Customer Experience Handbook: Structure and Metrics

Why your business needs a customer experience manual
The customer experience is not improvised. When each attendant decides on their own how long it takes to respond, to whom to transfer and how to record the history, the operation becomes a series of individual initiatives without continuity. The customer repeats information, waits without knowing why, and leaves the conversation unresolved. The customer experience manual transforms intentions into concrete rules and converts those rules into system configuration, so that the operation works in a predictable and measurable way.
This guide presents the complete structure of an operational manual: the seven mandatory components, such as defining deadlines by channel and category, scaling flow between areas, responsibilities, rules for transshipment between AI and human, and compliance metrics. At the end, you will know how to convert the manual into a system configuration and monitor compliance with indicators that reveal bottlenecks before jeopardizing the experience.
The seven mandatory components of the manual
A well-structured customer experience manual covers everything from contact input to post-resolution follow-up. It does not replace the service manual (which deals with posture and scripts), but defines what happens before and after the conversation: channels, categories, managers, deadlines, scheduling, and metrics. These seven components ensure that the rule doesn't depend on someone reminding.
1. Input channels and response expectation
List all active channels - WhatsApp, Instagram, chat on the site, e-mail, telephone - and define the expected first response time for each one. Customer expectations vary depending on the channel: 36% expect a response in online chat within 1 minute, 26% within 5 minutes; in messaging apps, 19% expect a response within 1 minute, 32% within 5 minutes; in e-mail, 9% wait up to 1 minute, 16% within 5 minutes, and 48% accept up to 24 hours, [according to data from Octadesk] (URL0). These numbers show that ignoring the difference between channels leads to unrealistic goals or unfulfilled promises.
When the operation uses a [unified inbox] (URL1), messages from different sources enter the same queue but remain identified by channel. The SLA configuration respects the origin: WhatsApp with an alert in 5 minutes, e-mail in 4 hours. Ignoring the expectations of each channel transforms the speed of one into a problem for the other.
2. Demand categories and responsible sector
Not every conversation has the same operational weight. A question about operating hours is resolved in minutes; a faulty replacement can take days. Define the main categories - pre-sale question, order, delivery, exchange, billing, technical support - and assign each one to a sector. This division appears as a call field or tag in the service tool and guides distribution, priority, and deadline.
Categorization reduces the volume of random transfers and allows you to track where the operation consumes the most time. When all conversations are mixed up, the general indicator hides that the “delivery” category is 200% above the expected volume while “pre-sale question” answers on time.
3. Deadlines by category and channel
Define SLA (Service Level Agreement) for first response and for full resolution. The first response acknowledges the receipt; the resolution closes the call. Chat and WhatsApp require a response in minutes, but the solution can take hours or days if it depends on technical analysis or stock replacement. Separating the two deadlines prevents the customer from being left with no return while waiting for a definitive answer.
The SLA differentiated by category recognizes that not everything has the same deadline. A product question can be resolved within 2 hours; an exchange, within 48 business hours; a technical defect, within 5 business days. Record deadlines, business hours, exceptions, and how to report delays to the customer.
4. Scheduling and transfer flow
Define when and how demand moves from one area to another. Escalation occurs when the first responsible person is unable to solve it - due to lack of autonomy, information, or access to the system. The flow needs to be clear: from sales to after-sales, from after-sales to technical support, from service to financial.
Every transfer point is a risk of losing context. The manual must state that, when transferring, the attendant records the reason, what has already been tried and the deadline agreed with the customer. This prevents the next sector from restarting the conversation from scratch, frustrating those who have already explained the problem twice.
5. Transshipment rules between AI and human
Automation organizes input but doesn't solve everything. Define when the bot transfers it to the human attendant: when it detects irritation, when identifying requests outside the trained scope, after three attempts without resolution, or when the customer asks explicitly. 63% of consumers say that, for a robot to serve well, it needs to connect them to a human when necessary, [according to Octadesk] (URL2).
The context passage is the critical point. The human agent must receive the conversation history, the data collected, and the stage of the process. When that doesn't happen, the customer repeats everything and automation becomes a barrier. Platforms such as BRAND[ combine automations with human participation] (URL3) and preserve the full context when making the transfer.
6. Records and shared history
Every contact generates data. The manual must define what to record: reason, category, responsible, deadline, actions taken, and result. These fields feed reports and prevent the company from treating the same customer as unknown with each new conversation.
The history shared between channels is the basis of omnichannel service. When the customer sends a message on Instagram, calls the phone and sends e-mail, all three contacts must be connected to the same record. Without this, the company loses continuity and the customer repeats the problem on each channel.
7. Compliance metrics and review routine
Establish indicators that show whether the manual is being followed: first response time, percentage of SLA fulfilled, resolution at first contact, CSAT satisfaction, volume of transfers, and average service time. These numbers reveal bottlenecks before damaging the experience.
The quarterly review of the rules keeps the manual up to date. Monitor the indicators on a monthly basis and adjust deadlines, managers, and categories as new channels, new products, or structural changes emerge. Record the date and reason for each change so that everyone knows which version is valid.
How to set realistic deadlines per channel
Unrealistic promises destroy trust. The deadline must balance the customer's expectation with the team's actual capacity. Start with the history: how long the operation takes today to respond and resolve on each channel. That number is the baseline. From there, define the goal of gradual improvement and the deadline that will be communicated to the customer.
If the email takes 12 hours for the first response but the customer waits up to 24 hours, report “we respond within 1 business day” and work internally to reduce it to 8 hours. If WhatsApp takes 15 minutes and the customer waits until 5, report “we respond within 30 minutes” while organizing distribution and screening to reach 5.
Deadlines differentiated by complexity
Not every demand has the same operating cost. Simple questions - time, address, payment method - can be solved by bot or in minutes. Complex requests — credit analysis, default exchange approval — depend on other industries and take days.
Define categories with their own deadlines: immediate service (up to 30 minutes), short term (up to 24 hours), medium term (up to 3 business days) and long term (up to 7 business days). Communicate the deadline to the customer at the start of the conversation and send updates whenever there is progress or delay.
Clear Responsibilities by Demand Category
The lack of a definite responsible person snowballs calls between areas. The manual should list each category and the group or person that assumes the demand. When there is a question about who solves it, the rule is: those who receive it, register it and forward it with the full context to the correct sector.
Use service groups in the tool to separate sales, after-sales, support, financial, and other industries. The [BRANDinbox] (URL4) allows you to create queues and assign conversations automatically, based on channel, time, or keyword rules. This ensures that each demand reaches the person responsible without relying on manual recall.
Escalation for manager or specialist
Some demands go beyond the autonomy of the first attendant. Define when and how to scale: amount above X reais, cancellation request, formal complaint, technical error with no known solution. The escalation cannot be random; it requires clear criteria, deadline and notification to the person responsible for the analysis.
Scheduling flow: how to avoid empty transfers
Context-free transfer wastes time and irritates the customer. The scheduling flow organizes the passage between areas: where it leaves, where it goes, what needs to be registered and who accompanies it to the end.
Example of a faulty exchange flow: customer opens a call on WhatsApp → attendant collects invoice and photos → transfers for technical analysis with full registration → technician approves or denies within 24 hours → if approved, logistics schedule collection → attendant confirms appointment with customer → logistics executes → financial processes reimbursement → attendant closes call and applies satisfaction survey.
Each stage has a responsibility and a deadline. When demand moves from one area to another, the system records who transferred, when, and why. This history allows you to track where the process crashes and correct it before becoming a public complaint.
Transfer and Rework Indicators
Keep track of how many conversations are transferred, between which areas, and how many return to the sector of origin. A high return rate indicates that the transfer was premature or that information was missing. Also measure the time between receipt and the first action of the destination sector: delay at this point reveals overload or lack of clarity about priority.
Automation rules: when the bot answers and when it becomes human
Automation reduces repetitive volume, but requires clear limits. The manual should define the scope of the bot: opening hours, FAQ, basic data collection, order status. Other than that, the conversation moves on to the human attendant.
The most common transshipment triggers are: the customer explicitly asks to speak to the attendant, the bot does not find an answer after two attempts, the customer shows irritation or urgency, demand involves negotiation or case analysis. [BRANDallows chatbots by rules and by AI] (URL5), with configurable handoff and preservation of context when transferring.
Preserving context in the transfer
The worst mistake is making the customer repeat everything when exiting the bot. The human attendant must receive the complete history: messages exchanged, data collected, category identified, and expected next step. This transforms automation into an accelerator, not an obstacle.
Platforms such as BRAND[ offer unified inbox] (URL6) that record WhatsApp, Instagram, Messenger, Telegram, e-mail, and website conversations in one place, with preserved context between channels and between bot and human.
How to measure compliance with the manual
Writing rules doesn't guarantee compliance. The manual needs indicators to show whether it is being followed. The main ones are:
- First response time per channel: reveals whether the incoming SLA is being met.
- Percentage of SLA met by category: shows where the operation is delaying.
- Resolution upon first contact: indicates whether the attendant has the autonomy and information to close the demand.
- Volume of transfers by sector: points out screening flaws or lack of clarity about those responsible.
- CSAT satisfaction at the end of service: measures the customer's perception of the result.
- Reopened calls: Signals incomplete resolution or lack of follow-up.
Follow these numbers weekly by channel, by service group, and by category. The overall average hides problems: one fast channel may be compensating for another with days of delay. [BRANDoffers real-time analytics] (URL7) with reports by agent, channel, and category, and export to BI.
Converting the manual into a system configuration
The manual becomes an operation when the rules enter the system. Each demand category becomes a call field or tag. Each responsible person becomes a service group. Each deadline becomes a configured SLA. Each standard reply becomes a quick message template. When the rule is in the tool, it's not up to someone to remember.
Practical Setup Example
Category: Default exchange Manager: After-sales First Response Time: 30 minutes Resolution Time: 48 business hours Flow: Attendant collects notes and photos → transfer for technical analysis → technician approves within 24 hours → logistics schedule collection → attendant closes with research SLA in the system: 30-minute alert (first response), 24-hour alert (technical analysis), 48-hour alert (final resolution)
When the call exceeds the alert, the system notifies the person responsible. This anticipates the problem before it becomes a complaint. The configuration replicates the manual so that the execution is automatic.
Continuous review based on data
The manual is not a fixed document. Review the categories, deadlines, and those responsible on a quarterly basis based on the indicators. If a category has an increasing volume and a deadline has expired, reevaluate the flow or expand the team. If a transfer between sectors always takes time, review the escalation criteria or reduce intermediate steps.
Record all changes with a date and reason so that the team knows which version is valid. Report changes at an alignment meeting and update internal documentation while adjusting the tool.
Impact of the manual on reputation and conversion
Operational compliance reflects directly on customer perception. Companies with an “Excellent” reputation or RA1000 seal on Reclame AQUI have solution rates above 80% and respond to complaints within 6 days, according to [analysis of the Reclame AQUI Award] (URL8). Those numbers don't come from sympathetic scripts; they come from clear processes, deadlines met, and follow-up to the end.
Almost [half of Brazilians consult Reclame HERE] (URL9) before buying, and [70% avoid brands with poorly answered complaints] (URL10). Speed also weighs in: [85% of consumers say that quick response influences the decision to buy again] (URL11). The manual does not guarantee sale, but the lack of it increases abandonment.
Frequently Asked Questions
What is the difference between a customer experience manual and a service manual?
The service manual usually deals with the attendant's posture and scripts. The customer experience manual goes further: it covers channels, demand categories, managers, deadlines, scaling between areas, and metrics. In practice, the service manual is a chapter of it. Without deadline and scheduling rules, the best script doesn't prevent demand from being lost between departments.
How to place the manual inside the service system?
Convert each rule into a configuration. Demand categories saw call fields, managers saw service groups, deadlines saw SLA by category and channel, and standard responses saw quick responses. When the rule is in the system, it's not up to someone to remember. The document still exists as a reference, but the execution remains in the tool.
How to know if the team is following the manual?
Follow the indicators linked to each rule: first response time, calls resolved on time, resolution at first contact, and satisfaction after service. Analyze by service group and category, not just the overall average. A good average can hide a category where deadlines run out every week.
How often should the manual be revised?
Review the indicators on a monthly basis and, each quarter, the rules. It is also worth reviewing whenever a new channel appears, a new demand category, or a change in structure, such as the creation of a post-sales team. Record the date and reason for each change so that everyone knows which version is valid.
Does the tool have to adapt to the manual or does the manual to the tool?
The tool must represent the operation. Before choosing, check if it allows you to configure categories, statuses, forms, groups, and deadlines with the rules that the company already has. Adjustments to the manual may take place, but by process decision, not by system limitation. A good test is to configure two or three critical categories during the evaluation and verify that the deadlines and those responsible are as described in the manual.
Does the manual work with multiple service channels?
Yes. One of the objectives of the manual is precisely to organize the entry when the customer can arrive by WhatsApp, Instagram, chat, e-mail or telephone. Each channel expects its own response, but the category, the responsible party, and the resolution deadline follow the same rules. [BRANDomnichannel integration] (URL12) centralizes conversations from multiple channels into a single box, preserving the full customer context.
How to ensure that the manual is applied in peak demand?
Peaks reveal weaknesses. The manual must provide specific SLA for high volume periods, inform the customer of the deadline change, and activate additional features - such as automation for screening and prioritizing urgent demands. Follow the indicators in real time and adjust the distribution between attendants as the queue grows.
Does automation replace the manual or complement it?
It complements. Automation performs parts of the manual — screening, data collection, FAQ response — but does not replace the definition of responsibilities, deadlines, and scheduling flow. The bot needs to know when to transfer to human, and the human needs to receive the full context. Without the manual, automation becomes a barrier; with it, it speeds up the operation.
References
- [Customer Experience Manual: How to Set Up Your Operation Manual | Octadesk] (URL13)
- [Customer service: what it is, what it does and how to organize | Octadesk] (URL14)
- [WhatsApp API budget for seasonal campaigns | Omnichannel] (URL15)
- [How to Safely Automate WhatsApp Calls | Huggy Blog] (URL16)
- [Fixed number in WhatsApp Business: how to validate and when to migrate | Octadesk] (URL17)
- [URA of service: how to direct calls by sector | Omnichannel] (URL18)
- [AI agent vs Chatbot: what's the difference and which one to use? | Huggy Blog] (URL19)
- [Reclame Award HERE: how it works and what it measures in service | Octadesk] (URL20)
Share
Open Instagram and paste the link in your post/story.