When Your Tools Do Not Talk to Each Other: An Integration Primer for Egyptian SMEs

By Sila · 13 September 2026 · 8 min read

Walk into most small businesses in Cairo and you will find the same set of tools: WhatsApp for customers, an Excel or Google Sheet for tracking, a point-of-sale or booking system, an accounting package, and maybe an online store. Each one works. The problem is that none of them know the others exist. A customer's order lives in a WhatsApp chat, then in a sheet, then in the accounting software, and a person carries it between each stop by hand. That carrying is what an integration replaces. This article explains, without jargon, how tools get connected, which connection to build first, and what tends to go wrong so you can plan around it.

How to tell you have an integration problem

Owners rarely say "we need an integration." They describe symptoms. If more than two of these sound familiar, the tools are the issue, not the staff.

• The same customer, order, or patient gets typed into two or three places, and the versions do not always match.
• Nobody is sure which sheet is the current one, so people ask around before trusting a number.
• Month-end reports are assembled by copying and pasting from several systems into one file, and the person who does it is the only one who can.
• Customers are asked to repeat information they already sent, because the person answering cannot see the earlier message.
• A tool was bought to fix a problem, but it created a fourth place to update.

Each symptom is a gap between two systems. The fix is not a bigger system that does everything. It is a connection that moves the record automatically from where it is created to where it is needed.

The four ways tools get connected

There are only four practical methods, and most SMEs end up using a mix. Knowing them helps you understand what a vendor is proposing and why it costs what it does.

1. Built-in integrations

Many tools ship with ready connections to popular partners. An online store can push orders straight to a shipping provider, and a calendar can send events to a messaging app. These are the cheapest option and should always be checked first. The catch is that they cover only the pairings the vendor chose, and rarely fit Egyptian tools like local accounting packages or delivery companies.

2. Workflow platforms

Platforms like n8n and Make sit in the middle and act as a translator. They watch one tool for an event, reshape the data, and send it to another tool. This is the layer that does most of the work in the systems we build, because it connects almost anything that has an API, runs on rules you can read, and can be changed in an afternoon rather than a rebuild. Both are affordable for SMEs, and n8n can be hosted on your own server if data location matters to you.

3. Custom code against an API

An API is simply a door a software product opens so other software can read and write its data. When a workflow platform cannot do something, or when speed and reliability matter, a small piece of custom code talks to that door directly. The WhatsApp Business API is the best-known example for Egyptian businesses: it is the only supported way to send automated messages at volume, and it needs a small custom layer or a workflow platform in front of it.

4. Files and email as the fallback

Some tools have no API at all. Older accounting software and some local POS systems fall here. The fallback is a scheduled export, usually a CSV file, which a workflow reads and pushes onward, or an email the system already sends that a workflow parses. It is not elegant, but it removes the re-keying, and it often runs for years without complaint.

What an integration actually consists of

Whatever the method, every working integration has the same four parts, and asking about each one is a good way to judge a proposal.

1. A trigger. Something happens in tool A: a new order, a paid invoice, a booking confirmed, a message received.
2. A mapping. The fields in tool A are translated into the fields tool B expects. This is where most of the thinking goes. Which tool holds the master version of the customer's name and phone number? What happens when a field is blank?
3. An action. Tool B is updated: a row added, a record created, a message sent, a task assigned.
4. Error handling. When tool B is down or rejects the data, the workflow retries, then tells a named person. An integration without this part is a time bomb.

The mapping step is where you, the owner, add the most value. You know that the phone number in the booking system is the real one and the one in the sheet is often a typo. You know that "paid" in the POS means cash received but "paid" in accounting means reconciled. Write these rules down before anyone builds anything.

Which connection to build first

Do not try to connect everything at once. Follow the record that moves most often and costs the most when it moves wrong. For a retail shop that is usually the order, from wherever it is placed to stock and accounting. For a clinic it is the appointment, from booking to reminder to visit record. For a professional services firm it is the client enquiry, from first message to proposal to invoice. For a logistics operator it is the shipment, from order to driver to proof of delivery.

A simple way to choose: list the three records your business handles most, then for each one count the number of systems it touches and the number of times a human copies it. Multiply. The highest score is your first integration. It is almost always the one with the most copying, not the one with the most sophisticated software.

An illustrative example

The figures below are made up to show the method, not a client result. Replace them with your own.

Suppose a delivery company in Cairo receives orders two ways: merchants send them over WhatsApp, and a few larger merchants upload a sheet. A dispatcher copies each order into a master Google Sheet, messages the driver, and at the end of the day the accountant re-enters completed deliveries into the accounting package. Say that is 120 orders a day and each order costs about four minutes of copying and checking across the three stops. That is roughly eight hours of typing per day, spread across two people, before anyone has done the actual work of moving parcels.

The first integration here is not an app for merchants. It is a workflow that reads the WhatsApp messages and sheet uploads into one order table, formats the address and phone number consistently, sends the driver assignment automatically, and posts completed deliveries to accounting once a day as a batch. The dispatcher still handles exceptions, which is the part that needs a human. Put your own staff cost per hour against the hours recovered, in EGP, and compare it with a one-time build fee and a monthly retainer. If the payback is longer than a few months, the workflow is wrong or the volume is too low to bother yet.

What goes wrong, and how to plan for it

• Duplicates. A customer messages twice, or an order is edited after it was sent onward. Every integration needs a rule for what makes two records the same, usually a phone number or an order ID, and what to do when it sees one twice.
• Phone numbers and Arabic text. Egyptian numbers arrive as 010 local format, as +20 international format, with spaces, or with Arabic digits. Names arrive in Arabic, English, and both. Normalising these at the mapping step prevents most matching failures later.
• Tools without an API. Ask before you buy. If your accounting or POS vendor cannot describe how other software gets data out, you will be on the file-export fallback for as long as you use it.
• Silent failures. A workflow that stops without telling anyone is worse than no workflow, because people stop checking. Insist on alerts to a real person, and a weekly check that counts records in and records out.
• Tool changes. Vendors change their APIs and pricing. This is the practical reason a monthly retainer exists: someone has to notice and fix the break before your team does.
• No owner. One named person inside your business should be able to explain what the integration does, in plain words, and who to call when it misbehaves.

Questions to ask before you sign

Whether you work with us or anyone else, these questions separate a durable integration from a demo.

1. Which method are you using for each connection, and why not the cheaper one?
2. Which system is the master for each field, and where is that written down?
3. What happens when a tool is down for an hour? For a day?
4. Who gets alerted when something fails, and how quickly?
5. Where does the data live, and can we take it with us if we leave?
6. What is included in the monthly retainer, and what counts as a new project?

Start with one gap

The businesses in Egypt that get real value from automation rarely replaced their tools. They kept WhatsApp, the sheet, the POS, and the accounting package, and closed the gaps between them one connection at a time, starting with the record that was copied most. If you can name the record your team re-types most often, you already know where to begin. If you would like a second opinion on which gap to close first, or want it built and maintained properly, talk to us at Sila. We connect the tools Egyptian SMEs already use, and the first conversation is free.