Which part of your sales funnel is ready for an agent?

How many new contacts does your business make in a day?

Most owners cannot answer that, and I mean that without judgment, because I have asked it hundreds of times. The answer usually arrives as a range, or as a story about last month, or as a confident number that turns out to be the number of messages received rather than the number of people actually spoken to.

Six years ago I wrote that a sales funnel is a flight plan: whoever plots the route and tracks it has a much better chance of arriving than whoever leaves themselves at the mercy of the weather. I closed that post by asking whether you could really run all of it by hand, on a spreadsheet or on the back of an envelope. In 2020 the honest answer was "not for long, get a CRM."

The question is the same today. The answer changed, and it changed in a way that is easy to get wrong in an expensive direction.

A funnel is a process, and processes have a readiness

I have written elsewhere about the four filters I use to choose the first process a business hands to an agent: how often the work happens, whether it already exists in written form, how cheap a mistake is to fix, and whether one named person owns the result.

A sales funnel is not one process. It is three, stacked, with very different answers to those four questions. That is the whole insight of this piece, and once you see it the decision stops being a matter of taste.

So let us run the filters, stage by stage, out loud.

Top of funnel: passes all four

The work at the top is capturing a contact, replying fast, asking the same handful of qualifying questions, and recording what came back.

Four for four. This is where a first agent belongs, and it is also where the pain is quietest, which is exactly why owners skip it. Nobody lies awake over qualifying questions. Meanwhile the lead that came in on Friday at six goes unanswered until Monday, and that is the single most expensive habit I find in small businesses, year after year.

Middle of funnel: passes, with a reviewer

The middle is follow-up, nurture, scheduling, and keeping the record honest as things move.

Frequency is high, the work is scriptable, and the owner is usually clear. The filter that gets tighter is the third one. A follow-up sent at the wrong moment costs you a correction. A follow-up that invents a delivery date, or repeats a discount nobody approved, costs you something you cannot take back. So the middle is agent work with a person reading what goes out, at least until the loop has proven itself over several cycles rather than one good week.

Bottom of funnel: fails three of four

The bottom is the negotiation, the exception on price, the objection that is not really about price, and the moment someone decides to trust you.

One out of four. I am not saying an agent cannot produce a competent negotiation email. It can, and that is precisely the trap: it will read well, and you will not find out it was wrong until the customer does.

What this looks like in a real week

A small service business, the shape I have sat with dozens of times. Contacts arrive from a form, from Instagram, from WhatsApp and from the phone. Nobody knows how many. Roughly a third get answered the same day, and the ones that arrive after hours are a coin flip.

The first agent does not sell anything. It watches the channels, replies within minutes to say a real human name will follow up and by when, asks the three or four qualifying questions, and writes the answers into one record. That is it.

What changes in the first month is rarely what the owner expected. It is not usually more sales, at least not yet. It is that for the first time there is a real number for how many new contacts arrive in a day, and a real number for how many were qualified out. That number is the thing you cannot buy, and it is what makes every later decision cheaper, including the decision of whether to hire.

Then, and only then, the middle: the follow-up nobody had time to send, with someone reading it before it goes.

The bottom stays yours. Not forever as a matter of principle, but until you have a written record of how those conversations actually go, which you do not have today.

The prerequisite nobody wants to hear

None of the above works if there is nowhere to write things down. I have made this argument on its own, because it deserves its own room: an agent does not remember, it looks things up, and a business whose real status lives in a WhatsApp thread gives it nothing to look up.

So if you read this and the honest situation is that there is no single place a contact gets recorded, the first project is not an agent. It is one page that says what a contact record must contain and who is responsible for it being true. That page takes an afternoon and it pays for itself even if you never build an agent at all.

Automating a funnel with no record does not organize your sales. It just produces the same confusion at a higher rate.

What this has to do with Holybiz

I did not learn this from agents. I learned it selling.

At Sizebay, the sales model was mine to prove: I made the first sales myself, with my own hands, before I taught two salespeople to make the next ones, and the only reason I could teach them anything was that I had written down what actually happened in those conversations. Before that, eight years inside Tupy, Latin America's largest foundry, building the systems the sale genuinely ran through, from the order to the tax book. And since 2019, accredited by SEBRAE, running the same interview through hundreds of small and medium companies.

The method has never changed: interview the operation, find out what the data already answers, build only what is genuinely missing. What is different now is that the thing you build at the end of it can be an agent instead of a spreadsheet. The judgment about where to put it is the same judgment it always was.