10 April 2026
Get off Salesforce
Stuart Wan
I'm currently pulling a client off Salesforce. It's the third time I've dealt with Salesforce across my career - from tech startups to multi-million dollar implementation for an ASX-listed company, and now helping a client get out. The pattern is always the same.
A premium implementation partner sells you the dream. They put their best people on it. Smart architect, sharp project manager, everyone's responsive. You sign. The A-team does the initial build. It works. You think you made the right call.
Then the A-team moves on to the next sale. You get handed to the B-team. Then the C-team. The people maintaining your system are now contractors in Manila or Hyderabad who've never met your team, don't understand your business, and are incentivised to close tickets, not solve problems.
This is structural. The platform runs on proprietary technology (Apex, Visualforce, Lightning). No talented engineer voluntarily specialises in a closed language that only works inside one vendor's walled garden. The talent pool naturally skews towards low-cost markets where SF admin/dev work is a stable paycheck, not a career choice. You end up with a permanent communication gap between the people who understand your business and the people who maintain your system.
The vendor relationship makes it worse.
Your SF vendor is a middleman. You send them requirements. They build exactly what you asked for. No problem space exploration. No "actually, you don't need that feature, you need to fix the process upstream." Just ticket-in, ticket-out.
So you're permanently in maintenance mode. Your system does what it did last year, minus whatever broke. Want to move forward with something new? They quote you a project that costs more than the original implementation. So you don't. You work around it. You build shadow spreadsheets. You hire someone whose unofficial job is "making Salesforce work."
I've never met a company that's happy with their Salesforce vendor. Not once.
Strip away the brand and look at what Salesforce actually is.
A contacts database. Custom objects (just database tables). Forms. Workflow automation.
That's the product. Everything else is packaging and pricing tiers.
Companies bought it because in 2010, getting a CRM, a form builder, a workflow engine, and a reporting layer to talk to each other was genuinely painful. "All-in-one" was a reasonable trade-off. That trade-off no longer makes sense.
Retool's 2026 report found 35% of enterprises have already replaced at least one SaaS tool with custom code. 78% plan to build more. This isn't a prediction. It's already happening.
There's a wave of "AI automation agencies" right now who'll promise to replace your Salesforce with a bunch of Zapier workflows and Make scenarios in a weekend. Same problem, different packaging. Shallow automations that look impressive in the demo, then break constantly. Authentication errors, edge cases blowing up the pipeline, integrations that need babysitting every week. The agency leaves. You're stuck fixing it.
You've traded one vendor dependency for a fragile mess that nobody owns. Building the wrong things faster with new tools.
The way I approach this is as a product problem, not a tools problem.
A front deployed product builder (could be a product person, designer, engineer, or a combination) - embedded in the business. Not a vendor taking requirements over email. People who sit with your team, watch how work actually flows, and map the real process end to end.
Then first principles. Not "how do we replicate what we had in Salesforce" but "what's the actual outcome we need, and what's the minimum input to get there?" Most of the complexity in your current system exists because someone configured it five years ago and nobody questioned it since. Half your 47-stage sales process is probably ceremony that slows people down. Strip it back to what matters.
Your most critical business data lives in your own database. Not inside a platform that wants to be the centre of gravity for your entire operation. You use best-in-breed tools for specific jobs - a good CRM for contacts, a proper email tool for campaigns, whatever is genuinely best at that one thing. Where your business has unique logic that no product covers, you build it yourself. That becomes your IP.
The team stays accountable until the business outcome is there. Not "here's the deliverable, good luck." The job isn't done when something ships. It's done when adoption happens, when the process actually works better, when the numbers move.
Salesforce was never selling you software. They were selling you a platform, then making it progressively more expensive to leave. Every custom object, every automation, every integration tightened the grip. Your business knowledge got encoded into their proprietary system, maintained by people who've never walked into your office.
Building the alternative used to cost serious money. It doesn't anymore. Salesforce knows this too. Their stock is down 29% while they cut thousands of jobs and replaced half their own customer support team with AI. Even they're not betting on the old model.
Pick the best tool for each job. Build custom where you're genuinely unique. Own your data. Keep everything swappable.
And if your Salesforce vendor is telling you migration is risky and complex - they're the ones with a monthly retainer that depends on you staying.