Most businesses considering automated customer service are working from assumptions rather than their own data. This engagement changes that — we look at what your customers actually ask before drawing any conclusions.
← Back to Marsh Crest FieldAfter four weeks, you will have a categorised breakdown of your customer enquiry volume — which questions have stable answers, which require human judgement, and what proportion could reasonably be handled without a person involved.
That final report includes a recommendation. Sometimes that recommendation is to proceed with automation. Sometimes it is to leave the process as it is. Either way, you will have looked at your actual data rather than guessing.
Categorised enquiry view
Your existing contact records sorted by enquiry type, volume, and complexity — built from your data, not from category assumptions.
Proportion estimate for automated handling
A specific figure — what percentage of your contact volume could be handled without human involvement, with the reasoning shown.
Written recommendation
A clear statement of whether to proceed, and if so, what scope makes sense. Including the case for doing nothing.
A lot of businesses arrive at automation decisions through a combination of vendor demonstrations and general industry noise. Someone attends a conference, reads a trade publication, or hears that a competitor has deployed a chatbot — and the conversation starts from there.
The trouble is that customer service enquiries vary enormously between organisations. One company's contacts are 70% routine status updates; another's are almost entirely complex, judgement-heavy complaints. The same tool performs very differently across those two situations.
Without looking at your own records first, you are essentially making a purchasing decision based on how automation works at some other organisation — which may or may not resemble yours.
There is also the customer side to consider. Customers who would still need a human — because their situation is unusual, or because they are in distress — end up in an automated system that cannot help them. That outcome is not neutral. It costs customer relationships and generates additional contacts that then require more staff time than the original enquiry would have.
The work begins with what you already have — your contact history, ticket logs, or correspondence archive — not with a product selection.
How enquiries arrive, who handles them, what types of responses are given. We map this from your records without disturbing it.
Enquiry types are sorted by stability of answer, frequency, and whether human judgement is genuinely needed. Volume figures are calculated.
Any recommendation to proceed includes a clear structure for which contacts remain with people and how automated responses would be monitored.
Software vendors assess your situation in order to sell you their product. Their interest is in finding reasons to proceed. Our interest is in giving you an accurate picture — which sometimes means finding reasons not to proceed.
We do not have a product to sell alongside this engagement. The output is a document. You can take that document to any vendor, or to no vendor at all.
Each phase has a clear purpose and a defined output. You will know what is happening and why at each stage.
We collect a sample of your contact records — email threads, ticket logs, chat transcripts, or whatever format you use — and carry out an initial pass to understand the overall distribution. No system access is needed; an export is sufficient.
Enquiry types are labelled and grouped. We identify which questions have answers that do not change, which depend on account-specific data, and which require a person to interpret the situation. Volume percentages are calculated for each group.
We look at what happens to customers whose enquiries would fall outside an automated scope. How many would there be? What do they tend to need? Is there a clear handover path? This part is often skipped in vendor assessments and is usually where the risk sits.
The full findings are written up and presented. The recommendation may be to proceed with a defined scope, proceed with conditions, or leave the current process unchanged. A walkthrough is included so you can ask questions about the methodology before we close.
This is the total cost of the engagement, agreed at the start. There are no additional fees for data complexity, extra meetings, or findings that take longer to work through than expected.
The fee is the same whether the recommendation is to deploy automation or to leave things as they are. We do not have a financial interest in which direction the analysis goes.
¥31,000 JPY
Four-week fixed engagement
Analysis of your existing customer contact records — no new data collection required
Categorised enquiry breakdown with volume percentages per type
Customer impact section covering the contacts that would still require a person
Written recommendation — including where the recommendation is not to proceed
Walkthrough session with the team at the end of week four
All documents retained by you — no ongoing access or licence required
We look at whether the correct response to a given enquiry type is consistent across customers and time periods. Stable answers are candidates for automation. Answers that depend on context, account history, or discretion are not.
Automating a type of enquiry only matters if it appears in meaningful volume. We calculate what each category actually represents in your contact mix — not what seems plausible.
For any realistic automation scope, some contacts remain outside it. We document who those customers are, what they tend to need, and how a human path would work alongside the automated one.
What to expect, honestly
A typical engagement finds that somewhere between 30% and 60% of contact volume suits automation. What sits in that range for your organisation depends on your specific enquiry mix, which is why the analysis is worth doing. Some organisations find the proportion is too low to make a deployment worthwhile. That finding has value too — it saves the cost and effort of a deployment that would not deliver much.
We do not have a product on the other side of this engagement. If the analysis suggests automation is not a good fit for your contact mix, that is what the report will say.
The category definitions, volume calculations, and decision logic are all documented in the report. You are not asked to accept conclusions without seeing how they were reached.
If you are not sure whether this engagement fits your situation, get in touch. We will discuss what you have and what you are considering — without any obligation to proceed.
You send us a message
Describe your customer contact situation briefly — what channels you use, roughly how many contacts you handle, what feels like it might be a candidate for automation.
We respond within two working days
We read each message carefully and respond with specific questions or a suggested next step. No template replies.
We agree scope and start date
If both sides are comfortable, we agree the data formats we need, confirm the start date, and begin week one. The scope is set in writing before any work starts.
Four weeks later, you have a report
The full categorisation, proportion figures, customer impact section, and recommendation are delivered. You walk away with a document you keep regardless of what you decide to do next.
If you are considering whether customer service automation might fit your organisation, the assessment gives you an honest answer grounded in your actual contact records. Get in touch to discuss what you have.
Send us a messageOther services from Marsh Crest Field
Setting up search across your procedures, manuals and archives so staff find existing material instead of recreating it. Suited to organisations with a decade or more of accumulated documentation.
Learn more →Introducing assisted drafting into your marketing or communications work, with defined review stages and clear rules on where it is not used. Includes style guide encoding and reviewer training.
Learn more →