Ask an SME owner where their margin is leaking and they'll point at procurement, payroll or the energy bill. Almost never at Excel. And yet that's often where the bleeding starts. Customer records living in three tabs at once. An order list a warehouse worker re-enters by hand into the ERP, again and again, because the webshop and the ERP don't talk to each other. A report that gets stitched together every Monday from four exports and a pile of VLOOKUPs by someone who really should be doing something else. It works. Until it doesn't. And something always goes wrong eventually.
Studies show that 88-94% of business spreadsheets contain material errors [1]. Not occasionally. Right now, in your files, almost certainly too. 70% of CFOs still rely primarily on Excel for planning and forecasting [2], while 44% of organisations name spreadsheet errors as their biggest planning risk [3]. We hold on to it anyway. Not out of love for Excel. More because the step toward connected systems feels opaque, and nobody enjoys breaking open a working process for something unfamiliar.
It doesn't have to be opaque. Connecting software in 2026 is more accessible, cheaper and faster than it was a few years back. Below is the approach that works for SMEs. In the Netherlands, where 81.5% of SMEs hit at least a basic level of digitalisation in 2024 [4], this is becoming standard practice rather than an experiment. No hype, just concrete choices.
Why Excel processes keep haunting SMEs
The Dutch SME sector is more digital than it looks. 81.5% reached at least a basic level of digitalisation in 2024, and the government is aiming for 95% by 2030 [4]. Behind those numbers sits a shadow layer. Among the smallest businesses, only 24% are genuinely digitalised in their day-to-day operations [5]. The rest run on scattered SaaS tools, a few shared folders, and a wall of Excels glueing it all together.
In Dutch SME manufacturing, 89% now have an ERP [6]. Sounds good. Until you ask whether that ERP talks to the CRM, the webshop, the accounting platform and the order portal. Then it goes quiet. Between those systems sits Excel. And an employee retyping.
The cost isn't invisible, just unmeasured. Up to 14 hours per week per team is lost to manually moving data between systems [7]. Knowledge workers spend up to 30% of their working week on reporting and hunting for data [8]. McKinsey estimated in November 2025 that 57% of working hours in office jobs is already automatable with technology that exists today [9]. Much of that automatable work is literally this: bridging systems that aren't connected.
We see it in every first conversation. The owner thinks the answer is to hire someone. In most cases, a well-chosen integration solves it better, for less money.
What 'connecting software' actually means
Connecting isn't magic. Two systems exchange data automatically, with no human in between. That's it. Three flavours matter.
An API (Application Programming Interface) is an agreement that lets software A request or send data to software B. Pull model: you actively ask for data. Almost every modern SaaS tool has one. Not every API is well-documented or stable, mind you, and you only notice the difference after a week working with it.
A webhook is the reverse route. Software B automatically sends a message the moment something happens. Push model, real-time. Webhooks are usually cheaper and faster than continuously querying an API. Zapier itself estimates that only 1.5% of polling calls actually return an update [10]. The other 98.5%? Wasted compute and wasted money.
iPaaS (Integration Platform as a Service) is a platform that ties APIs and webhooks together without you having to write code. The iPaaS market is growing from $7.85 billion in 2025 to $9.24 billion in 2026, with the SME segment growing fastest at a 32.1% CAGR [11]. For a reason. For light flows it's the quickest route.
Below that sits a fourth option: custom integration. Code written specifically to make two systems communicate directly. Not needed for every flow. For your core processes it's often the smartest choice. More on that shortly.
Step 1: Map your Excel flows — what's data, what's logic?
Before you connect anything, you need to know what an Excel actually does. Almost every "critical" Excel breaks down into two things:
- Data. Customer records, orders, stock, timesheets, invoices. Doesn't belong in Excel. Belongs in a system.
- Logic. Calculations, scenarios, reports, "let me whip up a quick overview". Can stay in Excel just fine, as long as the data flows in automatically from the source.
That distinction is where most projects go wrong. People replace Excel with another Excel that has different buttons. Or they rip Excel out and try to cram everything into a new system, including the scenario analyses only the leadership team still understands. Both approaches cost time and yield little. Pull the data out and put it in the right place. Let Excel do what it's good at: ad-hoc analysis and what-if scenarios. We've covered a complete approach to mapping manual work and a prioritisation matrix for workflow automation you can use here.
Practical test. Pick one Excel that gets touched every week by at least two people. Walk through every column and ask the same question: where does this data originally come from? If the answer is another system, that column belongs there. The Excel can have a live connection — not a copy.
Step 2: Pick the right integration method per use case
Not every integration deserves the same approach. A short overview:
| Situation | Best approach |
|---|---|
| Two standard SaaS tools with good APIs, light flow (<1,000 actions/month) | iPaaS / no-code platform |
| Real-time updates needed (order status, payments) | Webhooks, possibly via iPaaS |
| High volume (>10,000 actions/month) or complex data transformation | Custom integration |
| Legacy system without an API or with a bad one | Custom middleware |
| Multiple systems, dependent logic, your own business rules | Custom integration layer |
| One-off report or dashboard | Data straight from sources via a BI tool |
It's often pitched as "start cheap with no-code, scale up later". Honestly, that rarely works. For light edge processes, no-code is fine. For your core processes — order to invoice, lead to customer, time tracking to payroll — you hit limits much faster than you think. Quietly limits on action counts. Quietly limits on data transformation. Quietly licence costs that scale with your growth in ways you can't see in year one. And then you start over, with double the cost.
The choice between no-code, custom or a hybrid model is technical and strategic. It determines how well you can scale in two years' time, how dependent you are on an outside vendor, and what your maintenance costs look like. This is where it pays to bring in someone with experience before the first flow is built. Not after.
Step 3: Build it or have it built
Say you know which flow you're going to connect and which method fits. What are the routes?
Build it yourself on a no-code platform. For a team lead with technical inclination, doable for simple flows. Putting a lead from a form into the CRM. Creating a new customer in the accounting tool. Count on 1-3 days of work per flow, plus subscription costs that climb as volume grows. A marketing agency cut its client onboarding from 4 hours to 45 minutes by tying three standard tools (CRM, invoicing and Workspace) together [12]. Nice result. With the caveat that those were the lighter processes. The heavier side of their business stayed in custom builds.
Have a native integration configured. Many SaaS tools offer built-in connections to popular other tools. Not building, just configuring. Good for "standard" combinations. Limited beyond that.
Custom integration. Your own integration layer, built specifically for your stack and your business rules. More expensive upfront, cheaper to maintain at volume. A study of 47 SMEs showed that integrating CRM, automation and content tools cut weekly manual work from 45 hours to 12 hours — a 73% reduction [13]. Numbers like that don't come from loose no-code Zaps. They come from thoughtful architecture where you think upfront about edge cases, duplicate records, and that one customer who somehow exists twice.
Whichever route fits, never build without mapping. First know which fields come from where. How they get transformed. What happens when something fails. Sounds obvious. Gets skipped systematically.
Step 4: Test, monitor and build in error handling
An integration that works on the day it goes live is an integration that's silently failed four times in six months without anyone noticing. Building isn't the hard phase. Keeping it alive is.
Three things you need as standard:
- Logging. Every action gets recorded. What got through, what failed, when.
- Alerts. If an integration stalls or a queue backs up, someone gets notified within minutes. Not a monthly email.
- Retry & dead letter queue. Failed messages get retried. If they really won't go through, they end up in a separate pile for a manual check. Not silently lost.
Then there's GDPR. The moment personal data flows between systems, you need processor agreements and data minimisation. Not every iPaaS vendor belongs in a European context. Always check where the data is processed. A US hyperscaler with a European data centre is not the same as a European platform. We ask about it on every architecture decision.
When Excel actually stays the right answer
Not every Excel deserves to be replaced. Really. Excel is a fantastically cheap, flexible, widely available tool for analysis. Using it for what it does well — building models, running scenarios, one-off reports — is fine. It only becomes a problem when Excel becomes the source of truth for data that already exists elsewhere.
Rule of thumb: Excel is allowed to consume, not to store. Connected to your systems via a live data layer or dashboard, Excel is a perfectly good endpoint. As a storage place for customer records, stock or invoices, it isn't.
What it costs and how long it takes for SMEs
A realistic range for SME projects in 2026:
| Project type | One-time cost | Lead time |
|---|---|---|
| No-code integration between 2 standard tools | €1,500-€5,000 | 1-3 weeks |
| Custom integration between 2 systems with transformation | €5,000-€15,000 | 2-6 weeks |
| ERP-CRM-webshop integration (core of the business) | €15,000-€40,000 | 2-4 months |
| Reporting dashboard on connected data | €2,000-€8,000 | 2-4 weeks |
Plus monthly subscriptions for iPaaS platforms (typically €30-€200 for light flows, climbing with volume) or hosting for custom builds (from €50/month). Lead time is rarely set by the build itself. It's almost always set by the quality and accessibility of the existing APIs. We've had projects where two weeks of building was followed by six weeks of waiting for a third party to extend their API. Sometimes it's just part of the deal. We covered that cost and ROI reality in detail earlier and in our honest guide on AI implementation costs.
What you get back
The numbers are strikingly consistent. Companies that genuinely connect their software report time savings of 70-90% on the automated processes. A typical case: from 45 hours of manual work per week to 12, a saving of 33 hours. Week after week. Time your team can spend on client work instead of retyping [13]. Webhook integrations cut integration costs by up to 80% compared to polling solutions [14]. Multi-step automation, according to recent analyses, delivers 30-200% ROI in year one and up to 300% over the longer term [15].
Not future talk. This is what SMEs taking the step now report daily. The tools are here. The methods are settled. The difference between businesses that pull ROI out of this and businesses that get stuck isn't in the technology. It's in who you bring in to make the right choices upfront, and in the willingness to do that first round of mapping work properly before the first line of code is written.
Replacing Excel files with real integrations?
We map your systems, data flows and automation opportunities and build integrations that last for years. No scattered Zaps — an architecture that fits your growth. Get in touch →
Sources
[1] The Analytics Doctor, "7 Common Spreadsheet Errors", https://www.theanalyticsdoctor.com/spreadsheet-errors/
[2] CFOTech, "Seventy percent of CFOs risk errors by relying on Excel for finance", https://cfotech.co.uk/story/seventy-percent-of-cfos-risk-errors-by-relying-on-excel-for-finance
[3] Valorx / Ventana Research, "Still forecasting with spreadsheets — eliminate this costly mistake", https://valorx.com/resources/blog/still-forecasting-with-spreadsheets-eliminate-this-costly-mistake-in-weeks/
[4] Rijksoverheid, "More SME digitalisation for competitive strength", https://www.rijksoverheid.nl/actueel/nieuws/2025/03/07/meer-digitalisering-mkb-concurrentiekracht-totale-digitale-economie-onder-druk
[5] CBS, "Digitalisation and Knowledge Economy 2025", https://www.cbs.nl/nl-nl/longread/rapportages/2026/digitalisering-en-kenniseconomie-2025
[6] Aiden, "Digital transformation in Dutch SME manufacturing", https://aiden.eu/kennisbank/blog-hoe-digitale-transformatie-vorm-krijgt-in-de-mkb-maakindustrie/
[7] Aonflow, "The Hidden Costs of Manual Data Entry", https://aonflow.com/blog/the-hidden-costs-of-manual-data-entry-between-dynamics-365-and-business-central
[8] Simply Flows / Forrester, "Time wasted on repetitive tasks", https://simplyflows.com/time-wasted-on-repetitive-tasks-is-40-percent/
[9] McKinsey Global Institute, "Agents, Robots and Us (November 2025)", https://medium.com/@angelaf_56127/mckinseys-november-2025-bombshell-57-of-work-hours-already-automatable-e57bd9b706f1
[10] Merge, "Polling vs Webhooks", https://www.merge.dev/blog/webhooks-vs-polling
[11] Fortune Business Insights, "iPaaS Market 2026", https://www.fortunebusinessinsights.com/integration-platform-as-a-service-ipaas-market-109835
[12] Corey Duchemin, "Why Zapier is still a startup favorite in 2025", https://medium.com/@coreyduchemin/why-zapier-is-still-a-startup-favorite-in-2025-powering-growth-with-smart-automation-c9b3cbe2a265
[13] Rapid Architect, "SMB AI productivity case study", https://rapidarchitect.com/ai-real-life-case-study/
[14] JDM Software, "Webhooks vs APIs Business Guide 2025", https://www.jdmsoftware.com.au/webhooks-vs-apis-complete-business-guide-software-integration-2025
[15] Hackceleration, "Zapier Review 2026", https://hackceleration.com/zapier-review/
