Smartlook, or many others for logging, heatmapping, and user journey analysis.
• ScreamingFrog for website crawling and issues checker.
• Synup Listing Scan and BrightLocal for local SEO information.
• Semrush, Ahrefs, and SEO Review Tools for a variety of tools under a single location (keyword research, backlink checking, crawling, and more).

Most importantly, apply human expertise—experience, strategy, and judgment—to interpret the data in context.

At Marketpath, we’ve spent more than 20 years building and optimizing websites that score well in lab tests AND perform well for real users AND achieve real business goals. Whether you need to improve load times, increase engagement, or boost conversions, we can help you balance performance, usability, and effectiveness.

Final Thoughts

PageSpeed Insights is an excellent starting point for measuring and improving site performance but it’s not the final word. Use it as a guide, not a goalpost. Focus on delivering a site that is effective at attracting, engaging, and converting your audience, and that ultimately drives results for your business.

If you’d like to discuss how to make your website faster and more effective, contact Marketpath today. We’ll help you build a site that performs beautifully in both lab tests and real life.