What a manager does by hand — and what can be taken away

The routine nobody counts

When a manager's workload is measured, people usually look at the number of calls and meetings. The work done by hand between calls stays in the shadows — writing up the outcome of a conversation, setting a reminder, moving a customer to another funnel stage. It does not look like a separate task, so it is neither measured nor reduced. Yet this is exactly what separates a manager who handles a lead within an hour from one who is drowning in deal cards.

In a live GetGate project a manager frees up about an hour of working time every day — precisely because what used to be done by hand is now done by the sales robot and the CRM connected to it. This is not abstract time saving in general, but the concrete result of one project. Still, it shows that the manual part of the work is not a trifle but a full hour a day that can be won back.

Further on we break down which operations exactly make up this manual work. And separately — what of it is taken over by automation and what still rests on a human being, because there is no replacement for them.

Writing into the CRM by hand — where the time goes

The classic picture: a manager calls a customer, talks to them, and writes all the important points into the deal card by hand. What the customer said about budget, when it suits them to be called back, what objections they had — all of it has to be recalled right after the call and transferred into CRM fields. If there are many calls in a day, some details are lost or jotted down hastily, and later it is unclear what was meant.

The problem is not the manager's laziness but that writing up is a separate action which takes time away from the next call. While they are typing text into the card, they are not talking to a customer and not preparing for the next conversation. During CRM implementation this part can be closed with task automation and telephony integration — then some of the data reaches the card without manual entry.

The calculation stands apart: how many products or services belong to the deal, which group they fall into, what of this has to be counted for the report. In the GetGate project five product groups are counted automatically — this too used to be manual work for the manager or the accountant. Automatic counting does not change the deal itself, but it removes an extra step between the conversation with the customer and recording the result.

Moving dates and reminders by hand

Another layer of manual work is moving deadlines and amounts into the CRM from memory. The manager sees that the customer mentioned a certain sum earlier, enters it into the card, pushes the date of the next contact back — and counts on the software to remind them about the deal at the right moment. This works as long as the manager keeps the context of every deal in their head. But the more deals there are, the more often this shifting is done by eye rather than by fact.

The same happens with distributing customers across funnel stages. The manager roughly senses who should be moved further out — not a refusal, but not a hot lead either — and moves them by hand, relying on their feeling from the conversation rather than on a clear rule. A mistake in this move is not visible straight away: the customer simply drops out of sight, and is remembered when they have already gone cold or left for another seller.

A properly built funnel removes part of this uncertainty: the rules for moving deals and the reminders are set once during CRM implementation, instead of being invented from scratch for every customer. It does not take the manager out of the process, but it does remove the situation where a decision is made by eye at the end of a long day.

The first reply to the customer comes too late

Some leads physically cannot be handled by hand in time — because they arrive when the manager is already off shift. In a live GetGate project 27% of leads arrive in the evening and at weekends, and previously they were answered only by the morning. For a customer who wrote at ten in the evening, a morning reply is already late: they could have found another seller or simply cooled off.

Here the manual work is not that the manager does something wrong, but that they are physically not there when the customer is ready to talk. The problem cannot be solved by hiring more people for a night shift for the sake of a fraction of the leads — for most companies that does not pay for itself. What is needed is a way to answer the customer at once, without waiting for the morning and without spending a separate salary on it.

This is exactly the part of the work where manual labour is not merely slow but impossible on a human schedule. The solution is not to make the manager work at night, but to answer the customer the moment they get in touch through about the sales robot, and in the morning to hand the manager a warm customer along with the history of the conversation.

What the sales robot takes on

The sales robot answers and qualifies customers in every channel where they write: Telegram, WhatsApp, Instagram, Avito, the website. It recognises voice messages and photos, remembers the history of the dialogue and does not start the conversation over if the customer comes back a day later. All of this runs through AmoCRM, so the customer card fills itself in as the conversation goes, rather than after the fact by the manager's hand.

In a blind comparison, in 80% of cases the assistant's answer was no worse than a human's — this is the result of our own check on real correspondence, not a general assessment of the market. This is not about every possible dialogue, but about the standard questions a conversation with a customer usually starts from: what the service includes, how much it costs, when someone can come. These are precisely the questions that most often waste a manager's time — answering them requires no experience, only the patience to repeat the same thing for the hundredth time.

When the customer ripens into a substantive conversation — ready to discuss details, to bargain, to decide — the robot hands them over to the manager. This is the key boundary: the robot takes over first-touch handling and qualification, it does not replace the sale as a whole. In part of the deals in such a funnel human involvement is reduced to a minimum, but it does not disappear altogether — complex cases still call for a live conversation.

What stays with a person

Not everything in sales can or should be taken over by automation. A conversation where the customer hesitates and needs to be heard, bargaining over non-standard terms, working with objections that do not fit a script — that still rests on the manager. The robot copes well with repetitive questions precisely because they repeat; where every case is unique, a human is needed.

Part of the work with an already warm customer also remains: bringing them to payment, agreeing the details of the contract, explaining the nuances you cannot write into a script in advance. Here the manager's value is not the speed of typing but understanding the context of this particular customer and being able to decide on the spot. It is for this part that a manager's time is worth freeing from routine — so that it goes where a decision really does need a human, and not into recalling and entering a date in a card.

The responsibility for keeping the deal from stalling stays here too. Automation reminds and suggests the next step, but the final decision — to call again, to change the offer, to walk away from the customer — is taken by a person. It is this division that makes the pairing of robot and manager work: one covers volume and speed, the other complexity and responsibility.

Quality control shows where a human is still being lost

Even after first-touch handling has been taken off the manager, some calls and conversations still slip past any control — simply because a head of department physically cannot listen to every conversation in a day. AI listens to calls and correspondence, scores them against a checklist, looks for cases where the customer has in fact left although the deal is formally still open, and calculates a score for each manager. This is not a replacement for human control, but a way to see what would otherwise go unnoticed.

It also becomes visible here where the manager still does by hand what ought to be automated: moving dates by eye, forgetting to write up the outcome of a conversation, dragging out a reply to a customer. Such things are not always noticeable in the moment, but they accumulate in the statistics on quality control. The head of department gets not a feeling that something is off, but concrete cases that can be pointed to and discussed.

Pairing this with sales analytics adds another layer: plan against fact in real time, a breakdown by channel and by manager, alerts when figures sag. Together with quality control this turns scattered observations into a picture of exactly where customers are lost and how much time managers spend on what did not need doing by hand at all.

How this is introduced, step by step

It all starts with a review of the customer's sales department: we look at where exactly money and time are lost — at which funnel stage, in which channel, in which routine operation. Without that review it is unclear what to automate first: in one company most of the time goes into first replies, in another into moving deals between stages. There is no general recipe, there is the concrete picture of a concrete sales department.

Next the CRM is assembled around the customer's process, the robot is trained for the niche, and the channels where customers actually write are connected. This is not a one-off setup for the future but a working tool that managers start using straight away. In parallel the rules are talked through: what the robot now does and what stays with a human — without that division confusion arises over who is responsible for what.

After launch, plan and fact are watched, the funnel is fine-tuned and what worked is scaled up. This is a continuous process, not a one-time implementation: the market and the customers change, and the settings need tightening. If work on the sales department itself is needed in parallel — scripts, rules, analysis of the reasons for refusals — that is a separate direction, described on the page building a sales department.

Where to start the conversation about your own routine

Before introducing anything, it is worth honestly looking at a manager's day yourself: how much time goes into talking to a customer, and how much into writing up the outcome, moving a date, recalling what was promised. Often this is not recognised as a separate item of time spent until you start counting it step by step. The same question is worth asking not about one manager but about the whole department — the routine is similar for everyone, even if each of them describes it in their own words.

Not all routine is taken over by automation equally: some of it is taken by the sales robot, some by a properly built CRM, and some only becomes visible through quality control of calls. That is why the conversation usually starts not with choosing a tool but with going through the specific operations that are done by hand today. After that it is clear which of the six directions — sales robot, CRM, analytics, quality control, building a sales department or client acquisition — are needed first.

More breakdowns of situations like this are in the blog: there you will find concrete cases from practice, without general reasoning about automation in the abstract. If your sales department has a similar routine, the conversation about what to take out of it is worth starting with a concrete review rather than a ready-made solution.