Government website performance: how fast gov.pl sites are

Most of the Polish public institutions' websites that Chrome reports on are fast for real users. Of the 114 sites behind .gov.pl domains that Chrome has enough data on to assess on all three Core Web Vitals, 69% pass them on phones, against 62% of the 715 sites of a random sample of other popular .pl domains. The shared portal of the Polish government (opens in a new tab), where ministries and central offices have their sections, shows its main content in 1.4s. Government website performance splits in two: without the online services, public sites pass more often than the random sample, while the services citizens sign in to mostly fail, most of them because their layout shifts.

We took every .gov.pl domain in the Tranco list of popular sites (opens in a new tab), followed each to the site its homepage lands on, and looked that site up in the Chrome UX Report (opens in a new tab), Google's record of how fast pages load for real Chrome users. It is the same field data as our study of Core Web Vitals on Polish shops, and the same census that found which CMS public administration runs.

.gov.pl domains followed

189

sites with field data on phones

119

pass Core Web Vitals on phones

69%

LCP of www.gov.pl on phones

1.4s

Are Polish government websites fast?

A site passes Core Web Vitals when, at the 75th percentile of its visits, Largest Contentful Paint (LCP, when the main content appears) is at most 2.5 seconds, Interaction to Next Paint (INP, how fast it reacts to a tap) at most 200 ms and Cumulative Layout Shift (CLS, how much the layout jumps) at most 0.1 (web.dev (opens in a new tab)). On phones, 79 of the 114 public institutions' sites pass (69%, 95% interval 60% to 77%), against 446 of the 715 other .pl sites (62%, 59% to 66%): a little more often, though with this many sites the gap is within the margin of error. The clear difference is INP: 93% of public sites react to taps in time (87% to 96%), against 78% of the others (75% to 81%).

• gov.pl sites
• Other .pl sites

All three69% · 62%

LCP81% · 87%

INP93% · 78%

CLS86% · 84%

0%25%50%75%100%

Source: Chrome UX Report (Google); Outofplace census, 30 Oct 2026; Tranco list Y8YYGDownload CSVDownload PNG

Show the data

Item	gov.pl sites	Other .pl sites
All three	69%	62%
LCP	81%	87%
INP	93%	78%
CLS	86%	84%

The chart as an image, free to reuse under CC BY 4.0 with credit to Outofplace and a link to this page.

On LCP and CLS the two groups can't be told apart: 81% of public sites and 87% of other .pl sites have a good LCP, 86% and 84% a good CLS, both gaps within the margin of error. Among public sites, the online services are where layout shift is most common (below).

The average hides two kinds of site. Leaving out the online services we listed, 75 of the remaining 99 public sites pass on phones (76%, 66% to 83%), more than in the random .pl sample, though that sample wasn't stripped of its own online services. The services, as the section below shows, mostly fail.

How fast is gov.pl?

The gov.pl portal is where the government's ministries and many central offices publish: 25 of the domains in our sample, such as mf.gov.pl (finance), ms.gov.pl (justice) or men.gov.pl (education), redirect to a section of it. For CrUX they are one site, and on phones it shows its main content in 1.4s, reacts to taps in 117ms and barely shifts (CLS 0.00). On desktops its main content comes even sooner, in 1.1s, but there it fails: its layout shifts by 0.17, over the 0.1 threshold.

That makes one shared system responsible for the speed of a large part of the central administration: it passes on phones, and fixing its layout shift alone would make it pass on desktops too. CrUX also has page-level data for 21 of the portal pages these domains land on; they are in crux_pages.csv.

Which public institutions have the fastest websites?

The chart has the 20 most popular institutions' sites with field data, by their rank in the Tranco list; the table under it has all of them. Most sit in the green band, well under 2.5 seconds. Among the busiest, the fastest, the CEIDG business register, the National Health Fund (nfz.gov.pl), the tax portal (podatki.gov.pl) and a few others, are within a few tenths of a second of each other, too close to rank; praca.gov.pl, the public employment services' portal, is the only one of the 20 over 2.5 seconds.

ceidg.gov.pl0.7s

nfz.gov.pl0.8s

podatki.gov.pl1s

uokik.gov.pl1.1s

dziennikustaw.gov.pl1.1s

uodo.gov.pl1.2s

login.gov.pl1.2s

sejm.gov.pl1.4s

www.gov.pl (portal)1.4s

rpo.gov.pl1.5s

szukajwarchiwach.gov.pl1.7s

policja.gov.pl1.7s

ipn.gov.pl1.7s

geoportal.gov.pl1.7s

parp.gov.pl1.7s

biznes.gov.pl2.1s

pacjent.gov.pl2.1s

knf.gov.pl2.1s

lasy.gov.pl2.4s

praca.gov.pl3.3s

0s2s4s6s

Source: Chrome UX Report (Google); Tranco list Y8YYGDownload CSVDownload PNG

Show the data

Item	Value	n
ceidg.gov.pl	0.7s	1
nfz.gov.pl	0.8s	1
podatki.gov.pl	1s	1
uokik.gov.pl	1.1s	1
dziennikustaw.gov.pl	1.1s	1
uodo.gov.pl	1.2s	1
login.gov.pl	1.2s	1
sejm.gov.pl	1.4s	1
www.gov.pl (portal)	1.4s	27
rpo.gov.pl	1.5s	2
szukajwarchiwach.gov.pl	1.7s	1
policja.gov.pl	1.7s	1
ipn.gov.pl	1.7s	1
geoportal.gov.pl	1.7s	1
parp.gov.pl	1.7s	1
biznes.gov.pl	2.1s	1
pacjent.gov.pl	2.1s	1
knf.gov.pl	2.1s	1
lasy.gov.pl	2.4s	1
praca.gov.pl	3.3s	1

The chart as an image, free to reuse under CC BY 4.0 with credit to Outofplace and a link to this page.

A domain that leads elsewhere is listed under its own name: policja.gov.pl's figures are those of policja.pl, where it redirects.

Online services: where gov.pl falls behind

The services a citizen or a business signs in to, files with or pays through are a different story. Of the 18 we listed (tax, the national login, mObywatel, the patient account, ePUAP, the business portal, e-TOLL and others), 16 have field data on phones, and only 4 of the 15 we could assess pass (95% interval 11% to 52%, wide with so few). Of the 11 that fail, 10 shift their layout by more than 0.1. Tax (podatki.gov.pl), the patient account (pacjent.gov.pl), e-Doręczenia and podpis.gov.pl are the services that pass.

praca.gov.pl0.73

moj.gov.pl0.69

login.gov.pl0.50

ezamowienia.gov.pl0.50

mobywatel.gov.pl0.45

etoll.gov.pl0.42

puesc.gov.pl0.28

gabinet.gov.pl0.20

biznes.gov.pl0.16

e-konsulat.gov.pl0.11

pacjent.gov.pl0.09

podpis.gov.pl0.02

edoreczenia.gov.pl0.01

podatki.gov.pl0.00

ceidg.gov.pl0.00

e-sad.gov.pl0.00

0.000.200.400.600.80

Source: Chrome UX Report (Google)Download CSVDownload PNG

Show the data

Item	Value
praca.gov.pl	0.73
moj.gov.pl	0.69
login.gov.pl	0.50
ezamowienia.gov.pl	0.50
mobywatel.gov.pl	0.45
etoll.gov.pl	0.42
puesc.gov.pl	0.28
gabinet.gov.pl	0.20
biznes.gov.pl	0.16
e-konsulat.gov.pl	0.11
pacjent.gov.pl	0.09
podpis.gov.pl	0.02
edoreczenia.gov.pl	0.01
podatki.gov.pl	0.00
ceidg.gov.pl	0.00
e-sad.gov.pl	0.00

The chart as an image, free to reuse under CC BY 4.0 with credit to Outofplace and a link to this page.

Why are government websites slow?

Of the 35 public sites that fail on phones, 23 miss on LCP, 17 on CLS and 8 on INP (a site can miss more than one). With so few sites only one gap is clear: LCP is a more common reason than INP. CLS can't be told apart from either.

LCP too slow66%

INP too slow23%

CLS too high49%

Source: Chrome UX Report (Google)Download CSVDownload PNG

Show the data

Item	Value	95% interval	n
LCP too slow	66%	49%–79%	35
INP too slow	23%	12%–39%	35
CLS too high	49%	33%–64%	35

The chart as an image, free to reuse under CC BY 4.0 with credit to Outofplace and a link to this page.

Slow LCP often goes with a slow server. Of the 23 public sites whose LCP misses 2.5 seconds, 13 also have a time to first byte (TTFB) over 800 ms, the threshold web.dev calls good (opens in a new tab), against 15 of the 94 whose LCP is good. That is an association, not proof that the server is the cause, and the other sites with a slow LCP have a fast first byte. Several of the slowest wait over three seconds for the first byte alone: no image compression helps a page that hasn't started to arrive. For those, a CDN in front of the server or a cache for pages that are the same for every visitor is the first fix.

Phones vs desktops

Public sites pass as often on desktops as on phones: 68% against 69%. Other .pl sites gain on desktops, from 62% to 69%, which closes the gap between the two groups there. The main content still arrives later on a phone: for the median public site, phone LCP is 1.2 times the desktop one.

• Phones
• Desktops

www.gov.pl (portal)1.4s · 1.1s

sejm.gov.pl1.4s · 0.8s

uokik.gov.pl1.1s · 0.8s

nfz.gov.pl0.8s · 0.7s

uodo.gov.pl1.2s · 0.9s

podatki.gov.pl1s · 0.8s

praca.gov.pl3.3s · 2.3s

policja.gov.pl1.7s · 1.2s

pacjent.gov.pl2.1s · 1.9s

lasy.gov.pl2.4s · 2.1s

ceidg.gov.pl0.7s · 0.6s

parp.gov.pl1.7s · 1.3s

biznes.gov.pl2.1s · 1.5s

szukajwarchiwach.gov.pl1.7s · 1.4s

login.gov.pl1.2s · 1.1s

ipn.gov.pl1.7s · 1.4s

geoportal.gov.pl1.7s · 1.6s

dziennikustaw.gov.pl1.1s · 1.3s

rpo.gov.pl1.5s · 1s

knf.gov.pl2.1s · 1.4s

0s2s4s6s

Source: Chrome UX Report (Google)Download CSVDownload PNG

Show the data

Item	Phones	Desktops
www.gov.pl (portal)	1.4s	1.1s
sejm.gov.pl	1.4s	0.8s
uokik.gov.pl	1.1s	0.8s
nfz.gov.pl	0.8s	0.7s
uodo.gov.pl	1.2s	0.9s
podatki.gov.pl	1s	0.8s
praca.gov.pl	3.3s	2.3s
policja.gov.pl	1.7s	1.2s
pacjent.gov.pl	2.1s	1.9s
lasy.gov.pl	2.4s	2.1s
ceidg.gov.pl	0.7s	0.6s
parp.gov.pl	1.7s	1.3s
biznes.gov.pl	2.1s	1.5s
szukajwarchiwach.gov.pl	1.7s	1.4s
login.gov.pl	1.2s	1.1s
ipn.gov.pl	1.7s	1.4s
geoportal.gov.pl	1.7s	1.6s
dziennikustaw.gov.pl	1.1s	1.3s
rpo.gov.pl	1.5s	1s
knf.gov.pl	2.1s	1.4s

The chart as an image, free to reuse under CC BY 4.0 with credit to Outofplace and a link to this page.

Have gov.pl sites got faster in 2026?

The CrUX History API keeps each site's 28-day figures week by week. Across 110 public sites assessable in both January and October 2026, the share passing on phones went from 70% to 68%: no real change. 4 sites started passing and 6 stopped. Each site's weekly LCP is the sparkline in the table above.

How to check the speed of a government website

Anyone can read the same field data we used, for any site with enough Chrome traffic, without an account. Our guide explains why lab and field numbers differ.

1. Open PageSpeed Insights

   Paste the address into PageSpeed Insights (opens in a new tab). The top block, "Discover what your real users are experiencing", is the Chrome UX Report: 28 days of real Chrome visits, read at the 75th percentile. The lab score below it is one simulated load and often disagrees.

2. Switch between mobile and desktop

   The tabs at the top give each device its own verdict. Phones are the harder test.

3. Read the origin as well as the page

   Under the verdict, switch from "This URL" to "Origin". The origin covers every page of the site, the sign-in and the forms behind the homepage included: in our data, login.gov.pl and puesc.gov.pl shift far more across the whole site than on their front pages.

4. Find the metric that fails

   LCP over 2.5 seconds, INP over 200 milliseconds or CLS over 0.1. When LCP fails, look at TTFB in the same block: if the first byte alone takes most of the time, the fix is on the server, not in the page.

5. Follow it over time

   The CrUX History API (opens in a new tab) returns the same metrics week by week, up to 40 periods back, so a change shows whether a release helped.

How we measured

What this data can't tell you

CrUX reports only sites with enough Chrome traffic, so many small institutions, courts and offices drop out, and the sample leans to the busier ones: these are the public sites people use most, not every public site. Origin data covers every page of a site and every Chrome user who visited, not a particular page or person; Safari and Firefox users aren't in it. The comparison sample is drawn at random, not matched on traffic or purpose, and a shop and a ministry do different work. Differences here are between sites, not the effect of any one choice. The weekly history overlaps (each point is 28 days, a week apart), so neighbouring weeks aren't independent.

Further reading