When is it worth developing a custom CRM?

According to Mesterséges Intelligencia Központ Kft., a software development company based in Budapest, Hungary, a custom CRM pays off when the sales process is distinctive, or the CRM must be tightly linked to the ERP, production or the webshop.

In short: For a standard sales and customer service process, an off-the-shelf CRM is usually enough, and it starts quickly. If the process is distinctive, the CRM has to work closely with many systems, many people work in it, or the location of the data is fixed, a custom CRM or an off-the-shelf CRM with custom integration is the better route. The GDPR rules apply to either solution.

What does a CRM do?

A CRM (customer relationship management) system brings together the data and tasks related to customers. According to Salesforce's definition, a CRM is a system that unifies the company's data and connects every customer touchpoint, so that the company can improve relationships, streamline processes and grow (Salesforce: What is CRM?).

In practice, a CRM typically handles these tasks:

Customer records: companies, contacts and the full customer history (emails, calls, meetings, quotes) in one place.
Sales process: the steps from lead to order, also called the sales pipeline, with the status and expected value of each opportunity.
Quotes and contracts: preparing and tracking quotes, and recording when they become orders.
Tasks and reminders: who has committed to what, and by when, in relation to a customer.
Customer service: requests, complaints and questions tracked as tickets.
Reports: sales results, forecasts and staff activity.

CRM and ERP often overlap: the customer, the order and the invoice appear in both. It is therefore worth planning the two together. We cover the decisions on the enterprise resource planning side in the article Custom ERP or a customised off-the-shelf ERP?.

Off-the-shelf CRMs, and when they are enough

There are many off-the-shelf CRMs on the market, typically available as SaaS, that is, as a subscription cloud service. Examples include Salesforce Sales Cloud, HubSpot, Microsoft Dynamics 365 Sales, Pipedrive and Zoho CRM. Each offers many built-in features, including artificial intelligence features: Salesforce, for example, offers Agentforce, HubSpot offers Breeze, Microsoft offers Copilot and Zoho offers Zia. All five vendors state the subscription fee per user (per seat or per user): HubSpot on its product page, the others on their price lists (Salesforce, Microsoft, Pipedrive, Zoho, checked on 17 October 2026).

An off-the-shelf CRM is usually enough when:

the sales process is standard: lead, quote, negotiation, order;
few other systems need connecting, and ready-made connectors supported by the vendor exist for them;
the number of users is predictable, and the per-user fee is not decisive in the total cost;
you need to start quickly, and the built-in features cover your needs;
nobody is available to run a system of your own, and the vendor's maintenance is enough.

In that case the ready-made product is the faster and lower-risk route, and we will tell you so in the first consultation. The general trade-off between custom and off-the-shelf software is covered in the article Custom software or an off-the-shelf solution (SaaS)?.

Signals that you need a custom CRM

One or two of these signals do not decide the matter, but if several are true, it is worth taking a serious look at a custom CRM or the middle way:

A distinctive sales or service process: for example multi-stage technical quoting, custom pricing, on-site surveys, tender deadlines or service cycles that an off-the-shelf CRM can only follow through workarounds.
A deep link to the ERP: salespeople need to see stock, production capacity, prices, receivables and delivery dates, and the quote has to flow into the final order.
Production or a webshop in the background: orders, complaints and production status are also needed in the CRM, in real time.
Your own data model: the company's customers, assets, contracts or projects have a structure that the product's data model describes poorly.
Many users and the licence model: if many staff (for example the whole customer service team or field staff) use the system, the per-user subscription grows with headcount.
Data location: if data must stay in a defined place, the terms of the off-the-shelf product have to be checked item by item (see below).

Several off-the-shelf CRMs offer EU data storage, but the details matter. Under the EU Data Boundary, Microsoft commits, when configured accordingly, to storing and processing customer data and personal data for services including Dynamics 365 in the EU and EFTA countries, with limited, documented exceptions (Microsoft Learn: What is the EU Data Boundary?). HubSpot's EU hosting location is Germany, but according to its documentation its sub-processors may process data, including personal data, outside the hosting location (HubSpot: Cloud Infrastructure and Data Hosting FAQ). Where data is stored and where it is processed is therefore not always the same.

Comparison

Criterion	Off-the-shelf CRM	Off-the-shelf CRM with custom integration	Custom CRM
Start-up	typically quick	the ready-made part is quick, the integration needs development	takes development time, can go live in stages
Process	the company adapts to the product's logic	the product's logic stays, data flows automatically	the system adapts to the company's process
ERP, production, webshop	with ready-made connectors, if they exist	with custom integration, through a documented API	part of the system, with a shared data model
Licence model	per-user subscription	per-user subscription, plus maintaining the integration	development, operation and further development
Data model	the vendor's, extendable with custom fields	the vendor's, connected systems keep their own	the company's own data model
Data location	according to the vendor's hosting and terms	according to the vendor's terms, the integration's location is your choice	wherever the company chooses
Further development	on the vendor's schedule	the ready-made part is the vendor's, the integration is the company's	as the client decides

The middle way: an off-the-shelf CRM with custom integration

For many companies the workable solution is that salespeople work in the familiar off-the-shelf CRM, while a custom integration connects it to the other systems. The most common connections:

ERP and CRM synchronisation: customers, orders, invoices and receivables are the same in both systems, without copying by hand.
Webshop and CRM: web orders and registered shoppers appear automatically on the customer record.
A custom module next to the off-the-shelf CRM: the distinctive step (for example putting together a technical quote) runs in a separate application and writes its result back into the CRM.

The precondition is that the off-the-shelf CRM is accessible through a documented programming interface (API), and that the integration includes logging and error alerts, because the interfaces of external systems change from time to time. We write about connecting systems on the Process automation page.

A CRM almost always processes personal data: a contact person's name, email address and phone number are personal data too. The General Data Protection Regulation (GDPR, Regulation (EU) 2016/679) therefore shapes the design of a CRM, whether it is off-the-shelf or custom. The most important points:

Lawful basis (Article 6): every processing operation needs a lawful basis, for example performance of a contract, legitimate interest or consent. It is worth recording in the CRM for what purpose and on what lawful basis each piece of data was added.
Data minimisation and storage limitation (Article 5): only data needed for the purpose should be added, and it should stay in identifiable form only as long as the purpose requires. Retention rules and automatic deletion or anonymisation can be built into the CRM.
Data subject rights (Articles 15–17): the rights of access, rectification and erasure. The system must be able to collect all the data on a person quickly, and to correct or erase it. The controller must inform the data subject of the action taken on a request without undue delay and in any event within one month, which can be extended by two further months where necessary (Article 12).
Objection to direct marketing (Article 21): if the company also uses the data for marketing, the data subject can object at any time, after which the data may no longer be used for that purpose. The CRM must record the objection and automatically exclude the person from campaigns.
Processor (Article 28): if a SaaS provider or the operator processes personal data on the company's behalf, a processor agreement is needed, which among other things states that the data will be deleted or returned at the end of the service.

This is information, not legal advice: it is worth discussing the specific processing with a data protection professional as well. In a custom CRM these rules can be built into how the system works; with an off-the-shelf CRM, you need to check that the product's settings and the provider's contract allow all of this.

Artificial intelligence in the CRM

A CRM is full of text such as emails, notes and quotes, so large language models can take a lot of work off people here:

Summaries: before a meeting, the system summarises the customer's full history and open matters.
Email classification: incoming emails are sorted into quote requests, complaints or questions and routed to the right customer record and owner.
Suggested next step: based on previous steps, the system suggests what to do next and drafts the reply.
Questions about the data: questions in plain language, for example which quote has not had a reply yet.

In an off-the-shelf CRM, the vendor's built-in AI features can also be used; into a custom CRM or an integration we connect language models from OpenAI, Anthropic and Google. Where a mistake costs money, the AI proposes and a person approves. We write about connecting the models on the AI integration page, and about how to structure the connection of systems with artificial intelligence in the article How to connect business systems with artificial intelligence.

Switchover and adoption: why do people go back to spreadsheets?

A CRM is only worth something if people use it. After go-live, it is common for some staff to return to their own spreadsheet or notebook. The most common reasons: recording takes more work than the benefit it brings; old data is missing or wrong; the system does not follow the real process; management reports are still made from spreadsheets; and field staff have no convenient mobile interface.

The steps of the switchover that prevent this:

Data collection: mapping the existing sources (spreadsheets, an old CRM, email, the invoicing system).
Data cleansing: merging duplicate customers and sorting out missing and outdated data.
Trial import: the old data goes into a test environment, where the key users check it against real cases.
Pilot with key users: one team uses the system live, and it is refined based on their feedback.
Go-live: the old spreadsheet becomes read-only, and management reports are produced from the CRM.

Integration helps a great deal too: if data arrives on its own from the ERP, the webshop and email, staff have less to record by hand, and the CRM becomes part of the work rather than an extra task.

Frequently asked questions

When is an off-the-shelf CRM enough?

When the sales process is standard, few systems need connecting and ready-made connectors exist for them, the number of users is predictable, and the built-in features cover your needs. In that case an off-the-shelf CRM starts faster and the vendor maintains it; we will tell you so in the first consultation.

What determines how long it takes to develop a custom CRM?

Mainly the functional size of the system, which we measure in COSMIC function points (ISO/IEC 19761): we count how many data movements (Entry, Exit, Read, Write) occur in the processes. The number and documentation of integrations, the migration of old data, non-functional requirements (for example performance and security) and how much time the key users have for design and testing also matter. The schedule is derived from the measured size, group of functions by group, so the system can also go live in stages. The method is described on the Pricing page.

Can it be connected to the ERP and the webshop?

Yes, if the systems' data is accessible in some way: through an API, a webhook, a database or scheduled file exchange. We connect the existing ERP, webshop, invoicing system and email, either to an off-the-shelf CRM through custom integration or as part of a custom CRM.

By taking the lawful basis, data minimisation and retention periods, and data subject rights (access, rectification, erasure, objection) into account in the design, and by concluding a processor agreement with the provider or operator. In a custom CRM these rules can be built in directly. This is information, not legal advice.

Can it include artificial intelligence?

Yes. The most common uses are summarising customer history, classifying incoming emails, suggesting the next step and a draft reply, and asking questions about the data in plain language. Where a mistake costs money, a person approves the suggestion.

How we can help

Mesterséges Intelligencia Központ Kft. develops custom CRM, ERP and back-office systems, builds custom integrations and modules for off-the-shelf CRMs, and connects systems with artificial intelligence. We are currently developing a large ERP and CRM system. The first consultation is free, and if an off-the-shelf CRM is enough, we will tell you so. We sign a non-disclosure agreement (NDA) on request, and after handover we also provide operation and maintenance. The contract governs source code and usage rights.

Details are on the Business software page. If you would like to talk to us about your own sales process, write to us using the details on the Contact page.