Many SMB owners ask us the same question. Should we build this in Airtable or Notion, or do we need our own platform? The answer is rarely black and white. Airtable and Notion are excellent tools for 60 to 80 percent of the first phase. Only when you hit specific limits does a custom platform become the right choice. In this article we put the three tools side by side, explain where you reach a limit, and give a concrete decision checklist.
What are Airtable, Notion and Coda exactly?
Airtable is at its core a relational database with a spreadsheet interface. Tables, columns, links between records, and views on top. Strong for structured data with 5 to 20 users.
Notion started as a document system and grew into light databases. Strong where text and structure come together, such as handbooks and knowledge management.
Coda sits between Notion and Airtable. Doc-oriented like Notion, but with more powerful formulas and packs.
How do they differ from a custom platform?
| Feature | Airtable | Notion | Coda | Custom platform |
|---|---|---|---|---|
| Own UI and branding | Limited | Limited | Limited | Fully free |
| Complex business logic | Limited | Minimal | Reasonable | Unlimited |
| Record scale | Up to 250,000 per base | Hundreds to thousands | Comparable to Notion | Millions, no ceiling |
| Multi-tenant | Not native | Not suitable | Not suitable | Supported by default |
| Performance at scale | Slow above 50,000 records | Slow above thousands | Comparable | Scalable |
| Legacy system integrations | Limited via Zapier | Limited | Limited | Direct via API |
| Code ownership | No (SaaS lock-in) | No | No | Yes (your repo) |
| Monthly cost at scale | Scales with seats | Scales with seats | Scales with makers | Fixed (hosting) |
What do they cost at SMB scale?
| Number of users | Airtable (year) | Notion (year) | Coda (year) | Custom platform (year) |
|---|---|---|---|---|
| 5 users | €1,200 (Team) | €600 (Plus) | €720 (Pro) | €2,500 maintenance (after €12k build) |
| 20 users | €5,000 (Business) | €2,400 (Business) | €2,880 (Team) | €2,500 maintenance (after €15k build) |
| 100 users | €30,000+ (Enterprise) | €18,000+ (Enterprise) | €21,600+ (Enterprise) | €4,000 maintenance (after €18k build) |
At 5 users no-code is logical and almost free. At 20 users custom starts to get interesting. At 100 users custom is almost always cheaper.
When is Airtable the right choice?
- Product or inventory catalogue. 500 to 50,000 products, columns for suppliers, prices, stock.
- Content calendar and marketing planning. Marketing team of 3 to 10 people.
- Project tracker for agencies. 5 to 30 projects at a time.
- Light CRM for specific niches. Sales team of 3 to 8 people.
- Operational back office for 5 to 20 staff. Coaching practice or training agency.
When is Notion the right choice?
- Internal wiki and onboarding. Growing company (10 to 50 people).
- Customer knowledge base. SaaS company or consultancy.
- Sales playbook and proposal library.
- Light product roadmap. Product team of 2 to 6 people.
- HR handbook and personnel policy.
For hard reporting or customer-facing interfaces see dashboards and CMS solutions.
When do you hit the limits?
- Volume beyond 50,000 records per table. Airtable gets slow. Notion breaks around 5,000 to 10,000 records.
- Proprietary calculation logic that no formula can handle anymore. Dutch points-based rent calculation (WWS), complex sales-price stratifications.
- Multi-tenant scenarios. Each client a separate environment, data, permissions.
- Performance and concurrency problems. Long load times, automations queued.
- Tight integrations with legacy systems. ERP without a public API.
- Compliance and audit requirements. Notarial workflows, medical record requirements.
- Customer-facing interfaces with your own branding. Airtable Interfaces never quite feels like “your” product.
When do you need a custom platform?
- Sector-specific calculations. WWS points calculation for real estate (see Mastone case).
- Multi-tenant SaaS-like products.
- High scale or performance demands.
- Strict code ownership and vendor independence.
- Productising your work process.
Hybrid approach: combining often works best
- Notion for docs, custom for client flow. Law firm with knowledge management in Notion, intake flow in custom with Kleos integration.
- Airtable for content, custom for data. E-commerce plans marketing in Airtable, order flow in custom.
- Notion for handbook, custom for production. Manufacturer with quality handbook in Notion, production tracking in custom.
Which decision checklist do you use?
- How many records do you expect in 2 years? Below 10,000: Notion/Airtable. Above 50,000: custom.
- How many users, and what is their work pattern?
- Do you have proprietary logic that no tool supports out of the box?
- Do you need multi-tenant or customer portals?
- What is your budget over 3 years? Include seat growth in the math.
Performance benchmark: 10k vs 50k vs 250k records
Theory is nice, numbers are better. We tested the same dataset (a real estate portfolio with 25 columns) at three scale points in Airtable Business, a Notion database and a custom platform on a Hetzner VPS with PostgreSQL. Measured as load time for a filtered view and an aggregation query.
| Volume | Airtable view | Notion view | Custom (PostgreSQL) |
|---|---|---|---|
| 10,000 records | 1.2 sec | 3.4 sec | 0.3 sec |
| 50,000 records | 4.8 sec | Crashes after 12 sec | 0.4 sec |
| 250,000 records | Ceiling reached | Not feasible | 0.9 sec |
Airtable holds up fine to around 25,000 records. Above that filters and formulas slow down noticeably, and grouping on multiple columns becomes unusable. Notion breaks much earlier. A PostgreSQL database with a few indexes stays under one second well into the millions of rows.
Migration path: from Airtable to custom in 3 weeks
The question we get most is not “should I migrate” but “how do I do it without grinding the business to a halt”. We run this every month and the playbook is largely the same. Below is the version for a typical SMB team with 5 to 20 users.
- Week 1: data audit, schema mapping and PostgreSQL schema. Which tables are actually in use, which links between records and which formulas, plus building the PostgreSQL schema and seed script with a test import on a copy of the Airtable export.
- Week 2: build the screens, automations and role permissions. List and detail views for the most used tables, the remaining screens, automations and role permissions.
- Week 3: run in parallel and cut over. The team still works in Airtable while the new platform runs read-only alongside for validation, then cut over on a Friday, the weekend for the final sync, Monday live.
In 9 out of 10 cases this works without downtime. The biggest risk factor is not the tech but the human side, so plan 2 extra weeks for coaching and bug fixes after go-live.
Client example: training agency from Airtable to custom
A training agency in Utrecht (32 staff, 800 sessions per year) ran on Airtable Business in 2024 for planning, attendee admin and billing input. Cost: 4,800 euro per year on licences plus an estimated 6 hours per week of manual work to force Airtable into things it cannot do, like automatic per-attendee invoice lines and certificate PDF generation.
- Before (Airtable): 4,800 euro licences, 312 hours per year on workarounds, average 2 invoicing errors per month.
- Build custom: 18,000 euro one-off in 3 weeks, with Stripe integration and a PDF generator.
- After (custom): 3,200 euro per year for hosting plus maintenance, 40 hours per year on workflow work, near zero errors.
- Payback: 11 months at an internal rate of 65 euro per hour.
Important detail: the team initially wanted to stay on Airtable. The tipping point was not the price but the unreliability of complex formulas and the time spent correcting issues with clients.
Multi-tenant: why Airtable gets stuck here
Multi-tenant means multiple clients or branches each get their own isolated environment inside the same product. Airtable does not have this concept as a built-in feature. You can mimic it, but every attempt hits one of the limits below.
- Bases per client costs a fortune. A base per client means N times the seat licences. At 20 clients you head toward 30,000 euro per year.
- Filter tricks are not safe. A “tenant_id” field with filters does not meet data separation requirements. A user with a view link sees everything.
- Permissions are coarse. Read-only or edit, not field-level or row-level like PostgreSQL row-level security.
- Branded environments are missing. Client A cannot see their own logo or subdomain.
A custom platform solves this with a single PostgreSQL database, row-level security per tenant id and its own auth layer. In the Mastone case we built exactly that: a platform with an admin side and a separate investor portal where each investor sees only their own overview, without the hosting bill growing per user.
3-year TCO worked out: 20 users, realistic scenario
A fair comparison counts not just licences, but also the internal hourly cost lost on workarounds, plus the one-off build for custom. Here is the full picture for an SMB with 20 users, measured across 3 years.
| Cost item | Airtable Business | Notion Business | Custom platform |
|---|---|---|---|
| Licences / hosting 3 years | €15,000 | €7,200 | €7,500 |
| One-off build | €0 | €0 | €18,000 |
| Workaround hours per year (at €65) | €10,140 (156 hrs) | €7,800 (120 hrs) | €1,300 (20 hrs) |
| Tooling add-ons (Zapier etc.) | €2,400 | €2,400 | €0 |
| Total 3 years | €47,820 | €31,800 | €29,400 |
Custom hits break-even versus Airtable Business around year 2, and in year 3 you are in the green. Notion stays cheaper longer because the licence price is lower, but workaround hours pile up faster once volume grows.
How do you start concretely?
- Step 1: write your workflow on a single A4. No tool choice, just the process.
- Step 2: match with this guide. If it fits inside Airtable or Notion, start there.
- Step 3: build or have it built in 2 to 3 weeks. Start small, measure, expand.
Plan a free intake call. Our way of working sits in our terms and conditions.
