The usual pattern is enthusiasm followed by silence. A church announces translation, offers eight languages, tells everyone at once, and within a month it has quietly stopped — because one volunteer was carrying all of it and they had a fortnight's holiday.
A slower start survives. What follows is a three-month plan built around that single observation: almost every failure is organisational rather than technical.
Weeks 1–4: one language, one service, no announcement
Pick the single language most needed and run it for the people who need it, without telling the whole congregation.
This month is for finding your problems while the stakes are low. There will be problems, and they will be the same ones everyone has: the microphone feed is worse than you thought, the sign is in the wrong place, and a handful of names and terms come out mangled every week.
- Read a transcript properly. Not skim — read five minutes of it. If the text is fluent but wrong in places, that is an audio problem, and it is the only problem worth solving this month.
- Sit with the two or three people using it and ask what it is actually like. Ask specifically whether it keeps up, and whether it makes sense — those are different questions with different causes.
- Write down the words it gets wrong. Place names, the church's own vocabulary, people's surnames. Most systems let you supply a list, and it makes a measurable difference on difficult audio.
- Do not announce anything. A first impression is expensive to repair.
Weeks 5–8: a second operator, then the announcement
Now open up the languages you know are represented — and, more importantly, train a second person.
A system with one operator is a system with an expiry date. It will run beautifully until that person is ill, away, or has a difficult week, and the Sunday it fails will be a Sunday somebody was relying on it. Two people, both of whom have started a broadcast alone at least twice, is the minimum.
Move the code somewhere permanent this month too. A code that lives on a slide depends on somebody remembering to add the slide; a framed sign by the door depends on nobody.
Then announce it properly. One clear sentence in the welcome, repeated for three consecutive weeks, does far more than a single notice — people need to hear a thing about three times before it registers as available to them rather than as news.
Keep the wording an offer rather than a diagnosis: "if you'd rather follow in another language, there's a code on your notice sheet."
Weeks 9–12: look at the numbers, then widen
By now you have real data: how many joined, which languages, how long they stayed.
Expect at least one surprise. The most common is a language nobody planned for turning out to be the most used — often people who understand the spoken language perfectly well and find reading less tiring over forty minutes. Adjust to what people actually chose rather than to what you assumed in week one.
Watch the length of sessions, too. People joining and leaving after two minutes is a signal, and it nearly always points back at audio or at a delay long enough to make following tiring.
This is also the point to extend beyond Sunday. A midweek study, a parents' meeting, a members' meeting. These are frequently where the value is highest, because they are the rooms where someone has to participate rather than only listen.
What to avoid
- Do not announce before it is reliable. The congregation gets one first impression.
- Do not offer twenty languages on day one. It promises a level of service you have not tested, and an empty language list looks worse than a short one.
- Do not let it rest on one person. This is the single most common cause of failure, by a wide margin.
- Do not skip the audio work. Every other problem is downstream of it.
- Do not measure success by uptake alone. Four people, if they are the four who needed it, is a success.
At ninety days: the review that makes it permanent
Put fifteen minutes on a leadership agenda and bring three things: how many people used it, which languages, and what it cost. Add one sentence from someone who used it.
A trial that reports back becomes a permanent arrangement without another argument. A trial that quietly continues without a review is the one that gets cancelled in a budget meeting eighteen months later by somebody who never knew it was there.
What the second ninety days look like
Briefly, so you know where this is going. Months four to six are usually about the rooms rather than the technology: extending to the meetings that matter, adding the languages the figures revealed, and — if you keep an archive — making sure your website says, in those languages, that any of this exists. Most churches find the archive quietly becomes the reason new people arrive at all.
Who does what, written down
The plan above assumes somebody owns it. Churches that skip this step find that everybody assumed somebody else was doing it, which is how a working system stops without any decision being taken.
Three roles, and they can be three people or two:
- The operator. Starts and stops the broadcast, checks the microphone, watches the screen during the service. A rota job, and the one that needs two trained people.
- The owner. Holds the account, sets the languages, looks at the figures once a term and reports to the leadership. Usually staff, and it should be somebody who will still be there in a year.
- The person visitors ask. Whoever is on the door needs to be able to say what the code is for in one sentence. That is the whole training.
Write the three names down somewhere the leadership can see. It takes five minutes and it is the difference between a project and a habit.
Start the first four weeks
InMyTongue needs no hardware and no installation, and new accounts include free credit — enough to run a quiet first month before you commit to anything. Begin week one.