Requirements for a technical partnership: test them with your own outage
The integration between your webshop and the carrier is dropping orders again. You report it in the support portal, get an automated reply, and later that day someone calls who first wants to know how your systems talk to each other. You explained that last time too, to someone else. On a day like that you find out which requirements for a technical partnership you should have set when you chose your partner.
The presentation and the competitive quote that won you over are long forgotten by then.
Below are the requirements in the order we would test them, starting with the requirement that one party takes responsibility when your digital operation is under pressure.
One point of contact holds the chain together
Many organisations have development, hosting, maintenance and support spread across different suppliers. When something fails, the developer is quick to point at the server, the host at the code and the provider of an external service at the API. Meanwhile your customer is waiting for a fix.
So the first thing we would require is one technical point of contact who keeps an eye on how everything fits together, investigates the cause and stays in charge until the problem is solved. That party has to oversee the whole chain, including systems it did not build itself. For a webshop, the chain runs from stock management to the carrier; for a SaaS platform, from the application code through the database to server capacity.
A check you can do today: take your last outage and count how many parties you had to call before someone found the cause. If the answer is more than one, you know where your first requirement lies.
Technical partnership requirements start with your process
An app, internal system or API integration has only succeeded when a business process becomes faster, more reliable or easier to scale. A good partner therefore starts with your work and gets to frameworks and servers later. Where does your team lose time every day? Which manual step causes errors? What happens if an integration is down for an hour, which data is sensitive and which processes must never stop?
If someone skips those questions, you may end up with something that is technically tidy and falls short in daily operations.
Pay attention to what they advise against, too. Whether you need custom software at all should be open for discussion: sometimes an existing application is good enough, sometimes a limited API integration does the job where a whole new platform was proposed. For a growing business we would build in phases: solve the biggest bottleneck first, measure what that delivers and then expand in a targeted way. A partner who advises on impact and long-term maintenance will suggest that kind of phasing on their own.
Code and server belong in one analysis
Software that works well in a test environment can still be slow or unstable in production. That happens, for example, when capacity, caching, database settings, security or monitoring do not match the application. Your user only sees that the page will not load.
If the knowledge of the code sits with one party and the knowledge of the hosting with another, analysis takes far longer than it should. Whoever solves the problem needs to know how the application behaves on the hosting environment, which processes demand a lot of capacity and which changes carry risk.
With peak loads this matters even more. A publisher gets traffic after a news story, a webshop during a promotion, and some platforms process large data files at fixed times. Ask where the limit of your environment lies and how it is prepared for those moments. You want that answer before the first incident gives it to you.
Ask who picks up your ticket, and when
A support portal with automated replies is little comfort when your customers cannot place an order. You want to know who picks up the report, what the first action will be and when you will get a meaningful update. Responding quickly counts, and clear communication carries at least as much weight.
So ask how support actually works behind the promise of availability. Do you get direct access to people who know the technology? Are incidents logged and reviewed systematically? And do you get an explanation in plain language of what is needed to prevent a repeat?
A partner who promises to carry out every request immediately is one we would view with suspicion. Ask how priorities are set and who already knows your setup when your report comes in. You build that context in a relationship, not in a single ticket.
A backup proves itself when you restore it
Security often only comes up after an incident, an audit or a question from a larger customer. That is late. A technical partner should be able to explain from the start how access is managed, how updates are carried out, where backups are kept and how recovery is arranged.
Which measures fit depends on what you run. An internal planning tool needs something different from a platform with personal data, payments or confidential documents. These four questions are fair in every situation:
- How quickly can a backup be restored?
- Who has access to production data?
- How are vulnerabilities followed up?
- What happens when a component fails?
For the first question we would want to hear when a backup was last actually restored. A backup that is never checked gives a false sense of security. The same goes for monitoring: it only has value when a warning leads to action. Continuity lives in processes that are carried out, tested and improved on a regular basis.
Agree on extra work, and who decides
Custom development and maintenance do not always come with a fixed price. A change can grow once systems turn out to be connected or old code imposes limits. That comes with the territory, as long as your partner flags it early and makes the consequences clear.
Put in writing what falls under regular maintenance and what counts as new development. Agree how extra work and urgent requests are assessed, and who makes the call when speed and perfection cannot both be had.
You do not need to become a technical specialist to take part in that discussion, as long as your partner translates choices into business impact, with sentences such as: this approach goes live sooner and needs more maintenance later. Or: this extra investment lowers the risk of downtime during peak hours. If all you get back on a choice is jargon, ask what it will cost in maintenance later and which risk it removes.
Bring your last outage to the meeting
How a partnership will go is often visible before there is a proposal on the table. Notice whether someone asks about your operation, dependencies and risks, and whether they point out limits. Good signs are concrete: a conversation about your current situation, phasing and maintenance after go-live, about who your contact person will be and how incidents are handled. If a party says yes to everything straight away, without looking into it, that tells you enough as well.
Look beyond the first assignment while you are at it. Can this party handle both a small improvement and a complex platform project? Is there capacity for maintenance alongside development, and does communication stay direct as the partnership grows?
Before that meeting, write down your last outage: what went wrong and how long it took before someone found the cause. Put that situation to every party and ask how they would have handled it. At LJPc in Assen we build and manage software, apps, websites and hosting under one roof, so with us you can walk through an outage like that from the code to the server.