Which games should I take off my cafe PCs?

The machine told you there was no room. So you spent an evening walking the floor deleting shader caches, old installers and two titles you were fairly sure nobody wanted. That bought back forty gigabytes and cost you an evening. It was the third time this year.

Every article about games in a cafe is about what to put on. The install list, the titles trending this month, the thing a group of fifteen year olds asked for on Saturday. Adding is the enjoyable decision so it gets all the attention, which is why what to install is a subject of its own. Four years of adding with nothing ever coming back off is how you ended up here.

Here is what owners miss. A game you installed in 2022 that nobody has opened since 2023 is not sitting there for free.

What the dead titles are actually costing you

Four bills, none of which appear as a line item anywhere.

Disk. A title lives on every machine that offers it. A hundred gigabyte install is not a hundred gigabytes. Across thirty PCs it is three terabytes of storage doing nothing. If you run images or restore-on-reboot, it is also three terabytes you copy, snapshot and push out every time you rebuild. That is the part that stings, because managing images on shared machines gets slower with every corpse you carry.

Patch bandwidth. A launcher does not know which of your titles is dead. It updates them all. A game nobody has opened in eleven months still pulls its patches down your connection on the same Friday evening a paying customer is trying to play something else. You are buying bandwidth to keep a dead game current.

Seats. If you bought concurrent seats for a title, you are paying for seats on something nobody launches. The same goes for a pool of shared launcher accounts held for a game that stopped mattering two summers ago. That is money leaving monthly in exchange for nothing.

Attention. This is the one I would fix first even if disk were free. Your menu is a shop window. A customer sits down, sees eleven categories and a wall of tiles. The five titles that actually earn you money are competing for that first glance against everything else on the screen. Burying your good stock behind clutter is a choice you did not realise you were making. The same thinking runs through what a customer-facing screen should show.

Go and look before you touch anything

You already have the evidence. Somebody just has to open it.

In HandyCafe this lives on the Statistics page, under the Applications tab. It ranks applications by cumulative usage time across every client PC and every session. You get an ordered list of what your floor actually runs rather than what you remember it running. Usage is also grouped by category. Games, browsers, communication and productivity each show as their own slice.

The ranking gives you the order. The trend view gives you the direction. Direction is the part that decides things. Set a date range and you can watch how usage shifts across it.

That distinction saves you from the mistake everyone makes on their first pruning session. Over eight weeks a dead racing title and an out-of-season football title look identical, because both show nothing. Stretch the range across eighteen months and they stop looking alike. One dropped off and never recovered. The other spikes every autumn when the new edition lands, fades through spring and comes back. Delete the second one in July and you will be downloading it again in September while three regulars stand at your counter asking where it went.

The category cut answers a different question altogether. If games are a thinner slice of your floor's hours than you assumed and browsers are a fatter one, then you are not looking at a pruning problem. Ask what your cafe is now.

The number will mislead you if you let it

The list you are about to make is wrong in ways the screen will not show you. I would rather you hear that now than in September.

Hours played is not value. It is the closest proxy you have and it is not the same thing. Three ways it will lie to you:

Low hours can mean high spend. Five regulars who book a table every Thursday, order food, sit for four hours and pay the bill are worth more to you than a browser game with triple the recorded minutes spread thin across people nursing a single hour. Cumulative time does not know who paid what. Before you cut anything off that list, put it next to the Sessions and Computer Usage tabs and ask which machines and which nights the title belongs to.

Low hours can mean buried. If a title has been sitting three rows down inside a category nobody scrolls to, its numbers are measuring your menu layout rather than customer taste. Move it to the top of its category, drop it into Quick Launch or the Hero carousel for a fortnight and look again. If it is still flat, now you have learned something.

High hours can mean nothing came back. A title can run all week and still not be the reason anybody chose your room over the cheaper one. The tracker records launches and duration. It does not record whether anybody enjoyed themselves or whether one game is why a group of six keeps picking your place. No usage counter has ever measured that. A usage counter counts. Loyalty was never in that column.

Hide it. Delete it a month later.

This is the discipline that makes the whole exercise safe. Do not skip it because you feel confident.

Every app in the Client Menu has a visibility toggle. A hidden app is not displayed to customers and is not deleted. The files stay on the machines, the entry stays in your config and the title simply stops occupying window space. Changes push to connected clients instantly with no restart. You can do this at four in the afternoon from behind the counter without walking to a single PC.

So the routine is: hide the candidates, tidy the sort order of what is left so your earners sit where eyes land first, then wait a month.

If nobody asks in that month, you have your answer and you can uninstall with a clear head. If somebody asks in week two, you flip the toggle back and the cost of being wrong was one short conversation. Compare that with deleting first, where being wrong costs you a reinstall, a hundred gigabyte download and a customer who has already found out your cafe no longer has the thing they came for.

The seats you forgot you rent

While you have the usage ranking on screen, open Software Licenses next to it.

Software Licenses counts concurrent seats for a title and blocks overuse. It does not hand out credentials. What matters here is the cross-reference: anything you are holding seats for that does not appear anywhere near the top of your Applications ranking is a bill for a title your customers have stopped opening. The same check applies to shared launcher logins. If a provider pool exists for a game nobody launches, disable that provider and stop renewing. There is more on running those pools properly in managing shared game accounts.

What I would actually do

1. Open the Applications tab and set the range to the last eighteen months.
2. Write down everything with no meaningful usage in the last twelve, then check each one for an annual spike before it goes on the list.
3. Pull anything your best-spending regulars book time for back off that list, whatever the hours say.
4. For the ones you suspect are buried rather than dead, promote them instead of hiding them and give it two weeks.
5. Hide the rest. Reorder what is left so your five earners are the first things a customer sees.
6. Come back in a month. Uninstall what nobody missed, reclaim the seats and the disk, then put a reminder in your calendar to do this again in six months.

Twice a year is the right cadence. January and July, an hour each time.

An owner who never removes anything is postponing a decision. The postponement bills them every month. Your menu should look like the room you run today.