Chapter 06 · Workshop report: software
Why we built our own CMS
An honest account of why we built our own signage software, what it cost us and when an off-the-shelf system is the wiser choice.
We wrote our first playback system ourselves in 2016, alongside the enclosure for the first totem – there was simply nothing on the market that met our standards. Later, as the installations multiplied, we used off-the-shelf systems where they fitted. Until I stood beside the till in a bakery and watched the owner try to change the lunch menu: she clicked through four levels, found a timeline, moved an element by accident and ended up with a blank screen. The software could do everything – transitions, layers, video walls, permissions for a hundred sites. Everything except what this woman needed at six on a Tuesday morning. That was the day we started building our own again.
What bothered us about the off-the-shelf systems
I want to be fair: the systems of the time were not bad. They were built for somebody else – for chains with a marketing department, their own IT and people who use the tool every day. Our customers are bakeries, car dealerships, practices, municipalities; they spend twenty minutes a week with it.
Four things kept coming up:
- The licence ran per screen per month. Anyone operating many surfaces pays twice: for the hardware and for the right to switch it on.
- Most of the functions went untouched. Layers, scripting languages, animation curves – tools for designers, not for businesses.
- The interface assumed knowledge nobody had. After the training it went well for two weeks, then the branch manager called us again.
- Support sat in another time zone. A black screen on Saturday morning was dealt with on Monday afternoon.
On top came the question of where the content sits. For a menu board that does not matter; for meeting-room booking data it does.
What we wanted instead
We did not start with a list of functions but with five sentences on a sheet of paper. Each one was a decision against something else.
- A baker operates the system without training. If she needs a manual, it is the interface that is wrong, not the baker.
- Templates instead of design freedom: we set the grid, the typeface and the colour, the customer fills in text and image. Why that saves an installation is in the chapter on content.
- The data sits in Switzerland, on servers we can name. The revised data protection act requires knowing where data sits and who has access anyway.
- Every playout is logged. In our advertising network, that proof is the basis for billing; without it, third-party advertising cannot be sold seriously.
- The system knows our processes. A quotation becomes an order, which becomes a site with devices and a service history. If a player stops reporting in, the system knows who it belongs to.
The last point was the real reason. No off-the-shelf software knows your processes.
Start small and let it hurt
The first version could do exactly three things: upload images, put them in an order, send them to a screen. No schedule, no zones, no permissions. With that we went to three customers we knew well and said: try it, and tell us where it hurts.
It hurt in places we had not expected. Nobody missed transition effects. Everybody missed a preview showing what the slide would look like on the wall later. One branch manager uploaded the same image three times because after saving she could not see whether anything had happened. Things like that only show up where somebody works with it on a Tuesday morning under time pressure.
A rule came out of it: every new function needs a name and an occasion. Not «it would be nice» but «otherwise the municipality cannot show its referendum posters». We do not build functions without an occasion. That sounds strict and it saved us from a product that can do everything and helps nobody.
Five functions instead of fifty
After years in operation, actual use looks soberingly simple. The vast majority of installations need five things: upload content, put it in an order, control it over time, see the state of the screen, and reach somebody when it jams. Everything else is accessories.
Schedules beat effects. A bakery showing croissants in the morning and the menu at lunchtime gets more out of its surface than any animation. Automatic content – weather, opening hours, stock levels from the merchandise system – keeps the screen fresh even in a quiet week.
A screen that reports in on its own is worth more than any design feature. Our players check in at short intervals and send a screenshot on request. If a check-in fails to arrive, we find out before the customer notices. That has saved us many call-outs.
And less choice means fewer mistakes. Since the templates fix type size and line count, we hardly see illegible slides any more. Nobody has missed the freedom we took away.
What your own software really costs
I will not talk about the price here in figures but in obligations – those are more honest. Development is the smallest part and the only one that is finished at some point. Everything after it never stops.
What you take on if you build it yourself:
- Maintenance: player operating systems change, browsers change, interfaces to tills and calendars change. That is work no customer ordered.
- Security updates: a system reachable from the internet has to be maintained in the week a vulnerability becomes known, not next quarter.
- Availability: if our system stops, screens stop across German-speaking Switzerland. That responsibility cannot be passed on.
- Knowledge in more than one head: a system only one person understands is a risk.
For us it works out, because the effort spreads across many sites and across our own advertising network, and because the same system carries the processes from quotation to service call. For a business with four screens it never works out.
When an off-the-shelf system is the better choice
I do not recommend our system everywhere, and certainly not custom software to everybody. There are cases where an established product is clearly the better fit.
An international group with its own IT and sites in several countries needs a provider who fulfils contracts in different legal jurisdictions, presents certifications and brings a role model for hundreds of users. For that we are the wrong people.
Special cases with a fixed integration belong here too. If your branches depend on a particular till system and a proven signage extension exists for it, take that. Rebuilding an interface that already exists is expensive and thankless.
And very large networks sold programmatically need connections to trading platforms, standardised formats and counting methods that media agencies accept. Nobody builds that apparatus on the side; more on this in the chapter on the advertising network.
The rule of thumb is unspectacular: the more distinctive your processes, the more custom software pays off. The more standardised your requirement, the more likely it already exists off the shelf.
Questions to ask every supplier
Whether an off-the-shelf product or a provider with its own software: these questions cost ten minutes in a sales meeting and save a great deal later.
- Where is my data, on whose servers, in which country – and who at the provider can view it?
- How do I get my content back out? Is there an export, and in what format?
- How is the licence model built? Per screen, per site, per user – and what happens if I add or remove a screen?
- Is there a playout log, how long is it kept, and can I download it?
- How does remote maintenance work? Who can connect, is it logged, and can access be switched off?
- What happens if I change provider? Do the players belong to me, and can other software run on them?
- What happens if you no longer exist in five years?
The last question is uncomfortable, and that is exactly why I like asking it. Anyone who answers calmly and explains how, in the worst case, you get at your content and your devices, has thought about it.
Own system versus off-the-shelf system
I give the following assessment in conversation too – as a supplier who built his own, but without sugar-coating.
| Criterion | Own system | Off-the-shelf system |
|---|---|---|
| Time to the first picture | long, it has to be created first | short, open an account and start |
| Fit with your own processes | exactly as you work | you adapt to the product |
| Use by the untrained | can be deliberately kept simple | often built for specialists |
| Data storage | freely chosen, location determinable | predetermined, needs clarifying |
| Ongoing effort | maintenance in-house | licence per screen, grows with the network |
| Integration with till and stock | you build what you need | available if your system is included |
| International requirements | demanding | usually covered |
| Programmatic sales | very demanding | available from large providers |
| Dependency | on your own team | on the provider and its pricing model |
No single line decides it. Read the three that affect your everyday work.
Key pointYour own software pays off where your processes are genuinely your own – everywhere else you buy responsibility nobody ordered.