Answering an Inbound Enquiry Fast, Even in the Evening

How fast should you answer an inbound enquiry, really?

Short answer: aim for a useful first reply on the same day, ideally within the hour during working hours, and an immediate acknowledgement in the evening and at weekends. For a small business, what matters is that every enquiry gets an answer, not that the answer lands to the minute.

An inbound enquiry is a message a prospect sends of their own accord: a website form, an email, a WhatsApp Business message or a missed call. Their interest is at its peak the moment they send it. It then fades, and it fades faster when your competitors reply before you do. This article gives the published orders of magnitude, the workflow to put in place, what an AI can prepare, and what must stay in human hands.

What the published figures say, and their limits

Studies on lead response time are almost all American and old. They point the same way (faster is better), but they measure sales calls in large teams, not a small business receiving thirty enquiries a month. Read them as orders of magnitude.

One point is worth noting: the same study shows that a callback at five minutes qualifies fewer leads than one at a single minute (31.3% against 39.6%), but far more than a callback the next day. The gain is mostly between "the same day" and "the day after tomorrow", not between two and five minutes.

Why enquiries wait

In a small business, an inbound enquiry rarely waits out of negligence. It waits because nobody is assigned to pick it up at the right moment. Three causes come up again and again.

• It arrives outside office hours. A business owner works on their project in the evening or at the weekend, which is exactly when the form gets filled in. The reply goes out on Monday morning, after three other urgent items.
• It arrives through four channels. Form, email, WhatsApp, voicemail: each has its own inbox, and nobody watches them all at once.
• It needs sorting before any reply. You have to understand what the person wants, check whether it is your line of work, find out whether they are already a client. That sorting takes ten minutes, and it gets postponed.

The third point is the one an AI handles best. The first two are solved by a rule: a single intake channel and a named person per time slot.

The five-step workflow

Here is a workflow we describe for incoming request qualification, seen here from the manager's desk.

1. Single point of intake. Every channel feeds the same queue, even if it is just a shared spreadsheet. Nothing gets lost between two inboxes.
2. Immediate acknowledgement. A short message confirming receipt, stating an honest delay ("reply before noon tomorrow") and, if you wish, asking one or two useful questions. It reassures without pretending to answer.
3. Prepared sorting. The AI reads the enquiry, classifies it (new client, existing client, out of scope, cold solicitation), spots the urgency and the missing details, and prepares a three-line summary.
4. Draft reply. It proposes an answer based on your own templates and your real availability. It does not answer for you.
5. Approval and sending. A person reads it, corrects it if needed and sends it, or takes over by phone. Enquiries classified as cold receive a follow-up planned in advance.

This flow follows the rule we apply everywhere: the AI prepares, a person approves. The time saving comes from steps 3 and 4, not from automatic sending.

What can be automated, and what should not be

A worked example (assumptions, not a measurement)

Take a service business that receives 30 enquiries a month. Assumption: sorting and a first reply take 10 minutes of manual work per enquiry, which is 5 hours a month. With prepared sorting and a draft to review, approval takes about 3 minutes, which is 1.5 hours. The direct saving is about 3.5 hours a month, which is modest.

The real stake lies elsewhere. If ten of those enquiries arrive outside office hours and wait a day on average, an immediate acknowledgement and a reply prepared in the morning change the delay for most of them. If just one more enquiry in ten turns into a CHF 3,000 job, the gain far exceeds the time saved. These figures are calculation assumptions: measure your own before deciding anything. The formulas and prices section of our site shows how we frame this kind of project.

How an AI classifies an enquiry

Classification relies on a small number of categories that you define, not on free judgement. Four are enough to start: new client within your scope, existing client, out of scope, cold solicitation. For each enquiry, the AI states the category, the degree of urgency (a quoted deadline, a word such as "urgent") and what is missing to reply (budget, date, address, quantity). When it hesitates, it must say so and hand the enquiry to a person rather than decide. Plan to review its classifications during the first month: you will quickly see which enquiries it misreads, and you can adjust the categories.

What to put in the acknowledgement

A good acknowledgement has four elements: confirmation that the message arrived, a delay you can keep, the name of the person who will reply, and one or two questions that move things forward. An example structure, to adapt to your own tone:

Hello, your enquiry has arrived. Marie will reply before noon tomorrow. To save time, could you tell us by what date you need this and roughly what budget you have in mind? Thank you.

The message announces no price and promises nothing except the delay. If the delay is stated in the acknowledgement, it must be kept: a missed delay is worse than a long one.

Five common mistakes

• Promising a delay you cannot keep. State what your team can actually do, holidays included.
• Sending a generic reply. A reply that picks up nothing from the enquiry reads like an automatic message, and it is one.
• Letting the AI set a price. Prices come from a grid you have approved, never from free generation.
• Not planning an exit. As soon as a person replies, the automation must go quiet, otherwise the prospect receives two contradictory messages.
• Not measuring. Without a median delay before and after, you will not know whether the workflow is worth anything.

When this is not the right choice

Do not automate enquiry qualification in three situations.

• Fewer than ten enquiries a month. A reply template and a phone notification are enough, and cost less.
• Enquiries that are all different. If each one needs its own expertise (technical advice, bespoke consulting), prepared sorting saves you nothing.
• A relationship built on the person. If your clients expect to speak to you, an acknowledgement is enough and the rest must stay human.

In every case, start by measuring. For two weeks, note the time each enquiry arrives and the time of the first reply. If your median delay is already under four hours, your problem lies elsewhere, for example in the follow-up that comes next, which we cover in following up on a quote without nagging.

Where to start this week

1. Gather your channels into a single queue.
2. Write an honest acknowledgement, with a delay you can keep.
3. Measure the first-reply delay for two weeks.
4. Only then decide what is worth having an AI prepare.

If you want to see what this looks like on your own processes, the processes the studio automates are listed on the English home page.