Speed is a revenue metric, conditionally
The "100ms costs you 1% of revenue" statistic gets quoted far past what the underlying research supports. Speed matters enormously between 6 seconds and 2.5 seconds, and much less between 2.5 and 1.8.
Which is to say: if your LCP is 5s+, this is probably where your money works hardest. If it is already 2.2s, spend the money on merchandising instead.
What worked
Four things produced the clearest results across our client base. First, killing render-blocking third-party scripts — chat widgets, review platforms and heatmaps are the usual offenders, and deferring them is nearly free.
Second, properly sized and modern-format hero imagery. Not "compress the images" — a real responsive picture element with an appropriately sized mobile crop.
Third, self-hosting fonts with a sane fallback. Fourth, and biggest on WooCommerce specifically, auditing plugins that load their assets on every page rather than the one page they serve.
What did not
Aggressive JavaScript bundle splitting on Shopify themes produced almost nothing measurable for the effort involved. Neither did chasing a CLS score below 0.05 once we were already under 0.1.
Server-side rendering a store that was already fast made the codebase harder to maintain and the numbers no better.
A realistic sequence
Measure with field data, not lab data. Lighthouse on your laptop is a development tool, not a diagnosis.
Then: third-party scripts, images, fonts, plugins, in that order. Stop when the field LCP is under 2.5s on mobile and go do something that grows revenue instead.
If any of this is a live problem rather than an interesting read, say so. We would rather have a specific conversation than a general one.
— Sayan Saha, Founder
Keep reading.
More from the same team.
Shopify vs WooCommerce in 2026: an honest comparison from a team that builds both
We ship on both platforms every month. This is the framework we use to pick the one a client can still run in eighteen months.
Before you commission a redesign, run these four weeks of research
Four weeks of research costs a fraction of the wrong redesign. It either justifies the spend or saves you the whole budget.
The MVP question nobody asks: what happens if we remove this?
One question shrank a 31-feature list to nine and put the product in front of a paying customer in fourteen weeks.
Talk to a senior lead.
Send the specifics — platform, traffic, what is stuck — and we reply within one business day.