What Merchants Should Ask Their Agency About Store Speed (Beyond a Green Lighthouse Score)
Your agency sends a performance report. The Lighthouse score is 92, highlighted in green. The homepage looks fine, and the latest optimization work is marked complete.
Yet customers still say product pages feel slow on their phones. Someone notices a delay when selecting variants. A promotional banner occasionally pushes Add to Cart down the screen.
So is the store actually fast?
A Lighthouse score helps, but it does not answer that on its own. It measures one page under simulated conditions. Shoppers move across pages and devices on a storefront that changes whenever you update a theme, install an app, or launch a campaign. You do not need to become a performance specialist. You do need to know what to ask, what evidence to expect, and how improvements stay in place. Put these questions into your next performance review.
1. Why Isn’t a Green Lighthouse Score Enough?
Ask your agency: “Which pages did you test, and does this score reflect how customers shop?”
Google Lighthouse scores performance from 0 to 100. Ninety or above is considered good, and it is a practical way to spot problems and check whether development work improved a page. The catch: Lighthouse measures one page at a time, under simulated conditions.
Your homepage might score 94 while a product page with reviews, video, and recommendations performs much worse. Even a fast-loading page can frustrate shoppers if the size selector lags or the cart drawer opens slowly. One score also says little about yesterday, or about next week’s theme and app changes.
A good agency should show the pages tested, explain why they matter, and name where performance still needs work.
What to request: a breakdown by page type, with mobile separate from desktop.
2. Which Store Pages Should Be Tested?
Ask your agency: “Are we measuring the pages responsible for sales, or mainly the homepage?”
Performance checks should follow the shopping journey. At minimum, agree on:
- Homepage: especially when ads or brand campaigns send traffic there
- Product detail pages (PDPs): a best seller, a multi-variant product, and one with reviews or video
- Collection pages: a busy category with images, filters, and sorting
- Cart or basket: add, quantity change, and cart drawer
- Campaign landing pages: seasonal promotions and paid social destinations
On Shopify, interactions such as opening a cart drawer often need hands-on or scripted testing because they lack a separate URL. Shopify also controls much of checkout, so separate what your agency can change from platform-managed behavior. Five to ten representative URLs are enough for a smaller store; expand where sales, traffic, or complexity justify it.
What to request: a written list of pages and interactions, reviewed when commercial priorities change.
3. What’s the Difference Between Lab Data and Real-User Data?
Ask your agency: “Are these results from a simulated test or actual shoppers?”
Lab data comes from a controlled test such as Lighthouse. It simulates a visit under defined conditions, which helps diagnose issues or compare a page before and after a change.
Field data comes from real visitors. Google’s Chrome User Experience Report (CrUX) collects eligible Chrome experiences over a rolling 28-day period. Google PageSpeed Insights can show both: real-user measurements when available, plus a Lighthouse lab run. Google Search Console reports Core Web Vitals from field data, and Shopify’s Web Performance reports add another shopper-side view.
A strong lab result does not guarantee the same experience on older phones or slower connections. When a page lacks enough observations, Google may also report field data for the whole site rather than that URL.
What to request: lab and field results clearly labeled, with a note when real-user data is missing for a page.
4. What Mobile Core Web Vitals Targets Should We Agree On?
Ask your agency: “Are we meeting Google’s recommended thresholds on mobile?”
| Metric | What it means for your store | Good target |
|---|---|---|
| Largest Contentful Paint (LCP) | How quickly the main visible content, often a product image, appears | 2.5 seconds or less |
| Interaction to Next Paint (INP) | How promptly the page responds to a tap on a variant, filter, or button | 200 milliseconds or less |
| Cumulative Layout Shift (CLS) | How much the page unexpectedly moves while someone views it | 0.1 or less |
These are Google’s field-data thresholds at the 75th percentile: at least 75% of measured visits should meet each target for the relevant page or group. Mobile deserves extra attention; a page that feels fine on an office laptop may behave differently on an older phone. Ask for mobile numbers first, and which metric needs work.
A Lighthouse score alone does not prove good INP. The standard page-load test does not reproduce a full shopper journey, so real-user interaction data or dedicated interaction tests are needed. Talk Commerce has covered Core Web Vitals and their optimization; Google’s responsiveness metric has since moved from FID to INP.
5. What Happens to Store Speed After a Theme Update or New App?
Ask your agency: “How will we know if this new feature slows the store down?”
Shopify makes it easy to add apps and theme customization. That helps merchants, but each addition deserves a performance check. A familiar sequence: a review app, a new tracking pixel, a personalization widget, a product-gallery redesign. Each change may have a clear commercial purpose; together they can add scripts, images, fonts, or layout shifts that slow the store.
The useful response is not to reject helpful features. It is to measure impact. For material changes, a solid process usually includes a baseline on the affected product or collection page, the same tests on the same URL after the change (comparable settings, preferably several runs), a check that key interactions still feel responsive, and a recorded outcome before deciding on further work.
A simple change log that links theme updates, app installs, and marketing scripts to those checks makes trade-offs visible. If an app is commercially useful but measurably slower, you can accept that deliberately.
6. Who Watches Store Speed After Launch?
Ask your agency: “Who will notice a regression, and what happens when one is detected?”
A performance audit is a snapshot. Products, campaign assets, apps, and merchandising content keep changing after the project ends, so monitoring belongs in the maintenance agreement.
Two activities complement each other: scheduled lab checks on agreed URLs, and real-user monitoring of whether shopper experience is improving or getting worse. Shopify’s Web Performance reports and CrUX support the second. For regular multi-page PageSpeed checks, agencies can also use a tool such as Apogee Watcher alongside their existing stack; the Shopify Core Web Vitals guide walks through how to set that up.
The tool matters less than the process. Agree on frequency (for example weekly checks, plus after material changes and before major campaigns), thresholds (one workable rule: investigate if repeated mobile loading tests deteriorate more than 20% against baseline), and ownership (who receives the alert, investigates, and opens a ticket). You want that arrangement working before Black Friday, not after campaign traffic arrives.
7. How Should Store Speed Be Reported to Leadership?
Ask your agency: “Can you show us what’s improved, what’s worse, and what we’re doing next?”
Retail managers rarely need dozens of technical charts. They need a clear view of whether the storefront is acceptable and where investment is required. A one-page monthly report can cover:
| Report item | What to show |
|---|---|
| Pages covered | Product, collection, cart, and campaign pages tested |
| Customer experience | Mobile field metrics, with unavailable data marked |
| Changes over time | Lab performance versus the previous baseline |
| Priorities | Top three issues, business relevance, proposed fixes |
| Next steps | Owner, next test date, outstanding actions |
Keep field and lab separate: a lab change today may take time to appear in real-user reports. Showing conversion or add-to-cart beside speed can add context, though a speed change alone does not prove why sales moved. Reporting should support a decision, not celebrate a green number.
8. Store Speed Checklist: Questions to Send Your Agency
Copy this into your next agency brief, maintenance review, or campaign planning meeting.
- Page coverage: Which homepage, product, collection, cart, and campaign pages are included? Can we have the URL list in writing?
- Measurement: Will reports distinguish lab from real-user data and mark where field data is unavailable?
- Mobile targets: Are we checking LCP at 2.5 seconds, INP at 200 milliseconds, and CLS at 0.1 or better with the right real-user measurements?
- Changes: Will major theme changes, app installs, and marketing scripts trigger before-and-after checks?
- Monitoring: How often are priority pages checked after launch, and who reviews the results?
- Escalation: What deterioration opens an investigation or ticket, especially before important campaigns?
- Reporting: Can we get a monthly summary with trends, the top three actions, owners, and the next review date?
You do not need a perfect score on every page. You need agreed priorities, a reliable way to measure them, and someone responsible when performance deteriorates.
Start with one step: ask for the mobile results on your three highest-selling product pages, including real-user data where available, and compare them with the latest homepage screenshot. That conversation usually beats asking whether the Lighthouse score is still green.
About the Author
Markos Giannopoulos is the co-founder of Apogee Information Systems and Apogee Watcher, a PageSpeed monitoring tool for agencies and freelancers who manage multiple storefronts.