The short answer
In the custom CRM vs Salesforce vs HubSpot decision, most companies should buy: the platforms deliver pipeline management, email and calendar sync, mobile, permissions and reporting on day one, and no custom build recovers that head start on a standard sales motion. Building wins under five conditions — the process being managed is itself the competitive advantage, per-seat licensing at your headcount exceeds a three-year build-and-run budget, you need data residency the vendor cannot provide, your unit of work does not fit the account-contact-opportunity model, or spend on consultants, add-on SKUs and middleware already exceeds the licence. Absent one of those, buy the platform and spend the engineering budget where customers can see it.
Custom CRM vs Salesforce: why buying usually wins
The custom CRM vs Salesforce question usually resolves in favour of Salesforce or HubSpot, and anyone selling you engineering should say so first. A mainstream sales CRM is a solved product. Contacts, pipeline stages, activity logging, two-way email and calendar sync, offline mobile apps, role-based permissions, territory rules, forecasting, duplicate detection, bulk import, audit history and hundreds of connectors amount to fifteen years of product work, and they arrive the day you sign.
The failure mode of a custom CRM is rarely the initial build. Teams get a working system in a few months and are pleased with it. The failure comes two years later: the engineer who built it has moved on, mobile never happened, nobody implemented deduplication, reporting means asking someone to write SQL, and the sales team has drifted back to spreadsheets. A platform has a product team improving it regardless of your attention. A custom CRM gets whatever attention survives the competition with revenue work.
What does a platform CRM actually cost once it is running?
The licence is the visible number and rarely the largest. Salesforce publishes Sales Cloud in tiers: entry pricing around $25 per user per month, mid tiers around $100, enterprise and unlimited from roughly $165 up to several hundred, billed annually. HubSpot Sales Hub sits in a comparable band for Professional and Enterprise seats, with one structural difference — its marketing side prices on marketing contacts rather than seats, so that bill grows with your database. Both vendors change pricing regularly; verify current figures before modelling on them.
What sits underneath the licence is where budgets go. Capabilities that feel core often sit in separate SKUs — configure-price-quote, advanced forecasting, field service, extra sandboxes, additional API capacity. Then the human layer: an administrator or agency retainer, plus consultant days for every non-trivial change. Certified platform consultants transact in a wide market band, roughly $80 to $250 per hour by region and specialism, and worthwhile changes take more days than the requester expects.
There is also a maintenance tax specific to platforms. Salesforce ships three seasonal releases a year, and each is an opportunity for a customisation, a managed package or an integration to break. That is the price of a product that keeps improving, but it means a heavily customised org needs regression testing on a schedule the vendor sets.
The five conditions under which building a custom CRM wins
Building is right when one of these is clearly true. Not two of them slightly.
1. The process is your competitive advantage
If the way you qualify, price, schedule or fulfil is why customers choose you, encoding it in someone else’s object model puts a ceiling on it. A freight brokerage whose margin comes from matching loads to carriers, a lender whose edge is its credit workflow, a clinical group whose scheduling logic is the product — these are not managing contacts, they are running an operating system for the business. Bending a bought CRM flattens that advantage toward the industry average.
2. Per-seat economics have already tipped
Per-seat licensing is linear; a build is not. Work an example: 400 users at $150 per seat per month is $720,000 a year, before SKUs, admin salaries and consultants. Against that, a custom operational CRM in the $250,000 to $500,000 market range with maintenance at 15–25% per year looks very different by year two. The arithmetic only works when seat counts are high and per-seat value is low — hundreds of operational users each touching three screens. Fifty quota-carrying salespeople using everything the platform offers are the opposite case.
3. Data residency the vendor cannot offer
Some constraints are not negotiable with a SaaS vendor: a hosting region it does not operate in, deployment inside a customer’s own tenancy, an air-gapped environment, a regulator requiring the controller to hold the keys in a particular jurisdiction. Both vendors have expanded their regional and encryption options, so check before assuming — but where the requirement cannot be met, the decision is made for you.
4. Your unit of work is not an opportunity
Platform CRMs are built around account, contact and opportunity, with a deal progressing through stages toward a close date. If your unit of work is a shipment, a case file, a patient episode, a vessel or a loan tranche with several parties and a looping lifecycle, you will spend your life mapping it onto objects designed for something else. Custom objects help and are worth trying first. But when every report needs a workaround and every new hire must be taught what the Opportunity record really means here, the model is fighting you.
5. Customisation spend has already overtaken the licence
The most reliable signal, because it needs no forecasting — only last year’s invoices. Add up licences, SKUs, middleware, admin salaries and consultant fees. If the non-licence portion is larger and growing, you are already paying to build software, inside a container that constrains what you can build, on a codebase you cannot fully test.
Custom CRM vs Salesforce vs HubSpot compared
| Dimension | Custom build | Salesforce | HubSpot |
|---|---|---|---|
| Upfront cost | $60,000–$150,000 for one team’s workflow; $150,000–$500,000+ multi-team | Low licence entry, but enterprise implementations run to five or six figures | Lowest barrier; a small team is productive without an implementation partner |
| Cost at scale | Flat in headcount; grows with scope and maintenance, not users | Linear in seats and SKUs; enterprise tiers plus CPQ and sandboxes compound | Linear in seats, and in marketing contacts, so the bill tracks database growth |
| Time to value | Three to nine months before anyone logs a real record | Days for a basic org; months for a configured enterprise implementation | Fastest — a working pipeline in an afternoon, real adoption in weeks |
| Flexibility ceiling | None architecturally; the limit is budget and appetite for maintenance | High via Apex, Lightning Web Components and Flow, but bounded by governor limits | Moderate; custom objects gated to higher tiers, deep logic hits limits sooner |
| Operating burden | You own uptime, patching, backups, support and the roadmap | Needs an administrator or partner, plus regression testing three times a year | Lightest; often managed by an operations generalist |
| Who it suits | Firms whose workflow is the product, or with high operational seat counts | Complex sales motions, heavy territory and approval structures, partner ecosystems | SMB and mid-market teams valuing speed and adoption over depth |
Read the table by column. Salesforce wins where organisational complexity is high and the ecosystem is worth paying for. HubSpot wins where adoption speed beats configurability, which is more often than buyers admit. A custom build wins only where a structural constraint makes the other two columns expensive or impossible.
What does customising a platform CRM really cost?
Platform customisation is software development wearing different clothes. Apex classes, triggers, Lightning Web Components, Flow orchestrations and managed packages all need design, review, testing and release management, and they carry constraints ordinary code does not — governor limits on queries and CPU time per transaction, daily API allocations, and a deployment model built on sandboxes and a metadata pipeline.
The costs that surprise people are recurring rather than upfront. A customised org needs someone who understands it, either a salaried administrator or a partner on retainer, and every seasonal release needs regression testing. Managed packages introduce upgrade dependencies you do not control, so a package that goes unmaintained becomes a migration project with no budget line. Middleware adds a third vendor to every incident.
None of this argues against customising — a well-run Salesforce org that automates approvals and quoting can be the highest-return software in a company. It argues for counting it honestly. The fair comparison is not licence against build cost: it is licence plus SKUs plus admin plus consultants plus middleware plus release testing, against build plus maintenance plus hosting plus your own support rotation.
annualPlatformCost =
seats × pricePerSeatPerMonth × 12
+ addOnSKUs
+ adminSalary (loaded)
+ consultantDays × dayRate
+ middlewareAndConnectors
annualCustomCost =
maintenance (≈15–25% of build)
+ hosting
+ supportRotation
breakEvenYears = buildCost / (annualPlatformCost − annualCustomCost)If break-even lands beyond four or five years, buy. Software has an economic life, teams change, and a payback horizon longer than your planning horizon is not a payback.
What you must build yourself in a custom CRM
Custom CRM estimates come in low because people scope the domain logic and forget the platform underneath it. Pipeline stages and a record view are a fortnight. The rest is the product.
- Two-way email and calendar sync against the Microsoft Graph and Gmail APIs, including threading, permissions and users who revoke consent.
- A permissions model with record-level access, team hierarchies and field-level visibility, plus SSO and ideally SCIM provisioning.
- Deduplication, fuzzy matching and a merge flow that preserves history — harder than it sounds, and where data quality is won or lost.
- Import, export and bulk edit, because operations teams will not adopt a system they cannot correct in bulk.
- Search that finds the right record from a partial name, and reporting a non-engineer can drive without SQL.
- Mobile, an audit trail, notifications, and backups with a tested restore path.
Observed market ranges reflect that scope. A focused operational CRM replacing one team’s workflow, integrated with a couple of existing systems, sits in the $60,000 to $150,000 band. A multi-team system with quoting, forecasting, mobile, several integrations and a real permissions model runs from $150,000 to $500,000 and upward, with maintenance at 15–25% of build per year. Those are market observations, not quotes, and they move sharply with integration count and compliance.
The hybrid most companies should actually pick
The build-or-buy framing is a false binary, and the answer that works most often is neither column. Keep the platform as the system of record for what is standard — contacts, companies, pipeline, email, marketing — and build only the differentiated operational layer it models badly. A logistics business keeps HubSpot for the commercial relationship and builds load-matching and settlement; a lender keeps Salesforce for origination and builds the underwriting workflow.
The discipline that makes this work is deciding, field by field, which system owns the truth. Bidirectional synchronisation where both sides can write the same field is the commonest source of data corruption in integrated stacks: two systems with different validation rules will eventually disagree, and neither will be wrong locally. Give each field one owner, propagate changes one way with webhooks or change data capture, use idempotent writes with a stable external identifier, and reconcile on a schedule so drift is detected rather than found by a customer.
It also preserves option value. A hybrid shows you whether the custom layer earns its keep before you commit to replacing the platform, and if it does, you already own the hard part.
How to decide without a six-month evaluation
- Model three years of total cost each way: admin headcount, consultants, SKUs and middleware on the platform side; maintenance, hosting and support on the build side.
- Write down your top ten objects and how they relate. If more than two have no natural home in account, contact and opportunity, the platform will fight you.
- Split last year’s actual spend into licence and non-licence. If non-licence is larger and growing, you are already funding a build.
- Run a two-week configuration spike against your three most awkward workflows. It costs almost nothing and settles arguments faster than any vendor demonstration.
- Only then choose, and if the answer is a hybrid, name the owning system for every shared field before writing integration code.
Most companies that run that sequence buy, configure carefully and get on with their business, which is the correct outcome. The ones that build come out of it with a clear sentence explaining why, and that sentence is the best predictor of whether the project succeeds. If you want the numbers pressure-tested by people who have built both custom systems and the integrations into the platforms, ApexStack is happy to walk through it with you.
Frequently asked questions
- Is it cheaper to build a custom CRM than to pay for Salesforce?
- Only at high seat counts or heavy customisation. Salesforce cost is linear in seats while a build is largely flat in users, so the crossover depends on headcount and on how much you spend on add-on SKUs, administrators and consultants. Model three years of total cost on both sides, including maintenance at roughly 15–25% of build per year. If break-even falls beyond four or five years, buying is the better financial decision.
- When should a company build its own CRM instead of buying one?
- Build when one condition is clearly true: the workflow being managed is your competitive advantage, per-seat licensing at your headcount already exceeds a realistic three-year build-and-run budget, you have a data residency or isolation requirement the vendor cannot meet, your core unit of work does not fit the account-contact-opportunity model, or your spend on consultants and integrations has already overtaken the licence. Absent one of those, buy.
- What is the difference between Salesforce and HubSpot for a growing company?
- Salesforce offers greater depth — territory management, complex approvals, configure-price-quote, an extensive partner ecosystem — at the cost of needing an administrator and a longer implementation. HubSpot is faster to adopt and lighter to run, which matters more than most buyers expect, but its customisation ceiling arrives sooner and its marketing pricing scales with contact volume rather than seats, so the bill grows with your database.
- How much does a custom CRM cost to build?
- Observed market ranges run from about $60,000 to $150,000 for a focused system replacing one team’s workflow with a couple of integrations, and from $150,000 to $500,000 or more for a multi-team platform with quoting, forecasting, mobile access and a real permissions model. Budget maintenance at 15–25% of build cost per year. Integration count and compliance requirements move those figures more than anything else.
- Can you use Salesforce and a custom system together?
- Yes, and it is usually the best answer. Keep the platform as the system of record for contacts, pipeline and email, and build only the differentiated operational layer it models badly. The discipline that makes it work is assigning exactly one owning system per shared field, propagating changes in one direction with webhooks or change data capture, and reconciling on a schedule so drift is caught before a customer finds it.
- Why do custom CRM projects fail?
- Rarely at launch and almost always in year two. The initial build works, then the engineer who wrote it moves on, mobile access never arrives, deduplication was never implemented, reporting requires someone to write SQL, and the sales team drifts back to spreadsheets. A platform has a product team improving it regardless of your attention; a custom CRM only gets whatever attention survives the competition with revenue-generating work.
Talking about crm & erp systems?
Custom or heavily tailored operational platforms for businesses whose processes do not fit an off-the-shelf product.