Results
Kozelj.net: Google's speed score went from 41 to 94 out of 100
Kozelj is a Slovenian online store we upgraded and now maintain. Both versions are online, so you don't have to take our word for it: the old store still runs at old.kozelj.net, the new store at kozelj.net. We measured both with Google's own test (PageSpeed Insights, Lighthouse on a phone, several runs in October 2026). Google scores every page from 0 to 100: 90 or more is "good", 50 to 89 "needs improvement", and below 50 is "poor". The old store was in the poor band. The new one is in the good band on every score.
| Google score (0 to 100) | Old store old.kozelj.net | New store kozelj.net |
|---|---|---|
| Performance | 41 | 94 |
| Accessibility | 79 | 100 |
| Best practices | 88 | 100 |
| SEO | 50 | 100 |
| What a shopper on a phone waits for | Old store | New store |
|---|---|---|
| First thing appears on screen | 2.3 – 3.5 s | 1.4 – 2.3 s |
| Main content has loaded | 14.9 – 23.0 s | 1.8 – 2.9 s |
| Page looks complete | 9.5 – 11.8 s | 2.3 – 4.6 s |
| Page frozen, not reacting to taps | 0.33 – 0.76 s | 0.01 s |
In plain words: on the old store a shopper on a phone waits 15 to 23 seconds for the main content. On the new one it is there in about 2 seconds.
Run the test yourself: old store on PageSpeed Insights · new store on PageSpeed Insights.
Scores are the median of our runs in October 2026, timings the range we saw; Google's test varies by a few points from run to run, so your result will differ slightly. Before the new store went live we measured it at 99 on a staging copy; the live store, with its real traffic and third-party scripts, scores 90 to 99.
What can speed be worth? An example
The best-known study on this, by Deloitte for Google (37 brands, Europe and the US), found that making a retail site's mobile pages just 0.1 seconds faster brought 8.4% more orders on average.
Kozelj's main content now loads more than 12 seconds faster. But take only that smallest measured step, 0.1 seconds, and keep the order size the same. Here is what 8.4% more orders means for stores of different sizes. These are examples, not Kozelj's sales.
| If your store sells a year | With 8.4% more orders | Extra a year |
|---|---|---|
| $400,000 | $433,600 | +$33,600 |
| $500,000 | $542,000 | +$42,000 |
| $750,000 | $813,000 | +$63,000 |
| $1,000,000 | $1,084,000 | +$84,000 |
| $2,000,000 | $2,168,000 | +$168,000 |
The same study also saw people spend more per order, which this example leaves out.
This is an illustration based on published research, not a promise or a prediction for your store, and not Kozelj's own figures. What you actually earn depends on your traffic, prices, products and marketing. Source: Deloitte, "Milliseconds Make Millions" (2020).
For comparison, an upgrade for a store of these sizes costs from $4,500 once and from $395 a month after that, with the store on our server. Full pricing by store size.
More work
Further case studies are being published with client permission. In the meantime, reference work: nold.si, halopinki.si, kozelj.net — 400+ projects since 2005 through Limonet.
If you'd like to speak to a reference before hiring us, ask on the call and we'll arrange it.