How to Automate Client Follow-Ups for a Consulting Business

A consulting business rarely loses a deal in the meeting. It loses it in the ten days afterwards, while you are delivering for the clients you already have and the new enquiry quietly ages in your inbox.

The fix is not a bigger CRM or more discipline. It is a follow-up sequence that runs whether or not you remember it exists. This guide sets one up. Budget about two hours once, then it runs for every enquiry from that point forward.

What you are building

One automation with three parts. A new enquiry gets captured, an AI draft of each follow-up gets written in your voice, and the drafts get sent on a fixed schedule until the prospect replies or the sequence ends. You approve the drafts on the way out, or you let it send, whichever you can live with.

You need three things, all of which have free plans that cover this: a form or inbox that receives enquiries, an automation tool (Zapier, Make or n8n), and an AI assistant (ChatGPT or Claude).

Step 1: give every enquiry one front door

Automation cannot fire on something it cannot see. Right now your enquiries probably arrive four ways: a contact form, direct email, LinkedIn, and someone texting you because you met at a thing. Pick one destination that all four end up in.

The cheapest version is a single Google Sheet or Notion database with five columns: name, email, what they asked about, source, and date. Point your contact form at it directly. For the other three, spend ten seconds pasting the enquiry in yourself. That is the only manual step in this whole system, and it is the one worth keeping, because it forces you to read the enquiry.

Step 2: teach the AI your voice, once

The reason automated follow-ups sound automated is that people write the prompt in ninety seconds. Spend twenty minutes instead, and the difference is visible in every email the system ever sends.

Find three follow-up emails you actually sent that led to work. Paste them into your AI assistant with this instruction:

Save the output. That block of rules goes at the top of every prompt in this system. It is the difference between a sequence that sounds like you and one that sounds like a plugin.

Step 3: write the three touches

Three is the number that works for consulting. One touch is forgettable, two feels like a formality, and past four you are the person who cannot take a hint. Each touch has to earn its place by doing one of three jobs: re-anchor on the problem they told you about, make the next step smaller, or ask for a clear yes, no, or later.

1. 1Touch one, next morning. Recap the specific bottleneck they described in their own words, and name one thing you would look at first. Nothing else. This is not a pitch, it is proof you listened.
2. 2Touch two, four days later. Send something useful that costs them nothing to read: a short teardown, a relevant example, a checklist. Attach no ask beyond a reply.
3. 3Touch three, day ten. Ask for the decision directly. Yes, no, or a date to revisit. Give them permission to say no in the sentence itself, because that is what actually gets a reply.

Never send a follow-up whose entire content is that you are following up. If a touch has no substance, delete it from the sequence rather than padding it.

Step 4: wire it together

In your automation tool, build one scenario. The trigger is a new row in your enquiry sheet. The first action calls your AI assistant with the voice rules, the enquiry details, and the instruction for touch one. The second action creates a Gmail draft, or sends it, depending on how much you trust it in week one.

Then add two delays, four days and six days, each followed by the same pattern with the touch two and touch three instructions. Add one condition at the top of each: if the reply column in the sheet is filled in, stop. That single condition is what stops the system emailing someone who already said yes.

What good looks like after a month

Every enquiry that arrives gets three thoughtful touches without you scheduling anything. You spend your attention on the replies rather than on remembering who has not replied. And the enquiries that were never going to convert exit the pipeline on day ten instead of sitting in your head until Christmas.

This is one of the seven workflows in the AutomateFirst playbook, where it is written out with the exact prompts, the tool screens, and a checklist. If this guide was useful on its own, that is the point.