Time lost switching between systems
Erwin Berkouwer
Employees at Dutch SMEs switch daily between multiple systems (CRM, email, accounting, planning), which research on cognitive load shows can cause 20 to 40 percent productivity loss with a recovery time of 10 to 25 minutes per switch. AI agents and system connections can reduce this by automatically transferring data and letting one AI assistant query multiple systems, with estimated costs of 1,500 to 25,000 euros depending on complexity. Automation does not pay off with low volume, unstable processes, or systems without an API.
Constantly switching between CRM, email, accounting and planning tools costs SMEs far more time than it seems. This article shows what AI concretely solves, what it costs, and when it does not pay off yet.
The problem: constantly switching between systems
The average employee at a Dutch SME does not work in one system. They work in seven. Email is open next to the CRM, accounting runs in a separate tab, WhatsApp Business pings in between, and planning is still tracked in a spreadsheet. Every time someone jumps from one application to another, something gets lost: attention, time, and sometimes data too.
That switching feels small. Changing a tab takes a couple of seconds. But the accumulation of all those small switches is where it really hurts. Research on cognitive load shows that people can lose 20 to 40 percent of their productivity when switching between tasks, with a recovery time of 10 to 25 minutes per interruption before they are fully back in the task. Translate that to a workday with multiple systems and multiple interruptions, and there is significantly less time left for the work that actually matters.
Why this hits SMEs extra hard
Large organizations often have specialists per system. At an SME, one person often does everything: creating quotes, sending invoices, answering customer questions, keeping the schedule up to date. That person does not switch between systems twice a day, but dozens of times.
"On my planning board everyone looks fully booked. Yet deadlines slip and everyone feels rushed. The problem isn't the amount of work, it's the constant switching in between."
That is exactly the pattern we often see at UnifyAI clients: not too little capacity, but too much fragmentation.
The hidden costs nobody counts
What makes this extra tricky: nobody puts "switching cost" on the budget. There is no line item in the accounts for "time lost searching in the wrong system". Yet the effect shows up in other numbers: longer turnaround times for quotes, more errors in invoices because data is entered twice, and customers waiting longer for an answer because an employee first has to search three systems for the right one.
There is also a quality side to this story. Research on cognitive load shows that work interrupted by switching moments tends to be shallower and more error prone than work completed in one go. For administrative tasks such as invoicing or updating customer data, that means a real risk of mistakes that later cost even more time to fix.
What AI concretely does to solve this
AI does not solve the switching problem by adding yet another tool. It solves it by making systems talk to each other, so people have to switch less often. A few concrete applications we see in practice:
- AI agents that transfer data automatically. A new lead in the inbox is automatically created as a contact in the CRM, without anyone needing to copy and paste.
- One AI assistant that queries multiple systems at once. Instead of searching three systems for a customer's status, an employee asks one question and the AI advisor pulls the answer from the CRM, email, and accounting system at the same time.
- Workflow automation between existing applications. Quote requests that automatically trigger a task, a draft quote, and a calendar appointment, without anyone switching between four screens.
This type of automation is often housed in AI agents that run in the background and only step in when a human is needed to make a decision.
Importantly, AI is not bolted on separately from existing systems, it is built on top of or in between them. Most SMEs do not want to replace their CRM, accounting package, or planning tool, they want those systems to finally talk to each other. That is a fundamentally different approach than adding "yet another dashboard", which in the long run would create more switching moments instead of fewer.
A concrete approach: from scattered switching to one workflow
Most companies do not start by rebuilding their entire system landscape. They start by mapping where the most switching happens, then tackle one connection or automation at a time.
Step 1: Map the switches
A simple exercise: have an employee note for one day which systems they switch between and why. It often turns out that 70 to 80 percent of the switches revolve around three or four fixed combinations, for example email to CRM, CRM to accounting, and WhatsApp to planning.
Step 2: Connect the systems used together most often
| Situation | Without a connection | With an AI connection |
|---|---|---|
| New customer question via email | Manually retyped into the CRM | Automatically created as a lead |
| Quote approved | Open a separate invoicing program | Invoice is automatically prepared |
| Customer calls about status | Searching across three systems | One overview via the AI assistant |
| Planning changes | Spreadsheet and calendar updated separately | Automatic synchronization |
Step 3: Let AI take over the in-between steps, not the decision
The goal is not for AI to take over every decision. The goal is for AI to take over the switching work, so a human only has to handle the final, substantive part. That distinction matters for buy-in within a team: people accept automation faster when they still feel they are in control.
For companies unsure where to start, a short AI scan is often a useful first step to see which system switches cost the most time.
Costs: what to expect
Important to say upfront: the cost of this kind of automation depends heavily on how many systems need to be connected and how messy the source data is. The figures below are indicative.
| Type of project | Cost | Payback period |
|---|---|---|
| Simple connection (two systems, standard API) | around 1,500 to 4,000 euros | around 2 to 4 months |
| AI assistant spanning multiple systems | around 4,000 to 12,000 euros | around 4 to 8 months |
| Full workflow automation (multiple departments) | around 10,000 to 25,000 euros | around 6 to 12 months |
These figures are estimates based on typical SME projects and are not a guarantee. Custom work, outdated systems without an API, or many manual exceptions can significantly raise the cost.
When it does not (yet) pay off
Not every company benefits from this approach, and it is more honest to say so now than to find out afterward.
- If the volume is too low. Do you switch between two systems twice a day? Then the time savings from automation are likely smaller than the investment needed to build it.
- If the systems themselves will be replaced soon. Building a connection to a system that will be phased out in six months is wasted money.
- If processes are not yet stable. AI connections work best on processes that have been stable for a while. On processes that still change weekly, you risk the automation becoming outdated faster than it delivers value.
- If there is no API or export option. Some older systems simply do not allow a connection without expensive custom software. In that case a system replacement is needed first, not an AI project.
In short: if the switching is incidental, the best solution is sometimes just a better routine, not technology.
It is also worth realizing that automation is not a one-time snapshot. A connection that fits perfectly today may need adjustment a year from now because a system gets an update or a vendor changes. So do not only budget for the build cost, but also leave a little room for maintenance in your decision.
How to approach this in practice
Want to know where in your organization the most time is lost switching between systems? A good starting point is a conversation where we look together at your current systems and workflows, without immediately committing to a large project. Feel free to get in touch, or first read more about how AI consultancy works at UnifyAI.
Frequently asked questions
How much time do you typically lose switching between systems?
Research on cognitive load shows that people can lose 20 to 40 percent of their productivity when switching between tasks, with a recovery time of 10 to 25 minutes per switch. The exact percentage varies by role and by company.
Can AI eliminate all system switching?
No. AI can significantly reduce retyping and searching across systems, but not every decision or customer interaction can or should be automated. The goal is less switching, not zero switching.
Is connecting systems expensive to build?
That depends on the number of systems and whether they have a modern API. Simple connections are relatively affordable, see the cost table above as an indication. Outdated systems without an API are usually the biggest cost driver.
Do we need to replace our systems before using AI?
Not always. Many existing systems have an API or export function that is enough for a first connection. Only with genuinely closed, outdated software is system replacement often a necessary first step.
Where is the best place to start?
Start by mapping the most common switches in a normal workday. A short AI scan or an exploratory conversation usually quickly reveals the two or three connections that would save the most time.
Veelgestelde vragen
Korte, heldere antwoorden die je helpen sneller beslissen.
How much time do you typically lose switching between systems?
Research on cognitive load shows that people can lose 20 to 40 percent of their productivity when switching between tasks, with a recovery time of 10 to 25 minutes per switch. The exact percentage varies by role and by company.
Can AI eliminate all system switching?
No. AI can significantly reduce retyping and searching across systems, but not every decision or customer interaction can or should be automated. The goal is less switching, not zero switching.
Is connecting systems expensive to build?
That depends on the number of systems and whether they have a modern API. Simple connections are relatively affordable, see the cost table in the article as an indication. Outdated systems without an API are usually the biggest cost driver.
Do we need to replace our systems before using AI?
Not always. Many existing systems have an API or export function that is enough for a first connection. Only with genuinely closed, outdated software is system replacement often a necessary first step.
Where is the best place to start?
Start by mapping the most common switches in a normal workday. A short AI scan or an exploratory conversation usually quickly reveals the two or three connections that would save the most time.
