WRITING

Integrating with big platforms is easy until you need someone to answer

The code took days. Getting a human to answer took two weeks. In December a simple webhook with a big activities booking platform reminded me where integrations really live.

The code took days.

Getting a human to answer took two weeks.

In December I tried to connect to a big activities booking platform. The plan was simple. A webhook, so that every booking made on the platform would arrive in our system automatically. No copying. No spreadsheets.

I sent the first email following their own documentation. Nobody answered. A week later I wrote again, this time just asking who we should even talk to.

Silence. The professional kind.

Things only moved after the client called them. Suddenly they asked for our link. The first one returned an error on their side, so we sent another, and that one worked. In the same thread I asked whether there was any way to simulate a gift voucher purchase, so we could test without touching real bookings.

That question is still open.

Support for the support

In the middle of all that, a client's support chatbot failed.

We wrote to the vendor. Their answer: contact them through their own chat, after registering with our purchase code.

So the client's support tool could only get support through the vendor's own support tool. Like calling the fire brigade and being told to fill in a form at the station. Which is on fire.

I'm still not sure whether to laugh.

Integrating with a big platform is a technical task for days and a patience task for weeks.

The line nobody writes on a quote

I'll admit I underestimated this. Again.

When I estimate an integration, I estimate the code. How long to read the docs, write the handler, test the data. I almost never estimate how long I'll wait for a reply.

And the waiting is often longer than the work. It just never shows up in a plan, because nobody likes writing "two weeks: waiting" on a quote.

Back in January 2020 I wrote that you shouldn't build your business on a tool whose vendor doesn't answer. Five years later, I'm still relearning it. With bigger names.

What I'd do differently isn't technical. Ask for a test environment on day one, not after the first problem. Find the one door that actually works, before you need it. Tell the client honestly: "we're ready, we're waiting for them" is a fair update; a date that depends on someone else's inbox isn't. And keep a manual fallback, so nobody goes dark while you wait.

December had one more reminder of who sets our calendar. Our advisory firm confirmed that Verifactu had been postponed: to January 1st 2027 for companies and July 1st 2027 for the self-employed. After the certificate saga in November, welcome news. Also another date we didn't choose.

We've built Foxnice since 2014 without investors, and our products are paid for by client work. A two-week wait isn't just a delay. It's time we can't spend elsewhere, and a client wondering why it's taking so long.

Patient money teaches you to wait. It doesn't teach you to wait without a plan.

The code took days.

Next time, the plan will say how long the waiting takes.

P.S. If you work at a big platform and you're reading this: hello. We're still waiting on the voucher question.