Five scores decide how your website performs
Google Lighthouse is free. It runs inside Chrome. It grades any public website on five categories, and it takes about thirty seconds.
Four of those categories have shaped ecommerce for years. The fifth arrived this year, and it measures whether an AI shopping agent can read your site at all.
Here is how a site on the Essentials platform from Blueport scores.
August Haven, a Wisconsin furniture and design retailer, running on Blueport Essentials. Median of three homepage runs, mobile, default throttling, October 2026.
The Lighthouse report, straight out of Chrome.
What each score actually means
Most retailers have never seen their own numbers, because the tool sits in a developer menu and nobody translated the output into something a merchant cares about. Here is the translation.
Performance
How fast the site feels to a shopper on a phone. It tracks directly to bounce rate and conversion.
SEO
Whether search engines can crawl, read, and rank your pages. Your organic traffic lives here.
Accessibility
An automated check against WCAG criteria, which is where ADA exposure begins. No automated tool certifies compliance, but a low score tells you where to look.
Best Practices
The quality of the code underneath. It is the foundation the other three sit on.
Agentic Browsing
New this year. Whether an AI shopping agent can read your pages, follow their structure, and act on what it finds. Shoppers have started asking AI where to buy. This is the check that decides whether your store is in the answer.
What the gap looks like
Here is the same scorecard against a Top 100 furniture retailer running on a general-purpose ecommerce platform. Both are public sites. Anyone can run the test.
| Category | Blueport site | Top 100 retailer |
|---|---|---|
| Performance | 95 | 41 |
| Accessibility | 95 | 86 |
| Best Practices | 100 | 73 |
| SEO | 100 | 77 |
| Agentic browsing | Passed every applicable check | Did not |
Medians taken October 2026, mobile, default throttling. Three homepage runs for the Blueport site, five for the Top 100 retailer. The Blueport site ranged from 93 to 98 on performance. The Top 100 retailer ranged from 37 to 54. Its best run still landed 39 points below the Blueport site's worst.
The same report for the Top 100 retailer.
The two numbers underneath
A category score is a summary. The metrics behind it are where the gap actually lives, and two of them carry most of it.
Total Blocking Time
The stretch where the browser is too busy running JavaScript to respond to a tap.
Cumulative Layout Shift
How much the page moves while it finishes assembling. Google treats anything above 0.25 as poor.
Layout shift does double duty. It is a performance metric, and it is also one of the agentic browsing checks, because an agent that clicks a target which then moves has clicked the wrong thing. One flaw, counted in two categories.
Where the page gets built
Every ecommerce page gets assembled somewhere. Either the server sends a finished page, or it sends a mostly empty shell plus instructions and the shopper's phone does the assembly.
Many ecommerce builds lean on the second approach, especially headless storefronts and sites loaded with apps and scripts. It is faster to build that way. The cost lands on the device. Assembling a furniture category page with large images, pricing rules, availability by location, and financing offers is real work for a mid-range phone on store wifi.
It also lands on anything visiting the page that is not a person with a browser. An AI agent reading a shell and a pile of instructions has to execute all of it to learn what the page says. Often it does not, and the page looks empty.
Blueport Essentials pairs server-side rendering with traditional client-side rendering, choosing per page type. Catalog and product pages, the ones that need to be fast and legible to search and to agents, get rendered on the server. Interactive pieces that genuinely need to run on the device keep running there.
What agentic browsing tests
It is not a single score out of a hundred. It is a set of pass-or-fail checks, reported as a ratio, and the checks are specific.
Blueport has registered for Chrome's WebMCP origin trial
WebMCP is the program that lets a site declare what an agent can actually do on it. Search the catalog. Check availability. Start a quote. The standard is still being written, and participating while it is being written is the difference between inheriting it and catching up to it.
None of this is a roadmap item
Server-side rendering is live today. The agentic checks pass today. WebMCP registration is complete. The rollout to additional clients runs over the coming weeks, and retailers onboarding now receive all of it without asking for it.
Every Blueport Essentials client gets this the same way they get every platform update. New features and improvements ship on a two-week cycle, and they are included in the Essentials package.
No replatforming. No agency project. No separate line item.
The agentic checks Google added this year did not exist last year. The ones it adds next year are not written yet. Blueport Essentials clients get those as they land, included. Retailers on other platforms often buy them app by app, pay for a custom project, or hire someone to build them.
See your own scores
Send us your site. We will run the same five-category audit and send back your results, with notes on what is driving each number and what it would take to move it.
Request your scorecard