Automating answers to repetitive customer enquiries
A large share of customer service enquiries are a handful of questions asked in many different ways: where is my parcel, is this size available, please send me an invoice, how do I return an item. The answer to each of them is known and always the same.
How much this process costs you a year
The sliders start from values typical for this process. Swap them for your own — the result recalculates instantly, no account and no email required.
This process costs you
Realistically you will recover from PLN 0 to PLN 0 a year
Here is the math: salary times the employer cost multiplier 1.2 divided by 173 hours a month gives the hourly rate. That rate times the hours in a year gives the cost of the process. From the recovery we take a band of 40% to 70%, because part of the time recovered on paper comes back as supervising the automation and handling exceptions. This is a deliberately conservative range, not a promise.
Where the hours go
Each of these enquiries means reading the message, checking something in a system, writing a reply and logging the case. Three to five minutes. At a hundred enquiries a week that is several to over a dozen hours, more than anyone assumes before they actually count.
The second cost is interruption. Repetitive questions arrive throughout the day and pull people away from difficult cases that need focus. The result is that difficult cases wait the longest, which is exactly the opposite of what should happen.
What automation does
The enquiry is recognised and, if it falls within a known set, gets an answer straight away, with real data pulled from the system rather than a vague reply. Order status comes from the warehouse system, availability from the online shop, the invoice from the invoicing software. That is the difference between a useful answer and a bot that sends people back to a form.
Anything the automation cannot recognise with enough confidence goes to a person along with the context: who is writing, what orders they have, what their previous case was about. The confidence threshold is set deliberately, and it decides whether customers get good service or end up fighting a bot.
What it costs and when it pays back
Implementation usually falls between 3,000 and 10,000 PLN. The price depends on the number of enquiry types and the number of systems the answer data has to come from. Handling order status questions alone is cheap; a full set covering returns, complaints and invoices sits near the top of the range.
This is the fastest payback on the list, two to five months, because this process takes up the most hours of all six. The recovery rate, on the other hand, is lower than for reporting: some enquiries will always go to a person, and that is how it should be.
Contractors working with these tools
This process is most often built on the tools below. Each link opens a catalog narrowed to contractors who've declared it — rates, availability, and ratings from settled contracts.
Frequently asked questions
Usually 3,000 to 10,000 PLN as a one-off, within the automation range of 2,000 to 20,000 PLN. On top of that come the running costs of a language model, if the solution uses one. They depend on the number of enquiries and are worth agreeing before implementation, not after the first bill.
Frustration comes from a bot that cannot answer and cannot pass the case on. A sensibly built solution answers only what it recognises with confidence and hands the rest to a person along with the context. The confidence threshold is set deliberately and can be adjusted after the first few weeks.
Not if the answers come from your systems rather than from the model's general knowledge. This is the key difference when pricing: a solution that answers from a knowledge base and from order data costs more than one that generates text freely, and it is the only kind suitable for customer service.
It depends on how repetitive the questions are in your industry. The model we use on this page cautiously assumes 40% to 70% of the time is recovered, and that is a sensible starting point. A more precise figure can only be given after reviewing real enquiries from the past few weeks.