Custom Business Dashboard for SMEs: What to Track, What It Costs and How It's Built (2026)

Photo: Bidco Africa Chooses IBM Solution, Services To Support Expansion by j_cadmus, CC BY 2.0. Cropped.

Why SMEs struggle to see their own numbers

Ask most SME owners "How much cash did we collect yesterday?" or "Which branch had the best margin last month?" and the honest answer is "Let me check." The data exists, but it's scattered across an M-Pesa till statement, a bank portal, the POS system, three Excel sheets, the accountant's QuickBooks and the sales team's WhatsApp. Someone spends hours every week compiling a report that is out of date the day it's finished.

A custom dashboard does the compiling automatically, every few minutes, and shows the result on one screen, on your laptop or your phone.

What should a business dashboard show?

The right KPIs depend on your business, but most SME dashboards cover five areas:

Keep the first screen to six to ten numbers. Details belong one click deeper. A dashboard with 40 charts is just a spreadsheet with colours.

Start with decisions, not data

The best dashboards are designed backwards from the decisions they support. Write down:

1. The five questions you ask every morning or every week
2. What you'd do differently depending on the answer
3. Who else needs to see each number (branch managers, sales lead, accountant)

For example: "Which customers are more than 30 days overdue?" leads to "Call them today", so the dashboard shows an overdue list with phone numbers and a WhatsApp button. "Which products will run out this week?" leads to "Place an order", so it shows a reorder list with suppliers. Every widget earns its place by triggering an action.

Which data sources can be connected?

• M-Pesa: live via Safaricom's Daraja API callbacks, or by importing statements. See M-Pesa reconciliation.
• Banks: statement imports or bank APIs where available.
• Accounting: QuickBooks, Xero, Zoho Books, Sage and similar through their APIs.
• POS and e-commerce: sales, products and stock.
• Spreadsheets: Google Sheets or Excel files that teams already maintain.
• WhatsApp: enquiries and orders through the WhatsApp Business API. See WhatsApp CRM integration.
• CRM and forms: leads, pipeline and website enquiries.
• GPS trackers and field apps: vehicles, deliveries and field visits.

Most of the build effort goes into connecting these sources reliably and cleaning the data, for example matching an M-Pesa payment to the right invoice or merging the same customer recorded three different ways.

Custom dashboard vs Power BI, Looker Studio or Excel

BI tools are excellent, but they assume your data is already clean and connected, and someone has time to maintain the reports. For many Kenyan SMEs, the missing piece is the data plumbing and the actions, which is where a custom build pays off.

Adding AI to a dashboard

• Plain-language briefings: "Sales are 12% below last Tuesday, driven by the Thika branch. Three large customers haven't ordered this month."
• Anomaly alerts: unusual drops in sales, spikes in refunds, cash variances or idle staff, sent to WhatsApp.
• Ask-your-data: type "Which products sold best in Mombasa last month?" and get an answer and a chart.
• Forecasts: expected cash in the next 30 days and stock run-out dates.

We describe the full version of this idea in what an AI operations command center is.

How much does a custom dashboard cost?

Cost depends mainly on the number and difficulty of data sources, the number of user roles and views, and whether you need actions and AI features. A focused dashboard with two or three integrations is a modest project. A multi-branch command center with many sources, roles and automations is larger. Our OpsCommand Core Dashboard ($8,497) includes a live operations dashboard with up to three data-source integrations, starting with a free 7-day prototype. Running costs are usually just hosting and any API fees.

How a custom dashboard is built

1. Discovery (days): agree the questions, KPIs and users, and get access to data sources.
2. Prototype (about a week): a working dashboard on real data for the most important view, so you can react to something real.
3. Integrations and data cleaning: reliable connections, deduplication, matching rules.
4. Views, roles and actions: owner, managers, sales, finance, each seeing what they need.
5. Alerts and AI: WhatsApp or email alerts and daily briefings.
6. Training and handover: documentation, and full ownership of the code.

Common dashboard mistakes

• Too many metrics: if everything is important, nothing is.
• Vanity metrics: page views and followers instead of cash and conversions.
• Untrusted numbers: if figures don't match the bank, nobody uses the dashboard. Reconcile first.
• No owner: every KPI needs someone responsible for acting on it.
• Looking, not acting: build in alerts and action buttons.

Example: a three-branch hardware retailer

Picture a hardware retailer with branches in Nairobi, Thika and Machakos. The owner used to receive daily sales by WhatsApp photos of handwritten summaries and check M-Pesa statements at night. A custom dashboard now shows sales and collections per branch live, cash variances per till, top-selling and slow-moving items, low stock by branch, and overdue contractor accounts with one-tap WhatsApp reminders. Every morning at 7am the owner gets a short AI briefing on WhatsApp. Month-end now takes hours instead of days, and stock transfers between branches happen before items run out, not after.

Who should own the dashboard internally?

Assign one person, often the operations or finance lead, to own data quality: checking that integrations are working, fixing mismatched records and requesting changes as the business evolves. Without an owner, small data issues accumulate until trust is lost.

Security and access

Dashboards concentrate sensitive data: revenue, salaries, customer details. Use individual logins with roles, two-factor authentication for admins, encrypted connections, audit logs and regular backups, and make sure the system complies with the Data Protection Act. Never share a single "admin" password across the team.

Next step

Write down the five questions you'd want answered on one screen every morning. That list is your dashboard specification. Send it to us and we'll build a free working prototype on your real data within 7 days. For non-technical owners, AI dashboards for small business explains how to work with one day to day.

Frequently asked questions

Explore Kavenacc services & pricing →