How to Build a WhatsApp Business System: Architecture, Phases and Costs
There is a difference between using WhatsApp and running a business on it. The first is a phone with a green icon. The second is a system: one business number that every customer recognises, several staff working from one inbox, automated flows that capture orders and bookings as data, payments that confirm themselves, and records that survive when a salesperson resigns.
Most Nigerian businesses drift into the second without ever designing it. This guide is the design: what the components are, what you must decide before spending money, how the build phases run, and what it costs to do properly.
What a WhatsApp business system actually is
A WhatsApp business system is the combination of a verified business number on the WhatsApp Business Platform, software that lets a team handle conversations from one place, automated flows that collect structured information, and integrations that connect those conversations to the systems where your business actually keeps money, stock and customer records.
The defining characteristic is that conversations produce data. In an unsystematised WhatsApp operation, a completed sale leaves behind a chat thread. In a system, the same sale leaves behind a customer record, an order with line items, a payment reference, a delivery status and a support history.
What it is not. It is not simply installing the WhatsApp Business App, and it is not buying a chatbot. Both can be components; neither is a system.
The six layers of a WhatsApp business system
| Layer | What it does | Typical components |
|---|---|---|
| 1. Identity | Establishes who the business is on WhatsApp | Verified Meta Business account, business verification, display name, profile, catalogue |
| 2. Inbox | Lets humans work together on one number | Shared inbox with assignment, internal notes, labels, handover |
| 3. Automation | Handles the repetitive parts | Menus, interactive flows, quick replies, approved templates, routing and escalation rules |
| 4. Data | Remembers customers and transactions | CRM or database holding contacts, orders, bookings, tickets, consent records |
| 5. Integration | Connects WhatsApp to the rest of the business | Payment provider, stock or POS, delivery partner, website, accounting |
| 6. Reporting | Shows what is happening | Response times, conversion, order volume, category mix, agent workload |
Build them in order. A business that buys layer 3 without layer 4 automates itself into a slightly faster version of the same chaos.
Eight decisions to make before you build
These decisions determine cost, timeline and vendor choice. Settle them before requesting quotations.
- Which number will be the business number? It becomes the identity on your packaging, vehicles, website and Instagram bio. Changing it later is expensive in lost recognition.
- How many people will handle conversations, and in what roles? This decides your inbox licensing and your routing rules.
- What must a conversation produce? An order, a booking, a support ticket, a lead record, or several. This defines the data layer.
- Which systems must WhatsApp talk to? Payments certainly; possibly stock, delivery, accounting and your website.
- Who owns the customer data, and where is it stored? A vendor-hosted inbox means your records live with the vendor. Decide whether that is acceptable.
- What are your hours and service levels? These drive away messages, escalation rules and staffing.
- What is your consent and retention position? Under the Nigeria Data Protection Act 2023 you need a lawful basis, notice and a route to opt out. Decide how consent is captured and recorded.
- What does success look like in numbers? Response time, order capture rate, reduced reconciliation time. Without this you cannot judge the build afterwards.
Choosing your route: BSP, inbox product or Cloud API build
| Route | What it means | Best for | Trade-offs |
|---|---|---|---|
| WhatsApp Business App | Free app, no API | Solo operators | No team inbox, no scheduled sequences, no integrations |
| Off-the-shelf inbox on the Platform | A vendor connects your number and gives you an inbox with automation features | Teams wanting speed and low upfront cost | Monthly USD licensing, limited custom logic, data hosted by the vendor |
| Business Solution Provider plus automation tooling | A BSP manages the Platform connection while flows and campaigns are configured in their tool | Businesses needing templates, campaigns and flows without custom software | Vendor lock-in; verify Nigerian payment and support coverage |
| Meta Cloud API with custom software | Your own application talks directly to WhatsApp | Complex pricing, stock or multi-role logic; strong data ownership requirements | Highest cost, longest build, needs ongoing engineering |
A practical rule. If your order, pricing and stock logic can be expressed in an off-the-shelf tool, buy it and spend the saved budget on data quality. Build custom when WhatsApp must sit inside your operations rather than beside them, when you run several branches with shared stock, or when the data cannot leave your control.
The seven build phases
Phase 1: Discovery and process mapping (1 to 2 weeks)
Log a week of real conversations and sort them by type. Map the current journey for the two or three most common: enquiry to order, booking to attendance, complaint to resolution. Write the service levels and the escalation rules. This phase is where projects are saved or lost.
Phase 2: Identity and verification (1 to 3 weeks, partly outside your control)
Create or clean up the Meta Business account, complete business verification with your CAC documents, confirm the display name, and prepare the profile, catalogue and business description. Start this early: verification timelines vary and block everything downstream.
Phase 3: Inbox and routing setup (1 to 2 weeks)
Connect the number, configure the shared inbox, create queues, set assignment and escalation rules, write the menu, build the quick-reply library and agree the shift handover process.
Phase 4: Flows and templates (2 to 4 weeks)
Design the interactive flows that collect order, booking or support details as fields. Draft message templates and submit them for approval, allowing time for rejections and rewrites. Build the automated sequences with explicit stop rules.
Phase 5: Data and integrations (2 to 6 weeks)
Stand up or connect the data layer, then integrate payments, stock, delivery and your website. Payment confirmation by webhook is usually the highest-value integration and should be first.
Phase 6: Testing (1 to 2 weeks)
Test with real staff and real devices, not a demo script. Check the awkward paths: a customer who abandons a flow halfway, pays the wrong amount, sends a voice note instead of selecting an option, or messages at midnight. Test on a weak connection.
Phase 7: Migration, launch and training (1 week plus a settling period)
If you are moving an existing number to the Platform, plan it for a quiet trading period, because a number on the API cannot run in the App simultaneously. Export or record your catalogue, labels and saved replies first. Train staff on the inbox, not just the automation, and keep a daily review for the first fortnight.
Designing the data model
This is the part most vendors skip and most businesses later regret. Decide what each conversation must write into your records.
A workable minimum for a trading business:
- Customer: name, WhatsApp number, alternative number, delivery address or state, consent record with date and source
- Conversation: channel, first message time, category, assigned agent, first response time, outcome
- Order: reference, line items with quantity and price, delivery fee, total, payment status, payment reference, fulfilment status
- Booking, where relevant: service, resource or staff member, start and end time, deposit status, attendance outcome
- Support ticket, where relevant: category, priority, resolution, reopened flag
Two design rules matter more than the field list itself. First, capture data as fields, never as a pasted chat transcript, because fields can be searched, priced, counted and reported. Second, give every record a reference the customer can quote, since Nigerian customers frequently return through a different number or a different staff member.
Integrations that matter in Nigeria
Payments. Connect a provider such as Paystack, Flutterwave, Monnify or Remita, and use webhooks so confirmation is automatic. Where customers prefer bank transfer, a dedicated virtual account number per order removes manual reconciliation and the risk of edited screenshots.
Stock or POS. A catalogue that advertises unavailable items generates refunds, arguments and negative reviews. If you sell physical goods, stock accuracy is not optional.
Delivery. Dispatch and delivery events from partners such as GIG Logistics, Kwik, Sendbox or DHL, or from in-house riders, should trigger automatic status messages. This is the single most effective way to cut inbound support volume.
Website. Tracked click-to-chat links, forms that write into the same data layer, and consistent pricing between site and WhatsApp.
Accounting. Invoices and receipts flowing into your accounting system so month-end is not a reconstruction exercise.
CRM. If a CRM already exists, WhatsApp should write into it rather than becoming a second, competing customer database.
What changes for Nigerian businesses
The number is a brand asset. It is printed on vehicles, packaging, signage and shared by word of mouth. Migration planning matters more here than in markets where customers arrive through a website.
Business verification requires proper documentation. CAC registration documents and a consistent business name across your bank account, Meta account and website make verification smoother. Mismatched names are a common cause of delay.
Bank transfer shapes the architecture. Systems designed around card checkout do not fit most Nigerian SME customers. Design the payment flow around transfer confirmation first, with cards as an option.
Connectivity and data cost shape the flows. Keep flows short, text-first and tolerant of interruption. A customer who loses connection mid-flow should be able to resume rather than start again.
Power affects staffing, not just servers. Support rotas and response promises should reflect the hours your team can actually work reliably.
Staff turnover is a design constraint. Build so that a departing agent takes no customer records with them. This is one of the strongest arguments for a shared inbox and a proper data layer.
NDPA 2023 compliance is part of the build. Record consent with date and source, restrict agent access to what their role needs, set a retention position, and provide an opt-out that the system honours automatically. Verify specifics with the Nigeria Data Protection Commission or a qualified professional.
Example (hypothetical): a three-branch pharmacy group
This is an illustrative scenario, not a Linestech client account.
A pharmacy group runs three branches in Lagos. Customers message one number for availability, prices, refill reminders and delivery. Each branch also has its own number, which customers use inconsistently. Stock differs by branch, and controlled items require a pharmacist's judgement.
Design decisions taken.
- One customer-facing number on the Platform, with branch selection as the first menu step; branch numbers retained internally only.
- A shared inbox with three queues, one per branch, plus a fourth queue for prescription questions routed only to pharmacists.
- A flow capturing item, quantity, branch and delivery area as fields, with an explicit rule that prescription items always escalate to a pharmacist rather than being confirmed automatically.
- Integration with the branch stock system so availability answers reflect reality.
- Payment by dedicated virtual account per order, confirmed by webhook.
- A refill reminder sequence using approved templates, opt-in captured at first purchase and recorded with a date.
- Reporting on response time by branch and on unfulfilled requests by item, which became a purchasing input.
What made it work. Two things. The prescription escalation rule, which kept clinical judgement human, and the stock integration, which stopped the system promising what the shelves did not have. The automation itself was comparatively straightforward.
Regulatory note. Pharmacy operations in Nigeria are regulated, including which items may be sold and how. Any business in a regulated sector should confirm current requirements with the relevant authority, such as the Pharmacy Council of Nigeria or NAFDAC, before automating any part of a sales process.
What it costs and how long it takes
Figures below are indicative 2026 ranges. Actual quotations vary with scope, vendor, integrations and the naira exchange rate, because Platform charges, inbox licences and hosting are largely priced in US dollars.
| Build type | Scope | Indicative one-off cost | Typical timeline |
|---|---|---|---|
| Configured setup | Platform connection, verification, shared inbox, menu, quick replies, basic templates | ₦800,000 to ₦2,000,000 | 3 to 5 weeks |
| Standard system | The above plus interactive flows, payment integration with webhooks, CRM write-back, basic reporting | ₦2,000,000 to ₦5,000,000 | 6 to 10 weeks |
| Integrated system | Stock, delivery, website and accounting integrations, multi-queue routing, dashboards, multi-branch logic | ₦5,000,000 to ₦9,000,000 | 10 to 16 weeks |
| Custom build on Cloud API | Bespoke application, custom order and pricing logic, role-based access, full data ownership | ₦8,000,000 to ₦12,000,000+ | 12 to 24 weeks |
Recurring costs: Meta conversation charges by category in USD; inbox or CRM licences per user per month; payment provider transaction fees; hosting from roughly ₦150,000 per year for a small VPS and considerably more for a production application on cloud infrastructure; and maintenance, commonly ₦100,000 to ₦600,000 per month depending on scope.
When comparing vendors, insist on identical scope: the same number of flows, the same templates, the same integrations named individually, the same reporting, the same training days and the same post-launch support period. Two quotations that differ by several million naira almost always differ on integrations and support, not on messaging.
Maintenance and ownership after launch
A WhatsApp system is not a website that can sit untouched. Four things need standing ownership:
- Templates. Meta's categories and policies change; rejected or paused templates must be rewritten promptly.
- The reply library and flows. Prices, delivery areas and policies change constantly. Assign one owner.
- Number health. Watch quality rating and messaging limits after every campaign.
- Integrations. Payment providers and delivery partners update their APIs. Someone must monitor failed webhooks, or orders will silently stop confirming.
Name an internal owner at launch, even where a vendor provides support. Systems without an internal owner degrade within two quarters regardless of the support contract.
Mistakes to avoid
- Starting with tools instead of process. A week of logged conversations is worth more than three vendor demonstrations.
- Leaving business verification to the end. It is the most common cause of a delayed launch and it is not in your control.
- Migrating your main number without planning. Losing App access mid-trading day, with no exported catalogue or saved replies, is avoidable.
- Storing orders as chat text. Without fields there is no reporting, no stock check and no reliable fulfilment.
- Automating a regulated decision. Anything requiring professional judgement, whether clinical, legal or financial, must escalate to a qualified person.
- Ignoring the reporting layer. If nobody can see response times and category volumes, nobody can improve them.
- No stop rules on sequences. Messaging a customer who has already paid is the fastest way to earn a block.
- Treating consent as paperwork. Record it as data, with date and source, and let the system act on opt-outs automatically.
Conclusion
Building a WhatsApp business system is an operations project with a messaging interface, not a messaging project. The layers matter in order: identity, inbox, automation, data, integration, reporting. Skipping the data layer is the most common and most expensive error, because it turns every later improvement into guesswork.
Decide the eight questions before you ask for quotations. Start business verification immediately. Buy off-the-shelf unless your logic genuinely cannot be expressed in it. Integrate payments first, stock second and delivery third, because that order matches where the money and the support volume actually sit. Then name an internal owner, because templates, prices and integrations change every month whether or not anyone is watching.
Done this way, the system produces something more valuable than convenience: a business whose customer relationships, orders and history belong to the business rather than to whoever is holding the phone.
If you are ready to turn a busy WhatsApp line into a proper system, Linestech designs and builds WhatsApp Business Platform setups, integrations with Nigerian payment, stock and delivery providers, and custom order and booking software. Share how your conversations become orders today and we will map the layers you actually need.
Frequently asked questions
How long does WhatsApp business verification take in Nigeria?
It varies, and it is controlled by Meta rather than your vendor. Applications move faster when the business name matches across your CAC registration, your website, your Meta Business account and your supporting documents. Start the process in the first week of the project, because flows and templates cannot go live without it.
Can I keep using the WhatsApp Business App after moving to the Platform?
Not on the same number. A number registered on the WhatsApp Business Platform cannot simultaneously run in the App. Many Nigerian businesses keep a second App number for informal use or as a transition fallback, but the customer-facing number should be the one on the Platform.
Do I need a website as well as a WhatsApp system?
They serve different jobs. WhatsApp converts people who already know you; a website earns visibility in Google searches and gives customers something to verify before paying a stranger. A WhatsApp system with no website depends entirely on social platforms and referrals for discovery.
What happens to my chat history when I migrate the number?
Existing App chat history does not transfer into the Platform inbox. Export what matters, capture your labels, catalogue and saved replies as documents beforehand, and treat the migration as the start of proper record-keeping in the data layer rather than a continuation of chat history.
Can several branches share one WhatsApp number?
Yes, and it is usually better than publishing several numbers. Branch selection becomes the first step in the menu, conversations route to the branch queue, and reporting can compare branches. Internal branch numbers can remain for staff use while customers learn one number.
How do we handle customers who send voice notes instead of using the menu?
Present the menu at first contact, before the customer starts explaining. Accept that some will send voice notes anyway and route those straight to a human queue. The agent should record the outcome as structured data so the record stays complete even when the input was unstructured.
Is it safe to store customer data with an inbox vendor?
It depends on the vendor's security practices, hosting location and contractual terms, and on the sensitivity of your data. Ask where data is stored, who can access it, what happens on termination and whether you can export everything. For health, financial or other sensitive categories, take advice on Nigeria Data Protection Act 2023 obligations before committing.
What size of business justifies a custom build on the Cloud API?
Generally one where off-the-shelf tools cannot express the logic: complex pricing, multi-branch shared stock, role-based approvals, or a requirement that customer data never leaves systems you control. Below that threshold, a configured Platform setup with good integrations delivers most of the benefit at a fraction of the cost.
Sources and further reading
Figures, platform rules and regulations change. These are the primary references behind this article and the places to check before you act on it.
