How to ask 3,940 installers four questions, in 21 languages

Put four numbered questions in the body of the email, in the reader's language, with nothing to click. Send a short follow up about eight days later that carries the questions again, then one last one. Answer every reply in minutes, repeating two or three things the person said. In the Shelly Group programme this reached 3,940 installers and produced 360 replies and 225 survey answers.

This is a how to based on one real programme. It is not a sales campaign, and it is a good model for anyone who needs to hear from a base of partners, resellers, installers, dealers or franchisees who do not answer long forms. The facts and numbers are in the case study. The examples below are generalised.

What was the problem?

Shelly Group makes smart home and building automation devices and has a registered installer network in more than 100 countries. Many portal accounts were quiet. The company had little signal on who installed its products, on what kind of projects, or what stopped installers from installing more. A partner team cannot write to thousands of people in dozens of languages, read every answer and spot the one quietly fitting out a very large hotel.

What does the email look like?

The structure matters more than the exact words.

1. A team voice, not a named person. The email comes from a team, peer to peer, short, with no selling.
2. One line of context. Why you are asking, and that you read the answers.
3. Four numbered questions in the body. The four questions were abridged in the study as: what kind of projects do you mainly work on, roughly how many projects per month, what is your biggest challenge, and what is the one thing the company could do to support you better.
4. A request to reply to the email. Nothing to click, no form, no link.

Writing the questions into the email is the choice that matters. A survey link asks a stranger to leave their inbox for a page they do not know. A reply takes the same effort as answering a colleague. Of 210 complete surveys, 172 came back in one message.

What do the follow ups do?

They carry the load. In the June to August waves 13.6% of installers replied. With the first email alone the figure would have been 3.3%, so the follow ups made it 4.2 times as many. Three in four replies came only because of a follow up.

• First email: 3.3% replied.
• Follow up 1: 5.3% of those still silent.
• Follow up 2: 3.0%.
• Follow up 3: 3.6%.
• Follow up 4: 6.3%.

Share of all replies by touch: first email 24%, follow up 1 37%, follow up 2 17%, follow ups 3 and 4 together 22%. The first follow up was the best single touch. The median gap was 8.3 days after the first email, then about ten days.

Two rules came from the data. First, every follow up carries the questions again. When a reminder left them out, seven installers wrote back to ask for them. Since then every email contained them. Second, cap it. Touches four and five produced 22% of replies but also the only two requests to stop, across 1,966 installers and more than 6,700 emails. Shelly capped the sequence at the first email plus two follow ups. The post on how many follow ups sets this beside our other campaigns.

What about language?

Installers in 111 countries do not all read English at work. The agent wrote in 21 languages, and 51% of installers replied in a language other than English. See write in the local language for the numbers and the setup.

How fast should you answer?

Fast. The agent answered installers in a median 1.4 minutes, 70% within five minutes, and 50% of installer messages arrived outside office hours in Sofia time. Among the installers who converted, 67% got a reply within 15 minutes of their first message. Each reply repeated two or three things the installer had said, so the person knew it had been read.

An answer that takes a day reads as an automated acknowledgement. A reply in a minute that picks up their own words reads as attention.

When do you offer a call?

After, and only after. Once the survey is answered, the agent adds one optional line offering a 30 minute call, with a booking link. Nine of the 17 calls booked came from that line, three within 33 to 40 minutes. Offering a call before the survey turns a listening exercise into a pitch, and the answers dry up.

What does a person need to handle?

The agent triages and a person closes. Replies about pricing, buying direct, distribution, returns, support problems and partnership questions get a short holding reply and go to the company's staff. Big projects are flagged: the study describes installers working on a very large hotel or an integrator serving many client companies, and those threads reach a person quickly. 66 converted threads were handed to a person.

Rules the agent works by, generalised:

• Do not promise prices, codes, dates or certification status.
• Commercial and support topics get a holding reply, then a person.
• An opt out closes the thread for good.
• No booking link before the survey is answered.

What does "conversion" mean here?

Be careful here, as the study is. 238 of the 360 installers who genuinely replied did at least one of four things: answered the survey, booked a call, surfaced a big project, or engaged with a thank you code. That is a measure of engagement. No sales or revenue is claimed, and you should not read it as one.

What does a made up example look like?

This is an invented example, not a customer. Say you sell commercial coffee machines through 800 dealers in six countries. Your four questions could be: what kind of venues do you mainly fit out, roughly how many machines do you place per month, what most often stops you placing more, and what one thing could we do to help. Each is short, each has a free text answer, and each would change something you do, whether training, stock or support.

Then run the same rhythm as the Shelly programme: the first email, a follow up about eight days later with the four questions repeated, one last follow up, and then stop. Read the first twenty replies yourself before you trust any summary. The patterns that matter, such as the same obstacle named by many dealers, show up quickly when you read raw answers.

What can go wrong?

• The questions are too many. Four was the number used here. Every extra question lowers the chance of a complete answer in one message.
• The follow up drops the questions. In the study seven installers wrote back to ask for them, so repeat them every time.
• The reply is slow. An installer who gets an answer in a day has already moved on. The median reply in this programme took 1.4 minutes.
• Nobody owns the handoff. Pricing, support and big project threads need a named person, otherwise the best leads sit in a queue.
• Success is overclaimed. Count engagement as engagement. The study claims no revenue, and a survey programme should not either.

How do you copy it?

1. List your base and the language for each country.
2. Write four questions you would actually act on. If an answer would not change anything, drop the question.
3. Put them in the email body, numbered.
4. Send from a team voice with no sell.
5. Follow up about eight days later, with the questions again, then once more. Stop.
6. Answer every reply within minutes, repeating their words.
7. Offer one optional call after the survey is answered.
8. Hand pricing, support and big projects to a person.

The playbook has the full sequence and the template has the email.