APIs, Zapier, or Custom Code? How to Pick the Right Automation Layer

Three Tiers of Automation. Most Companies Pick the Wrong One.

120+ projects. One consistent mistake: wrong tool. Zapier when you need APIs. Custom code when Zapier would work. Low-code platforms that become technical debt. Here's the framework that separates waste from ROI.

The truth is simpler than you think: there are three tiers, each with clear strengths and weaknesses. Let me walk you through them.

Tier 1: No-Code Tools (Zapier, Make, Integromat)

These platforms connect two or more apps by sending data between them. Zapier costs $20–$500 per month depending on task volume. You get speed: you can build a workflow in minutes: and almost zero maintenance.

Use no-code when: You're connecting off-the-shelf apps (HubSpot to Slack, Stripe to Google Sheets, Jira to email). Data volumes are under 10,000 events per month. The workflow logic is straightforward: "When X happens, do Y, then do Z." Your team has no developers available. You need something working this week, not next quarter.

The limit hits fast. No-code platforms struggle with conditional logic. If you need to say "If customer lifetime value is over $10,000, AND they've purchased in the last 90 days, AND they're not in Canada, then route to a specific sales rep," Zapier gets messy. It's doable, but you're nesting conditions, adding branches, and within a month your workflow looks like spaghetti.

I worked with a mid-market SaaS company that tried to automate their entire lead qualification process in Zapier. By month two, they had 47 Zaps running. By month four, nobody understood the whole system, and when something broke, it took days to debug. We ended up rebuilding it.

Cost range: $20–$500/month. Maintenance: Low until complexity creeps up. Scaling limit: ~10,000 tasks/month before cost and complexity explode.

Tier 2: Low-Code Platforms (Retool, n8n, Supabase)

These sit in the middle. You get more control than no-code but don't need full software engineers. Retool costs $50–$500/month. n8n can run self-hosted for free or $100+/month in the cloud.

Use low-code when: Your workflow has complex branching logic. You need to transform data, run custom calculations, or write JavaScript for business rules. You want to build internal tools (dashboards, admin panels) alongside automations. Your team has one or two people who can code a little but aren't software engineers.

Low-code platforms shine when you have messy data. A client in healthcare needed to match patient records across three different systems, handle exceptions where names didn't align, and flag records for manual review. Zapier couldn't do it. Custom APIs would've cost $30,000. Retool did it in two weeks for $300/month. The business logic is complex, but Retool lets you write JavaScript without managing infrastructure.

The trade-off: you need someone to maintain it. When you have 50 workflows running, and one breaks, who fixes it? With Zapier, you call support. With Retool, that someone is on your team. You own the platform.

Cost range: $50–$500/month. Maintenance: Medium: you need someone who can debug logic and manage changes. Scaling limit: Depends on your database and data volume, but generally handles 1M+ events/month without breaking.

Tier 3: Custom Code (APIs, Scripts, Microservices)

You hire a developer or team to build it. Costs range from $10,000 (a simple script) to $500,000+ (a full platform). This is what you choose when the other tiers can't cut it.

Use custom code when: You need to integrate 10+ systems in a single workflow. Your data is so messy that no platform can handle the transformation. You need real-time processing at scale (100,000+ events per second). You have edge cases that no pre-built solution anticipates. Your business logic is proprietary and competitors can't know how it works.

A fintech client needed to process 500,000 credit card transactions per day, validate them against 15 different fraud-detection services, enrich them with customer behavioral data, and route them to different backend systems. No platform, no matter how fancy, could handle that. We built custom APIs in Python, running on Kubernetes. Cost was significant: $150,000 upfront, $5,000/month to maintain: but the alternative was losing transactions or running on hardware that cost more.

Custom code also gives you control. Want to change how the system behaves? You own the code. Want to debug a specific edge case? You have the source. Want to integrate a new vendor? It's a code change, not a new subscription.

The downside: it's slow to build, expensive to maintain, and if your team leaves, you need to either rehire them or hire someone new who has to learn the system from scratch. I once took over a custom automation system that was written poorly, and it took two months just to understand what it did.

Cost range: $10,000–$500,000+ upfront. Maintenance: High: you need dedicated developers. Scaling limit: Whatever your infrastructure can handle.

The Decision Tree

Step 1: How much data? If fewer than 1,000 events per day, start with no-code. If 1,000–100,000 per day, low-code is probably right. If 100,000+, seriously consider custom code.

Step 2: How complex is the logic? If it's "connect two apps," no-code works. If you need to transform data or run calculations, low-code. If you need proprietary algorithms or multi-system orchestration, custom code.

Step 3: What's your team skill? No-code if you have no developers. Low-code if you have one person who can code. Custom code if you have a dedicated team.

Step 4: What's the maintenance cost? If you can't afford to have someone monitor the system, no-code is cheaper (Zapier handles the maintenance). If you're comfortable owning a platform, low-code or custom code make sense.

Real Examples of Choosing Wrong

A B2B SaaS company spent four months building custom APIs to synchronize customer data across Salesforce, Stripe, and their data warehouse. The system worked, but it cost $80,000 and they needed a developer to maintain it. A year later, they realized Zapier could've done it for $200/month. They'd chosen based on the assumption that "in-house is always better." It wasn't.

Another company built a Retool dashboard that grew so complex they couldn't scale it. They added more features, more data sources, more users, and eventually the system was so slow that people stopped using it. What should've been a low-code solution became a custom code job anyway.

The pattern I see most: companies start with no-code, outgrow it faster than they expect, and scramble to migrate to low-code or custom code. That migration costs time and money. The smarter move is to pick the tier you'll need in 12 months, not the tier you need today.

What to Choose Right Now

If you're starting today, here's my honest take:

• No-code first: Unless you know you need custom code, start with Zapier or Make. You'll learn what your automation needs actually are. You can always upgrade later.
• Low-code second: If no-code breaks, move to low-code. You keep most of your workflows intact while adding custom logic where you need it.
• Custom code last: Only choose this if you've already tried the other two and they didn't work. Or if you know from the start that you need to process 100,000+ events per second.

I've seen companies waste six figures on custom solutions that Zapier would've handled fine. I've also seen companies outgrow no-code and have to rebuild everything. The right answer depends on your data volume, your logic complexity, and your team's skills.

Richard Batt has delivered 120+ AI and automation projects across 15+ industries. He helps businesses deploy AI that actually works, with battle-tested tools, templates, and implementation roadmaps. Featured in InfoWorld and WSJ.

Frequently Asked Questions

How long does it take to implement AI automation in a small business?

Most single-process automations take 1-5 days to implement and start delivering ROI within 30-90 days. Complex multi-system integrations take 2-8 weeks. The key is starting with one well-defined process, proving the value, then expanding.

Do I need technical skills to automate business processes?

Not for most automations. Tools like Zapier, Make.com, and N8N use visual builders that require no coding. About 80% of small business automation can be done without a developer. For the remaining 20%, you need someone comfortable with APIs and basic scripting.

Where should a business start with AI implementation?

Start with a process audit. Identify tasks that are high-volume, rule-based, and time-consuming. The best first automation is one that saves measurable time within 30 days. Across 120+ projects, the highest-ROI starting points are usually customer onboarding, invoice processing, and report generation.

How do I calculate ROI on an AI investment?

Measure the hours spent on the process before automation, multiply by fully loaded hourly cost, then subtract the tool cost. Most small business automations cost £50-500/month and save 5-20 hours per week. That typically means 300-1000% ROI in year one.

Which AI tools are best for business use in 2026?

It depends on the use case. For content and communication, Claude and ChatGPT lead. For data analysis, Gemini and GPT work well with spreadsheets. For automation, Zapier, Make.com, and N8N connect AI to your existing tools. The best tool is the one your team will actually use and maintain.

Put This Into Practice

I use versions of these approaches with my clients every week. The full templates, prompts, and implementation guides, covering the edge cases and variations you will hit in practice, are available inside the AI Ops Vault. It is your AI department for $97/month.

Want a personalised implementation plan first? Book your AI Roadmap session and I will map the fastest path from where you are now to working AI automation.