Two Months of EF-Map Usage Data, and How EO-Map Compares

Back at the end of July I wrote up eleven months of EF-Map's usage counters. That one was mostly the long view, routes and page loads going back to August 2025. Since then the stats have got a lot richer. On 3 August I started recording what happens after a panel is opened, not just that it was opened, and the performance counters that went in at the end of July now have two months behind them. EO-Map, the EVE Online map I cloned from EF-Map, has also been live long enough to compare against. So this is two months of all of that, plus Google Analytics and Search Console, for both sites.

A quick note on what's being counted. EF-Map's own stats are anonymous counters with nothing identifying attached. A panel was opened, a frame was drawn on time, that kind of thing. Google Analytics runs alongside in statistical-only mode, and you can switch it off in Display Settings. The windows differ depending on what started when, so I've said which one each number uses. Panel and engagement numbers run from 3 August to 4 October. Anything comparing the two sites' own counters runs from 30 August to 4 October, minus five days I'll come to in a second. Search Console lags a couple of days, so it stops on 2 October.

A few things I've had to work around. EF-Map's page-load counter has been running a bit below its first-rendered-frame counter since mid-August, and well below it since mid-September, which shouldn't be possible and looks like a variation on the counting bug I fixed in August. So I've used rendered frames as the measure of a map load. Google Analytics also recorded 343 "users" from Singapore who spent a combined 34 minutes on the site, about six seconds each, 205 of them on two days in August. Whatever that is, it isn't people using the map, so I've taken Singapore out of every EF-Map Google Analytics figure below.

EO-Map has its own oddities. Its counters went live on 29 August, and most of that first day looks like my own video capture tooling running against the site. Then from 23 to 27 September its loads jump, and they don't look like the rest of its traffic. Most of them come from one screen size and one CPU class, under half get any input at all, and resource-loading errors run at several times the normal rate. Google Analytics shows a real surge of Russian, Ukrainian and Belarusian visitors on some of those days, so at least some of it is people, but I can't tell how much. So I've left 29 August and 23 to 27 September out of both sites' comparison numbers, which leaves 31 days.

How busy is it?

Over those 31 days EF-Map's map rendered 3,490 times. That's about 113 a day, and it never dropped below 64 or went above 258. EO-Map rendered 2,928 times, about 94 a day. So EF-Map is still doing more map loads, but not by much against a site that's less than seven weeks old. The shape is completely different though. EF-Map is flat. EO-Map goes from 20 loads on its quietest day to 459 on its busiest.

Google Analytics turns that round. In September it counted 654 people on EF-Map and 1,304 on EO-Map, so EO-Map reached about twice as many people. EF-Map's 654 had 1,903 sessions between them, nearly three each, and EO-Map's 1,304 had 2,416, under two each. EF-Map has a smaller crowd that keeps coming back. EO-Map gets more people, and most of them come once.

As for how EF-Map is doing generally, it's quieter than it was. Those 654 people in September compare with 1,337 in March and 1,083 in July, so it's roughly half what it was in March, and about level with May. Search hasn't moved much either way. Google sent 516 clicks in the first 30 days of this window and 518 in the last 30.

The game's calendar does show up. The week Cycle 7 launched, starting 28 September, EF-Map did 1,029 map loads against 812 the week before, and launch day, 29 September, was its busiest day since the end of August. There was also a lift in the week of 14 September, 875 loads against 587 the week before. That lines up with the Rooks and Kings EVE Frontier video on the secret necropolis going up on 15 September, but Google Analytics shows next to no YouTube referrals that week, so I can't tell you whether that's the reason. EO-Map's two biggest days in the comparison, 13 and 22 September, both came with a burst of Reddit referrals.

Where people come from

Most traffic to both sites is direct, which in practice means bookmarks, typed addresses, Discord and anything else that doesn't pass on where you came from. After that the two sites split completely.

Where sessions came from

Share of sessions in Google Analytics. EF-Map 3 August to 4 October with Singapore excluded, EO-Map 21 August to 4 October.

After direct traffic, EF-Map is a search site. Almost a third of its sessions come from Google, Bing and DuckDuckGo. The biggest single referring site is EVE Datacore, 357 sessions from just 24 people, so a handful of regulars come in that way. EO-Map is a social site. Reddit and YouTube between them brought in over a fifth of its sessions, and its first Reddit spike on 26 August landed people straight on the Market Supply panel. AI assistants are a rounding error for both so far. ChatGPT sent EF-Map 16 sessions in two months and EO-Map 47, though those 47 came from 11 people.

Google search, both sites

From Search Console. EF-Map 3 August to 2 October, EO-Map 18 August to 2 October.

Search is where EF-Map's head start shows. A year of articles and tool pages means it turned up 47,341 times in Google over two months, and around half of its page impressions were blog articles. After the front page, the most clicked page on the whole site is the Blueprint Calculator landing page, 59 clicks, then the killboard with 49. EO-Map is a month and a half old. Where Google shows the search term, 24 of its 36 clicks were people searching its own name, and 115 of its 135 clicks landed on the front page. It is growing though, 70 clicks in its last two weeks against 36 in its first two. One I found interesting is that EF-Map shows up for "eve map" 1,486 times and "eve online map" 810 times, and gets clicked on 17 and 7 of them. People looking for an EVE Online map are being shown the EVE Frontier one, and the EVE Online map page on EF-Map, which points them at EO-Map, picked up 20 clicks of its own.

What people open, and what they actually stay in

This is the bit I was most curious about. Since 3 August EF-Map has recorded 11,204 panel opens. 72% of those were someone deliberately clicking a panel open. The rest were panels opened as part of something else, the route list appearing when you calculate a route, or a shared link opening a panel for you. But opening a panel only tells you someone was curious. So each panel now also records roughly how long it stayed open, in four bands, under 10 seconds, up to a minute, up to five minutes and over five.

Two caveats on the timing. A panel that was open when you last left the site comes back open, and that visit gets timed too, so some panels have more timed visits than opens. And time open isn't time spent looking at it, someone can leave routing open on a second monitor all evening. So read this as a rough measure of whether a panel holds people, not a stopwatch.

The most deliberately opened thing on EF-Map is the Tools Library, 1,731 deliberate opens, and it's also the one people leave fastest, 64% of visits over in under 10 seconds. It's a list you pick tools from, so a short visit isn't a bad sign on its own, but EF-Map doesn't count what gets picked, so I can't tell you how many of those people found what they wanted. EO-Map does count it, and since it started on 13 September, leaving out the odd days, it logged 213 tool picks against 355 Tools Library opens.

Event Search sits at the top of the chart, but treat it with care. Its 215 timed visits come from only 5 counted opens, so almost all of them are the panel coming back open from a previous visit, or opening some way the counter doesn't see, and I can't tell which. The interesting ones are the quiet panels that hold people. Gate Activity was only opened 86 times in two months, but half its visits went past a minute and 22% past five. System Finder, 163 opens, same story, 49% past a minute. And the Blueprint Calculator keeps a quarter of its visits past five minutes, from 297 opens, which is one of the highest five-minute shares on the site. At the other end, Smart Assemblies gets opened a lot, 732 times, but only 21% of visits go past a minute. The killboard sits in between, 1,225 opens and 36% past a minute.

Routing is the busiest panel that also holds people, 44% of visits past a minute, but it's where you can see people drop off as well. Of 3,217 times the routing panel was open, 2,380 had a start system filled in, only 629 got a destination, 371 got as far as a calculated route and 33 had the route copied out. Most people open routing, have a look, and never set where they're going. I don't know yet if that's people checking a distance and leaving, or the panel not making the next step obvious enough.

EO-Map's panels tell you a bit about what EVE Online players are after. Over the 31 comparison days, Force Projection, which maps the jump bridge zones from the latest Cradle of War rules, was opened 344 times with half of its visits past a minute, and Kill Replay holds people nearly as well at 44%. The Activity panel is the most opened thing on EO-Map, 624 times, but only a quarter of its visits go past a minute. EO-Map also gets far more of its panel opens from shared links, 19% against EF-Map's 6%, which fits with Reddit posts linking straight into a panel.

How long people stay

Session length and return visits

From each site's own counters over the 31 comparison days, except the last row, which is Google Analytics over the same windows as the traffic sources above. Session length is the time from load to leaving the page.

More than half of all visits to either site are over inside a minute, which surprised me at first. Some of that will be people checking one thing, a distance or a system name, and leaving, which is a perfectly good use of a map. The row I'd look at is the one about input. About a quarter of EF-Map loads, and nearly a third of EO-Map loads, never get a single click, scroll or key press after the map appears. Whatever those people came for, they didn't touch the map.

EF-Map has the longer tail. 7.2% of EF-Map sessions run over an hour against 4.4% on EO-Map, and returning users account for 70% of EF-Map's sessions in Google Analytics against 42%. EF-Map's figures include my own use and EO-Map's don't, which will nudge EF-Map up a bit, and I can't separate myself out to say exactly how much. Even so, EF-Map looks like a tool a smaller group of people keep coming back to. EO-Map, so far, is more of a place people come to look at something and move on.

Does the map run smoothly?

Since the end of July EF-Map has measured frame delivery while you're actually moving the camera around, and EO-Map inherited the same code. It's judged against your own display, capped at 60 frames a second, so a 144Hz monitor running at 60 counts as smooth and remote desktop capped at 30 doesn't count as struggling. Slow frames, the hitches, are recorded separately, because an average hides them completely. Only 48% of EF-Map loads and 50% of EO-Map loads end up with a measurement. It needs a stretch of real camera movement before it measures anything, and some sessions never send a result at all, so this leaves out a lot of people who loaded the map and left.

Frame delivery

Sessions with a measurement over the 31 comparison days, from each site's own counters.

For the vast majority of people both maps run fine, 94 to 96% smooth. Where EF-Map loses out is hitches. Only half of integrated graphics sessions on EF-Map are smooth with no hitches at all, against 64% on EO-Map, and even on dedicated graphics it's 74% against 82%. EF-Map also has more sessions with frequent stalls, even though far fewer of its sessions are on integrated graphics. EF-Map draws over 24,000 star systems and EO-Map 5,485 solar systems, but I haven't tested whether that's what makes the difference.

Performance Mode looks like it does what it's meant to. On integrated graphics, EF-Map sessions that had been switched into Performance Mode automatically were smooth 93% of the time, against 84% for integrated graphics sessions still in the default mode. EO-Map shows the same thing, 93% against 88%. The samples are small, 223 and 109 sessions on EF-Map, and it isn't a controlled test, the people in each group aren't the same people, but it points the right way. There's more on how it works in the Performance Mode write-up.

Time to first frame

Over the 31 comparison days, from opening the page to the map drawing its first frame. A first visit means the map database had to be downloaded, a return visit means it came from the browser cache.

Load times are the one I don't have an answer for. On a first visit EF-Map gets more loads on screen in under 2 seconds than EO-Map, 63% against 43%, but it also has more that take over 8. On a return visit EO-Map gets 74% on screen in under 2 seconds, while EF-Map's 66% is barely better than its own first visits. A cached load should be a lot faster than a fresh download, and on EF-Map it mostly isn't. EO-Map has its own oddity, 8% of its cached loads take over 8 seconds, more than its first visits. Separately, 3.5% of the EF-Map loads that reported their graphics had no WebGL at all, so the map couldn't draw anything for them, against 0.9% on EO-Map.

Who's on the other end

Audience and hardware

Rows marked GA are share of users in Google Analytics. The rest are share of the loads that reported their hardware, from each site's own counters over the 31 comparison days. Screen width is in browser pixels, so a 4K monitor at 150% scaling counts as 2560 wide, and only Chromium browsers report RAM.

I guessed the hardware would be pretty much the same on both, and it's close. Around half the loads on both report 32GB of RAM or more, and over half are on 13 or more CPU threads. Where they differ, EF-Map is the higher end of the two, with 17% of loads on screens wider than 2560 pixels against 9%, and 27% on 17 or more CPU threads against 21%. The bigger differences are in how people get there. EO-Map is twice as mobile, 36% of users against 18%, and a third of EO-Map loads are in a browser window under 800 pixels tall against a fifth on EF-Map. Only 21% of EO-Map loads are touch-first, so that isn't all phones.

The one I can't explain is the browser and OS split. By map loads, Firefox is 16% of EF-Map against 10% of EO-Map, but by users Google Analytics has them level at about 9 to 10%, and I can't tell you why the two disagree. EF-Map users are also twice as likely to be on Linux and nearly three times as likely to be on a Mac. I don't know why, and I'm not going to pretend I do. On languages EO-Map is more international, with Russian at 11% against EF-Map's 6%. The US is the biggest country for both, 36% of EF-Map's users and 29% of EO-Map's, followed by the UK and Germany on both.

The blog

The blog counts its own views on the server, no Google Analytics, and filters out anything that identifies itself as a crawler. From 5 August to the end of 4 October that's 14,113 views from people and 17,977 from crawlers, so 56% of everything hitting the blog is robots. Search engines, AI crawlers, link previews. And that's only the ones that admit it. On 4 October the "human" count jumped to 471, and by early morning on 5 October it was already at 891, when every other day since 2 September has been somewhere between 74 and 415. Those views landed on almost every one of the 126 articles a handful of times each. That's not 900 people reading, that's a crawler pretending to be a browser. There were similar jumps on 29 August and 31 August to 1 September, but I don't keep per-article numbers that far back to check.

Most read EF-Map blog articles, 6 September to 3 October

Views counted by the blog itself with crawlers filtered out, over four weeks that avoid all of those jumps. 4,550 views across 125 articles.

The Cycle 6 roundup at the top had a big day on 28 September, 127 views the day before Cycle 6 ended, and the EVE token article had 90 on 27 September. The one that surprised me is the Grok article. It's an AI coding write-up, nothing to do with the game, and it was read between 9 and 20 times a day from 6 to 18 September before tailing off. The plain-English articles from August, the review, EVE Online against EVE Frontier, the token, who makes the game and the system requirements, take five of the top ten between them.

Google is not where most blog readers come from. Search Console shows 339 clicks onto blog articles in two months, while the blog counted about 4,500 views in just four weeks, and some of those will be crawlers that got through the filter. The blog deliberately doesn't record where visitors came from, so I can't tell you where the rest are from. What search does show is which articles people find on their own. The review gets the most clicks, 41, then EVE Online against EVE Frontier with 39 from 2,979 impressions, the system requirements article with 31, and the Three.js starfield write-up, a year old now, with 29. EO-Map's blog is only a few weeks old and has had about a hundred human views, so there's nothing to compare there yet.

So yeah

Two months in, EF-Map is quieter than it was earlier in the year but steady. It's found through Google, and it's used by a smaller group of people who keep coming back. EO-Map reached twice as many people in September, it's found through Reddit and YouTube, and most of them visit once. The bits that stand out to me as worth a closer look are the routing drop-off, why cached loads on EF-Map aren't faster, the hitches on integrated graphics, and the counters themselves, the page-load one on EF-Map and whatever was going on with EO-Map in late September. If there's something you'd like counted that isn't, let me know.

Related Posts

Half a Million Light-Years Saved: Eleven Months of EF-Map Usage Data is the July write-up, the long view of the same counters going back to August 2025.

I Cloned EF-Map to Make an EVE Online Map, and It Mostly Worked is how EO-Map came about, and why the two sites share so much.

EF-Map Performance Mode: Two Free Fixes and One That Costs Frames covers how Performance Mode works and what it costs on weaker hardware.

EVE Frontier Cycle 7 First Week: Flight, Mining and Fog of War is my take on the cycle that launched in the middle of this data.