How to deliver a product well — the Xiaopai implementation handbook

You cannot talk about product delivery without talking about one role: the implementation consultant.

The job asks for real professional skills — project management, business analysis, software, system integration, training. It also asks for softer things: communicating well, solving problems, and caring about the customer.

Because so much of it is personal "soft skill", there is very little material to learn from. What does exist is hard to reuse, since industries and products differ so much.

Xiaopai is a lightweight production-management product, and our goal for delivery is three words: fast, accurate, decisive. This handbook exists to carry that idea from the top all the way down.

It is written for people who, like me, have just started as implementation consultants, or who are curious about the role. If you are a veteran, it may be too basic — treat it as a short reference. If you are new, it may clear up some of the confusion you run into on site.

The handbook has three parts:

• Principles: the basic ideas — what an implementation consultant is, why implementation matters, and what makes a good one
• Cases: a few representative customers we met during implementation, with successes and failures
• Practice: how a consultant maps the business process and uses the system to solve the customer's real pain

#Principles

#1. What is an implementation consultant?

Many people think a consultant sells time: teach the customer to use the system, close the project, wait for the commission.

That is only part of the work. Stop there and you are a "system trainer", and the customer's hidden problems pile up. At best they call you every day asking how to use things. At worst the business process was never mapped properly and the system sits unused. Once you are in that position, you are entirely on the back foot.

My view: a company that hires a consultant pays 10% for time and 90% for experience.

That is why senior consultants cost more. They solve problems quickly and professionally, and can even foresee the problems a company will hit as it grows, and warn about them in advance.

So the real goal of an implementation consultant is to understand what the customer needs, analyse the root cause behind it, draw on past experience, and give the solution that fits this customer best.

#2. Why do implementation at all?

"Three parts software, seven parts implementation" — I do not fully agree, but delivery is undeniably a key stage in a product's life.

• It is the channel for hearing customers. For a B2B product, feedback is the main source of iteration. You cannot build something customers love by guessing from an office.
• It is the hinge of the customer relationship. A good consultant ties the customer's business to what the product produces, and builds habits that make the product stick. Take the work-order traveler: most manufacturers design their own, and it follows the material through every step from part to finished good. Print it from the system and the customer says, "Not bad." Print it in exactly the layout they wanted and they say, "Wow, your system can do that?" More on system outputs and deliverables in the Practice part.
• It is driven by value. That is what keeps us at something, and makes us want to do it well. Value can be material or it can be a sense of meaning. If you want to become a well-rounded person, enjoy working with people and like to be stretched, this is a good role. Years from now someone may ask, "Of all the jobs, why implementation?" and you will answer, "Because to me it was worth doing."

#3. What makes a good implementation consultant?

• An empty cup. Stay humble. Customers are the best teachers, and the "difficult" ones are the strictest and most useful. Do not lose your footing with them — ask, in the spirit of learning. Be even more alert with customers who seem easy: dig for what they really need, or the gaps in your plan will surface after go-live.
• Communication. State your ideas clearly, listen to the customer, and feed back to the team promptly. When passing on a request, ask "why" a few times and find the real need. Never become a mere messenger for requirements.
• Project management. Coordinate the moving parts and keep control at the key milestones. If something blocks the project, borrow outside force — escalate and let the boss apply pressure — and do whatever it takes to bring the project to a close on schedule.
• A view of the whole. It is easy to sink into the details of one project. When that happens, step back and look from higher up: let small things go, and keep a high-quality close as the goal.
• Ownership. Be responsible for the project and put the customer at the centre. If a customer raises something on a day off and you cannot solve it right away, still reply quickly and say the team is on it. Details decide outcomes, and customers gladly pay for good service.

#4. What does "go-live" mean?

Go-live is not the customer signing a go-live confirmation. As the vendor's consultant, we need the customer to use the system deeply. As the product evolves, a consultant should spot the features that deepen usage and teach them promptly. Only when the customer's daily production can no longer easily be separated from the system has it truly gone live.

#Cases

These are cases from Xiaopai's implementation work — some successes, some failures. The successes carry experience; the failures carry hard lessons.

#Case 1: a custom automation-equipment maker (Customer A)

Customer A has been in business for over twenty years, covering equipment repair, automation engineering, energy-saving technology, automation product sales and machine building. It mostly makes non-standard, custom equipment. An earlier ERP rollout had failed and its level of digitalisation was low.

1. Pain points

1. Materials

• No agreed material-coding rule; key materials were not managed separately
• No material-issue process, so materials were often diverted and other machines ran short during production

2. Purchasing

• Purchasing could not follow up in time, and supplier delays were not recorded, so plans drifted away from production's schedule

3. Production

• Labour hours and progress were managed loosely; managers estimated production time from experience
• Lines often stopped to wait for materials
• Products were non-standard and every machine was assembled differently, so new staff often assembled them wrong

2. First implementation

1. Materials

• Set a coding rule: capitalised initials of the product plus a serial number
• Treat each machine as a project with its own virtual warehouse; all stock movements happen there
• Turn on batch management, with one code per item for key materials, strictly following the flow: purchase receipt, production issue, production receipt, sales shipment

2. Purchasing

• Add custom fields to purchase-order lines — arrival time, current status, delayed or not, reason, logistics — so buyers record them and the production lead can plan around arrivals
• Anyone can check purchasing status without phoning around, and the data also helps improve the supplier list

3. Production

• Manage the machining department separately by piece-rate reporting, to quantify progress
• Create a virtual step, "staff work", to record miscellaneous daily tasks

3. Second implementation

1. Materials

• Add a "bin" field matching where the material physically sits
• Compute material requirements from the BOM and issue purchase plans in one click
• Drop the virtual warehouses and batch management

2. Production

• The production lead issues both finished-goods and semi-finished work orders: machining issues semi-finished orders from the BOM and reports quantities made that day; assembly issues finished-goods orders and reports a percentage to show assembly progress

3. Warehouse

• After importing items, add opening stock in bulk with an adjustment document
• Receive against purchase orders
• Issue materials against work orders
• Receive finished goods against work orders

4. Industry terms

1. Machining: removing material precisely with machine tools to shape raw stock; done manually or by CNC.

2. Complete machine: the fully assembled machine.

3. Non-standard: proven equipment that has been fixed into a design and a series is a standard product. Anything produced, modified or made to order from that base at a customer's request is non-standard.

4. X, Y, Z axes: CNC motion is described with a left-handed Cartesian system. The axis parallel to the spindle is Z; rotations about X, Y and Z are A, B and C.

5. Five-axis machining: beyond X, Y and Z there is a rotary axis and a tilting axis. The rotary axis can turn a full 360°; the tilting axis only swings within a range, such as ±90°.

Its drawback: compared with 3+2 positioning the spindle is less rigid and precision errors are larger, so it is unsuitable in some cases — for example when the tool is too short or the holder too large to avoid vibration at any tilt.

Its strength: no special fixtures are needed, which lowers fixture cost; fewer re-clampings improve mould accuracy; and a longer effective cutting edge reduces cutting force, extends tool life and lowers cost.

6. Finishing tool: a tool for finish cuts, usually a spring finishing tool on a lathe.

5. What we learned

The first implementation was in September 2022. The customer needed almost three months to get comfortable with the system, and even then the result fell well short of expectations.

Customer A is a typical non-standard machinery maker, with complex and unstable BOMs. A company like this inevitably has two kinds of work orders: semi-finished orders that count machined parts, and finished-goods orders that track assembly of the whole machine.

Given what the customer could absorb at the time, the first round only covered finished-goods orders. The planned quantity was the number of machines ordered; each step's owner estimated assembly progress and reported a percentage; and the production lead kept one long-running order to log everyday work.

In early April of the following year, once the customer was fluent with the system, we ran a second round. It dealt with the leftover problem of semi-finished orders: the customer can now choose a finished-goods or a semi-finished order as the situation requires, and production progress finally became transparent.

#Case 2: a plastic-packaging maker (Customer B)

Customer B is a young packaging company making plastic packaging and other plastic products, films, paper-plastic bags, textiles and paper goods. The owner supported digitalisation and had tried a similar production system before, without getting it live.

1. How they worked before

Production was managed loosely through shared online documents. Sales orders, scheduling and reporting statistics all sat with the production lead. Only some steps were paid by piece; the rest by hours. The warehouse only handled raw-material receipts and finished-goods shipments.

2. Pain points

1. Orders

• Once an order was released, nobody could see progress in real time; it took rounds of asking, which was slow

2. Production

• Work orders moved chaotically; they wanted every worker to see the process requirements on a device

3. Warehouse

• Stock was kept by hand; they wanted the system to show the balance of every item

3. The plan

1. Orders

• Use the order-progress report to see overall progress per order

2. Production

• Fix a coding rule: capitalised initials of the item name plus a serial number
• Create work orders by picking items from the order, linking order and work order
• Design a work-order traveler and attach the process drawing to the work order
• Workers scan the code on the work order to report

3. Warehouse

• Receive finished goods by picking from the work order, linking receipt and work order
• Ship by picking from the order, linking shipment and order
• Use the stock-by-warehouse report to see every balance

4. Industry terms

1. Odd lot: the loose quantity left over when a shipment does not fill a whole packing unit.

2. Blank bag: a bag with no printing at all, which can be sold as a semi-finished good downstream. Printing and packaging is mostly made-to-order, and the blank bag is one of its few standard items.

3. Lamination: composite packaging joins two or more materials through one or more dry-lamination passes to give a specific function — usually a base layer, a functional layer and a heat-seal layer. The base layer mainly provides appearance, printing and moisture resistance.

Note: in Chinese it sounds very close to the name of another process, box gluing, and the two are easily confused.

5. What we learned

The first implementation was in February 2023. After a month of use the gaps in the plan showed. In June 2023 the owner said that after nearly half a year of trying, it was not a fit, and they would stop using Xiaopai for now.

1. On the first site visit we did not stress how much material coding matters. After a while the customer found that a standard item like the blank bag had ended up with several codes, and could not be received into stock at all.

2. The shop-floor workers were mostly older and found reporting hard. Even with team leaders reporting on their behalf, it was a struggle in practice.

3. The warehouse had to receive finished goods from what was on the traveler, but with nobody reporting on the floor, the traveler never reached the warehouse — so nothing could be received.

#Case 3: a smart-equipment maker (Customer C)

Customer C develops smart equipment, motor-manufacturing equipment, electromechanical devices, motors and optoelectronic and sensing products. It already ran a mainstream ERP for purchasing, sales and inventory, so it had a digital foundation, and the owner had clear ideas about making the shop floor smarter.

1. How they worked before

Production orders were released from the ERP. The floor was highly automated: one line had more than 30 machines, most with data-acquisition devices. But production was not managed down to the step, so progress could not be tracked.

2. Pain points

Orders:

1. Too few orders were being delivered on time

Production:

1. Paperless reporting was needed, including workers' scattered small tasks
2. Abnormal situations should raise alerts promptly
3. Reporting needed a "pause", to measure actual machine running time
4. Once goods reached the customer's site, assembly progress could not be controlled

3. The solution

Orders:

• Use the dashboard's management metrics to watch average order cycle time and on-time delivery rate

Production:

• Configure data capture per step: workers can upload photos and descriptions from their phones when reporting
• Configure notification rules with triggers and recipients: when something abnormal happens, the right people are notified in the system and on WeChat
• Add custom fields to the report form — paused or not, description, reason: when progress is abnormal, the worker selects "pause", gives the reason and uploads a photo for the production lead to review
• Add a "work-order type" field to work orders: when equipment must be installed at the customer's site, the production lead opens a finished-goods order marked as an installation order; installers report as they go, and the lead sees assembly progress through the order

4. Industry terms

1. Made-to-order: another word for "customised"; the Chinese term used by this customer comes from Hong Kong and Taiwan and is less common on the mainland.

5. What we learned

The first implementation was in late October 2022 and the system went live in mid-November, with a single day on site. At a follow-up in late November the customer was happy with the product and the team, and decided to add the warehouse module.

1. The company is a subsidiary of a listed company and already had a digital foundation. Its contact person was very efficient at preparing opening data and learning the system.

2. To make the product stick we normally suggest printing travelers from the system. But for customers with long production cycles and few work orders, a traveler can become a nuisance. Adjust to the actual situation.

#Practice

For a new consultant, establishing the first impression of a "professional consultant" on site matters a great deal. It makes projects close faster, and to some extent it also binds the customer more tightly to the product. This part covers two things: controlling the key milestones, and deliverables and system outputs.

• Controlling the key milestones: from taking on a project to closing it, how to stay in control at each milestone
• Deliverables and system outputs: how implementation can tie the customer and the product closer together

#Controlling the key milestones

Xiaopai is lightweight, and a consultant usually runs several implementations in the same period. So you can neither get lost in detail and lose speed, nor rush to close and ignore the logic between business processes. Balancing the whole and the parts is a required skill. This section covers the key stages of a project, and what to do when one of them stalls so the project still goes live on time.

1. Taking on the project

When an implementation task arrives, do the following:

1. Read the customer information sheet, then talk to the salesperson. Cover at least: how the key people think about management, the current production model, the state of digitalisation, the pain points, headcount and general skill level.
2. Share the attendee confirmation sheet in the customer group, collect contact details for the decision-maker and the contact person, and book the on-site date.
3. Send the company name, people's names and phone numbers to operations so the account can be opened, and post the details in the customer group.
4. Send the pre-go-live preparation sheet, and insist the customer prepares the base data before the visit — otherwise the visit will be far less efficient.
5. For longer projects, prepare a kickoff deck tailored to the customer beforehand.
6. Print the coaching sheet, training confirmation and go-live confirmation in advance; have them signed on site and bring them back for filing.

2. On site

Kickoff and discovery: most managers in manufacturing are used to the way things are, and changing a habit is hard. Bring the owner into the kickoff so people take it seriously and feel less daunted. Whether to hold a kickoff depends on the subscription tier, the people and the environment.

• Subscription tier: for longer projects, gather the relevant people for a kickoff; otherwise start implementing directly
• People: if the organisation is muddled, roles overlap heavily and responsibilities are unclear, a kickoff will not achieve much
• Environment: if the site has no place to hold a meeting, go straight to discovery and stay flexible

Driving the project: you will meet every kind of customer. Some learn slowly and need patience; some raise very professional requests or even find flaws in the system. Whoever it is, keep the initiative. Some common situations:

• Slow to learn but willing: usually no digital foundation, and Excel is their daily tool. Changing habits is extremely hard. Write this person a dedicated manual, down to the steps of each function and the meaning and purpose of every field.
• The owner wants the system, middle managers resist: check whether the plan is sound and really makes their work more efficient. Listen to the front line, collect and assess their needs. If neither the product nor the plan is the problem, go to the owner to push it through.
• The owner is watching, middle managers resist: usually we have not met expectations. Do not panic and rush. Focus on one or two problems and solve the real pain. If you can, show a before-and-after comparison so the value is tangible.
• The customer is busy and leaves you waiting: quite common. Use the time to organise material or map the process. But past a few hours, tell the contact person or the owner that implementation days are a cost they are paying for, and ask them to make time.

3. After go-live

Support: once the go-live confirmation is signed, the project is closed, and later questions are mostly handled online. Some common situations:

• A question outside working hours: reply as soon as you can — speed matters enormously to the experience. A standard reply works, such as: "Your feedback has been passed to the product team. If it makes it into a release, I'll let you know right away."
• Sharing releases: after each release, post the changes in the customer group. If it includes something this customer asked for, mention the people concerned, and where needed run a short online session to get them up to speed.

#Deliverables and system outputs

1. System outputs

The point of system outputs is to keep the customer's business from drifting away from the product. When planning for a customer, have their business documents printed from the system. Building the habit is what makes the product stick.

Documents that should be planned into the system:

• Standard edition: work-order traveler, daily data report
• Standard edition plus warehouse: work-order traveler, receipt voucher, shipment voucher, daily data report

Work-order traveler (machinery example)

The traveler is indispensable in production. Guide the customer to print it from the system. If front-line staff struggle with the system, plan for team leaders or the production lead to report on their behalf rather than having workers do it themselves.

Settlement note (printing and packaging example)

Shipment or settlement notes are usually printed in triplicate and handed to different people — warehouse, finance, gate — who each use it to settle, count stock or release goods.

Daily data report

Push data tailored to each role's interests. It matters a great deal for building habits and stickiness.

2. Implementation deliverables

The point of deliverables is to give implementation a sense of ceremony: the customer should recognise the professional service, and accept the implementation fee.

Deliverables by stage:

• Online stage: attendee confirmation sheet, pre-go-live base-data sheet
• On-site stage: coaching sheet, go-live confirmation, sign-in sheet, business process diagram, role manuals, material-coding rules, implementation summary

Attendee confirmation sheet

• Confirms contact details for the owner and the project contact, so accounts can be opened
• Fixes the schedule so the customer can prepare

Pre-go-live base-data sheet

• Has the customer prepare data during the online stage, making the visit more efficient
• Before the visit, the consultant helps check the base data in the sheet

Coaching sheet

• Signed by the customer's contact after the on-site work
• Externally it tells the customer what was covered; internally it is filed so managers know what was done

Go-live confirmation

• Signed by the customer's owner after the on-site work
• Externally it confirms the project is closed; internally it is filed as evidence if a dispute arises later

Sign-in sheet

• Signed by everyone who attended
• Externally it makes attendees take the training seriously; internally it is filed as evidence if a dispute arises later

Business process diagram

• After the visit, map the customer's process onto Xiaopai's functions and share the diagram

Role manuals

• After the visit, write a manual for each role based on the plan and share it as a PDF

Material-coding rules

• Agreeing material codes is the most important step in implementation and must be settled with the customer during it
• In machinery, for instance, different codes carry different meanings and help with quick lookup, stock counts and period settlement
• If the customer has no concept of coding, generate codes from serial numbers

Implementation summary

• After the visit, post what was done that day in the customer group
• Externally it tells the customer what was covered and what comes next; internally it keeps managers informed

Note: this article is adapted from an internal handbook I wrote at the time. Customer names and exact dates have been anonymised, and the original flowcharts and document samples are not included because they contain customer material.