What the two terms actually mean
A SaaS CRM is software you subscribe to: the vendor hosts it, updates it and holds your data on their infrastructure, and you configure it within the limits they set. A self-hosted CRM runs on infrastructure you control — your own servers, or a cloud account in your name — and whoever operates it is responsible for hosting, updates, backups and security. Custom-built CRMs are nearly always self-hosted in this sense; so are some open-source products.
The comparison, factor by factor
| Factor | SaaS CRM | Self-hosted CRM |
|---|---|---|
| Cost shape | Recurring per-seat or per-tier fees that grow with headcount | Upfront build or setup, then hosting and maintenance |
| Time to start | Days: sign up and configure | Weeks to months, depending on how much is built |
| Where data lives | The vendor's infrastructure, in regions they offer | Infrastructure you choose and control |
| Fit to your workflow | Configurable within the product's model | As close as you are willing to build |
| Updates and fixes | The vendor ships them on its own schedule | You schedule them, or whoever maintains it does |
| Security and backups | Mostly the vendor's responsibility, within a shared model | Yours — or your operator's — in full |
| Leaving | Export what the vendor allows, in the format it allows | You hold the database and the code |
| Compliance evidence | Vendor certifications and a data processing agreement | You produce the evidence yourself |
The cost nobody puts in the spreadsheet
Self-hosting does not remove the work the vendor was doing; it moves it to you. Someone has to apply security patches, rotate credentials, test restores from backups, watch for failures and answer when it breaks. If that someone is a named person with the time and the skill, self-hosting is a sound choice. If it is "IT, when they get a moment," it is a risk you have not priced.
SaaS has its own unpriced cost: the workaround layer. When the product almost fits your process, the gap fills up with spreadsheets, manual re-keying and one person who knows how it all connects. That cost is real, and it grows with the business.
When each one wins
- SaaS wins when your process is close to standard, you want to start this week, and nobody on your side wants to operate software.
- Self-hosted wins when client data must stay on infrastructure you control, when licence fees would grow faster than the value you get from them, or when your workflow is unusual enough that configuration cannot reach it.
- A hybrid is often right: start on SaaS to learn what you actually need, keep your data exportable from day one, and revisit when the workarounds start costing more than a build.
The questions that decide it
- Who, by name, will patch, back up and monitor this a year from now?
- Is there a legal, contractual or client requirement about where this data lives?
- How many seats will you have in three years, and what does the per-seat price do to the total?
- How different is your real process from the product's model, measured in workarounds today?
- If you left in two years, what would you be able to take with you?
If most of the answers point one way, that is your decision. If they split, buy SaaS first: it is the cheaper experiment. When a custom build is the better path I will say so, and when a subscription is, I will say that instead.
Where this applies
More notes

