Custom CRM Software Development, Against a Published Licence

Most comparison articles on custom CRM software development start with the build and treat the licence as the alternative. The arithmetic runs the other way. The shelf is the known cost. The build is the unknown cost, and it is the one that decides whether the project pays back.

This page works through the numbers you can check today, then sets out the three situations where a build genuinely earns its keep, and the much more common reason teams start a build that never finishes.

The published seat prices, side by side

Four products publish per-seat prices a procurement officer can copy into a spreadsheet. The figures below are taken from each vendor's own pricing page, checked September 2026, and printed as USD per seat per month billed annually unless the vendor says otherwise.

The Asyntai row is ours, so read it as a disclosure. Asyntai charges by messages answered for the whole site, not per seat, which is why a ten-seat sales team is not the natural comparison on this page. The CRM rows above are the ones a custom CRM build is usually weighed against.

Two things to notice before any arithmetic. First, HubSpot charges a one-time onboarding fee on top of the seat price: $1,500 for Professional and $3,500 for Enterprise. That changes the first-year number. Second, Pipedrive and HubSpot price by seat while Zoho and Salesforce price by user, and the two are not quite the same thing in practice. A seat you cannot add a colleague to is cheaper on paper than a user licence the vendor counts on a per-login basis.

The ten-seat team, on each published ladder

The cleanest way to put a number on the alternative is to assume a fixed team size, then read the annual cost straight off each published ladder. Ten is a common size for a sales team that has outgrown a spreadsheet but not yet earned its own platform engineer.

That is the published cost of the alternative. The build quote for the custom CRM is not in any of these tables, because no builder publishes one. It has to be asked for.

What a custom CRM build actually costs

The honest answer is that no public number exists. A small build by a freelancer, two months of work to ship a working first version, is commonly pitched somewhere in the tens of thousands of dollars. A build by a software agency with a project manager, a designer, a back-end developer and a QA engineer is commonly pitched somewhere in the low six figures. None of those figures is in any pricing page, which is the first thing to be clear about.

What can be said from the published prices is what the build has to beat over its life. A Salesforce Core licence for ten users is $23,400 in year one. Over three years, with no price rises, that is $70,200. A Zoho Professional licence for ten users is $2,760 in year one, which is $8,280 over three years. The build has to come in under those numbers, plus the years of developer time you keep paying after launch, or it does not pay back.

Put simply: the shelf is cheap, and the build is expensive. A custom CRM only wins when the shelf genuinely cannot do the job.

The three situations where a build is justified

Most lists of reasons to build a custom CRM are padded. Here are the three that genuinely push a team off the shelf.

1. The workflow cannot be drawn. Some processes are genuinely strange. A brokerage that prices each deal against five live feeds and routes it through one of three desks depending on the counterparty. A wholesaler whose pricing depends on the buyer's last twelve purchases and the warehouse the goods leave from. A field-services company whose quote turns on the engineer's first available slot. Salesforce and Pipedrive can be bent into many of these shapes, but the bend reaches a point where the configuration cost is most of a build anyway, and the result is harder to hand to the next administrator than a written application would be.

2. The data must live on your side. Regulated industries, defence suppliers, healthcare networks and some financial services firms cannot put customer data on a vendor's cloud, even one that is SOC2 audited and HIPAA-aligned. A custom build on owned infrastructure, with the documents to prove it, is the answer those buyers are actually buying. The shelf is not an option here.

3. The shelf is going to charge more next year than the build will. This is rarer than the sales pitch suggests, but real. A team on a usage-based price that grows with contacts, emails or API calls can outrun the cost of a build that holds its headcount flat. The arithmetic has to be done on the vendor's own published tier ladder, with the team's actual growth assumption written in. A claim that "Salesforce will cost us $X over five years" is only as good as the price ladder it was built on.

The far more common reason a build starts

The reason most custom CRM projects actually begin is none of the three above. It is a process nobody has written down yet.

A team goes looking for a CRM, finds the shelf, cannot quite match it to the way the business actually works, and decides the gap is the vendor's fault. The gap is usually their own. The steps a deal goes through, who has to approve what, which fields matter and which are noise, what counts as a qualified opportunity and what does not, are decisions that have not been made. They have been carried in the heads of two or three senior people, and nobody has written them down.

A build does not fix that. A build hides it. The developer takes the verbal process and turns it into a feature, the feature ships, and the team uses it for six months, and then someone changes the process and the build does not change with them. The shelf would have done the same job with a workflow that could be edited in an afternoon.

The honest test before a build starts is whether the team can write down, on one page, the lifecycle of a deal from first touch to close, with the fields and approvals that matter and the ones that do not. If they cannot, the problem is the process, not the software.

What the shelf actually gives you, in plain language

The four products above are not the same shape, and it is worth knowing what each one is for.

Zoho CRM is the broadest for the money. The free edition covers three users, the paid ladders run from Standard at $14 per user per month through Professional at $23 to Enterprise at $40 and Ultimate at $52, all billed annually. It also sells the same tiers billed monthly, at higher headline rates of $20, $35, $50 and $65. For a ten-seat team on the Professional tier billed annually the figure is $2,760 per year. For a ten-seat team on the Professional tier billed monthly it is $4,200 per year. The annual saving on ten seats at that tier comes from paying annually instead of monthly, and the size of that saving is worth working out on your own seat count before procurement picks the payment cadence.

Pipedrive is simpler and pipeline-shaped. The tiers are Lite at $14 per seat per month, Growth at $39, Premium at $59 and Ultimate at $79, all billed annually. There is no free tier, only a 14-day trial. LeadBooster and other add-ons are priced separately and sit on top of the seat price. For a ten-seat team on Premium that is $7,080 per year, and the only number on top of it is whatever the add-ons cost.

HubSpot Sales Hub is the one with the onboarding fee. Free tools are $0 per seat for up to 2 users. Starter is $7 per seat per month billed annually or $20 billed monthly, with 500 credits. Professional is $90 per seat per month billed annually or $100 billed monthly, with 3,000 credits, plus a required one-time $1,500 onboarding fee. Enterprise is $150 per seat per month with 5,000 credits, plus a required one-time $3,500 onboarding fee. Extra credits on annual billing are $0.01 each. For a ten-seat team on Professional in year one the seat bill is $10,800 across the year, plus $1,500 in onboarding, so $12,300 in year one and $10,800 every year after.

Salesforce Sales Cloud is the most expensive and the most configurable. Free Suite is $0 per user. Starter Suite is $25 per user per month. Pro Suite is $100 per user per month billed annually. Core is $195 per user per month billed annually. Advanced is $395 per user per month billed annually. Max is $550 per user per month billed annually. AI can be added to Core and above. For a ten-seat team on Pro Suite the figure is $12,000 per year. On Core it is $23,400 per year, which is the number a build has to beat over its life to make sense.

The onboarding fees, hidden in plain sight

HubSpot is the only one of the four that charges an onboarding fee on top of the seat price, and it is worth its own paragraph. The fee is $1,500 on Professional and $3,500 on Enterprise, both required, both one-time. For a ten-seat team on Enterprise in year one the seat bill is $18,000 across the year, plus $3,500 in onboarding, so the year-one total is $18,000 + $3,500 = $21,500, before any add-ons or extra credits. That is the headline number to use when comparing a build against HubSpot, because the build quote will not include a comparable fee.

Zoho, Pipedrive and Salesforce do not publish a comparable fee on their pricing pages, which is not the same as saying the on-call cost is zero. Implementation partners charge for their time. The difference is that those costs are quoted by the partner and not printed on the vendor's page, so they do not appear in the headline comparison.

The freelance build, and what it leaves out

A common route to a custom CRM is a freelancer or small studio. Two months of work to ship a working first version is a reasonable starting assumption. The build usually covers a sales pipeline, a contact record, a few custom fields, an email integration and a basic reporting screen. It does not usually cover:

• Mobile. A usable phone interface, with offline edits that resolve against the server later, is its own project.
• Integrations with the rest of the stack. The CRM does not live alone. It needs to talk to the marketing platform, the support inbox, the accounting system and the calendar. Each integration is its own small project.
• Permissioning that holds up under audit. A sales team that needs to see some records and not others, with the audit trail to prove it, is more code than most freelance builds ship.
• Reporting that the board will accept. A funnel report and a forecast by rep are not the same thing, and the second is harder than it looks.
• Ongoing maintenance. The freelancer ships the first version and disappears. Someone has to keep the lights on for the next three years.

None of these is a reason not to build. They are reasons to write the brief so the build includes them, and to budget for them, because the published shelf prices include them and the freelancer's quote usually does not.

What changes the answer

Three inputs move the result more than any other.

Team size. Per-seat pricing scales linearly with team size, and the build's ongoing cost does not. A team of fifty on Salesforce Core pays 50 × $195 × 12 months per year, and the same team on a build that holds its headcount flat looks very different by year three. The build only wins past a team size that has to be worked out on the user's own numbers, with the vendor's published tier as the input.

Process complexity. A two-stage pipeline is on every shelf. A twelve-stage pipeline with branching rules, regional desks and conditional approvals is on none of them, and the closer you get to that shape the more the build is doing work the shelf cannot. The honest test is whether the team's process fits on one page. If it does not, the shelf will struggle.

Data residency. A regulated buyer cannot put customer records on a vendor's cloud, and that removes the shelf from the conversation. The build is the only answer, and the build's price is the price of doing business in that market.

The mistake most teams make

The mistake is treating the licence as the cost and the build as the investment. The licence is the cost, and the build is also the cost. The difference is that the licence is published and the build is not.

A team that runs the arithmetic on Salesforce Core for ten users over three years and gets to $70,200 has a number to beat. A team that runs the arithmetic on a build and arrives at a quote from one builder has one data point, and that data point is the number the builder wants to win the work at. The honest comparison needs at least two build quotes, with the same brief, written the same way, before either is a real number.

It also needs the maintenance figure written in. A build that costs less than the shelf in year one and the same in year two will cost more in year three, four and five, because the shelf keeps getting updated and the build does not. The published prices are for this year's shelf. The build price has to be projected forward the same number of years, with a maintenance line.

When a build genuinely earns its keep

There is a short list of situations where the arithmetic favours the build, and it is shorter than the sales pitch suggests.

• The workflow cannot be drawn on the shelf. Tested by writing it on one page. If the page does not fit, the shelf cannot hold it.
• The data must live on your side. Tested by the regulator's rule, not by the vendor's whitepaper.
• The shelf will outprice the build over the project's life. Tested on the vendor's published tier ladder, with the team's growth assumption written in, and the build's maintenance written in next to it.

Outside those three, the shelf wins, and the question to ask is not which shelf but which tier. A ten-seat team on Zoho Professional at $23 per user per month billed annually is $2,760 per year, which is a number a build has to beat over its life to make sense. A ten-seat team on Pipedrive Premium at $59 per seat per month billed annually is $7,080 per year, and the same test applies.

A worked example, on Asyntai's own numbers

Asyntai is an AI chat agent that answers visitor questions on a website, priced by messages answered rather than by seat. It is in this article only because the comparison is the reader's, and a reader weighing a custom CRM against a shelf is also weighing whether some of the work the CRM is doing should be done somewhere else entirely.

The arithmetic is short. Asyntai Starter is $39 per month for 2,500 messages. Standard is $139 per month for 15,000 messages. Pro is $449 per month for 50,000 messages. None of those is a per-seat number, because Asyntai is priced by messages answered for the whole site.

The point is not that one is cheaper than the other. They are different products doing different jobs. The point is that some of the work a custom CRM is being scoped to do, namely capturing the visitor who lands on the website with a question the sales team will answer three days later, is work that is already done elsewhere for less than the cheapest CRM seat. A custom CRM that includes a chat layer is competing with both the shelf CRM and a conversational AI layer that captures visitors for $1,668 per year for the whole site, and that competition is worth running before the brief goes out.

How to run the comparison honestly

Three steps, in order.

1. Write the workflow on one page. The lifecycle of a deal from first touch to close, with the fields and approvals that matter and the ones that do not. If the page does not fit, the shelf will struggle and the build is the only answer. If it fits, the shelf is in the running.
2. Get two build quotes on the same brief. One quote is a number a builder wants to win the work at. Two quotes on the same brief, written the same way, are a market. The shelf's published tier is the third number.
3. Project all three over the project's life. Three years is a fair horizon for a small build, five for a larger one. The shelf's price has to be projected at the vendor's published rate, with the team's growth assumption written in. The build's price has to be projected with a maintenance line, because the shelf keeps getting updated and the build does not.

The shelf wins on published numbers most of the time. A custom build wins when the workflow genuinely cannot be drawn, or when the data has to live on your side, or when the shelf will outprice the build over the project's life. The most common reason a build starts is none of those, and the most common reason it finishes over budget is that the brief was the team's verbal process and not a written one.