---
title: "Custom CRM development"
description: "Custom CRM development: a CRM built around how your business sells and delivers, not a generic tool your team works around. Scoped by a £2,000 audit."
canonical: https://agmm.co.uk/services/custom-crm-development/
updated: 2026-10-13
updated_basis: the date the page itself states
publisher: AGMM (https://agmm.co.uk/about/agmm.json)
lang: en-GB
---

[All services](https://agmm.co.uk/services/)

Customer systems

# A CRM shaped around your business, not the other way round.

A custom CRM is built around how your business sells and delivers, so the team keeps it up to date and the numbers can be trusted. AGMM starts with a fixed £2,000, 14-day audit to decide whether to extend, replace or rebuild, then prices any build by proposal.

[Book the free 30-minute fit call](https://agmm.co.uk/ai-constraint-audit/#book) [See every department](https://agmm.co.uk/services/#departments)

Or call [0161 543 7189](tel:+441615437189)

By [Sam Wall](https://agmm.co.uk/about/sam-wall/) and [Adam Gillett](https://agmm.co.uk/about/adam-gillett/) Checked on 13 October 2026

Two of the papers this work handles. Demonstration with sample documents. In the Audit we draw your real lines.

Customer record Existing customer

Fenwick Group

Facilities management, 60 staff

**Owner**: A. Khan

**Last order**: March 2026

**Open work**: 1 proposal

**Contact**: J. Moore, Operations

1. 12 Oct Enquiry from the website form
2. 3 Sep Call with J. Moore
3. Mar 2026 Order delivered

The problem

## The CRM describes a business you do not run, so nobody keeps it up to date, and the numbers in it cannot be relied on.

Off-the-shelf CRMs assume a sales process. When yours does not match, the team invents workarounds, the data stops being trustworthy, and the reporting stops being worth reading. A custom system starts from how your business works.

The failure is quiet and it compounds. Somebody stops trusting a field, so they keep their own note. A second person sees the CRM is wrong and stops updating it too. Within a year the system holds a version of the business nobody recognises, and the reporting built on top of it is being reconciled by hand before anyone will present it. By then the argument has moved on from the software to the people, which is unfair to both. The useful question is not whether the team is disciplined. It is whether keeping the record accurate is extra work, because if it is, it will not happen, and no amount of training changes that.

Now picture yours

## The same work, three ways.

Pick a line of work, press a stage, then step through the papers it handles. Each step shows the real document, and where a person stays in charge.

Sales Every one in one place

Email Mon 12 Oct, 16:48

**From**: Fenwick Group, website form

**Subject**: **Enquiry: help with our onboarding process**

We have grown to 60 people and onboarding is all email.

Could someone call us this week?

1. From email, the website, the phone or a referral, into one place.
2. Linked to everything you already know about them.
3. Every enquiry gets an owner and a next step, so none sits idle.
4. AI drafts, classifies or extracts. A person approves anything that leaves the building or changes a commercial position.
5. Sent, and the record updates itself.

1. **As it runs now**

   Enquiries in more than one place, a private spreadsheet beside the CRM, and a report nobody trusts until somebody reconciles two systems by hand.
2. **Organised**

   Customers, enquiries and follow-ups in one view, with stages that match how you sell and deliver. The data stays right because keeping it right is not extra work.
3. **Agentified**

   Enquiries file themselves against the right customer, every one gets a next action, and a person approves anything that leaves the building.

*A person decides Runs on its own Back with notes Demonstration. Sample documents, no client data. Timings compressed.*

Signs you have it

You will recognise some of these.

- Your team keeps a private spreadsheet because the CRM does not fit.
- Enquiries arrive in more than one place and no single view shows all of them.
- Reporting needs somebody to reconcile two systems by hand first.
- You pay per seat for features nobody uses, and lack the one thing you need.
- Nobody can say with confidence which enquiries have not been followed up.

What we build

## What does the work include?

1. ### One place for customers and enquiries

   Customers, enquiries, follow-ups and the work your team needs to do next, in one view rather than four.
2. ### Your pipeline, your stages

   The stages match how you sell and deliver, so the data stays accurate because keeping it accurate is not extra work.
3. ### Connected to the rest of your tools

   Enquiries, email, quotes and job information flow in, so the CRM reflects reality without anyone copying between systems.
4. ### Reporting you can act on

   Built around the questions you ask, rather than a dashboard of everything the software can measure.

What it costs

## What does it cost? The audit is £2,000; the build is proposed after it

**1The AI Constraint Audit**: **£2,000** 14 days, fixed

**2The build**: **By proposal** Priced against what the audit found

**3Ongoing support**: **By proposal** Only if you want it

The only price published on this page is the diagnostic: £2,000 for the AI Constraint Audit, fourteen days, three sixty-minute working sessions. A CRM build is proposed after it and priced in its own document. The order runs that way because the expensive part of a custom CRM is rarely the screens. It is how many kinds of user it has to serve, how many systems it has to stay in step with, and the state of the data being moved into it. The audit fee is not credited against the build, deliberately, so nothing about the recommendation is paid for by the thing it recommends.

## What does the £2,000 cover before anything is built?

Four documents that are yours whatever you decide next: a business-wide constraint map, an AI opportunity map, an economic case for each option worth considering, and a ranked order for spending. On a CRM the useful part is usually the economic case, because it establishes whether the mismatch between your process and your current tool is costing you, or whether it is an irritation a configuration change would settle for a fraction of the money. If it is the second, you keep the documents and we have not sold you a build.

The whole ladder, including what a build costs and what we do not bill for, is on [what it costs](https://agmm.co.uk/what-it-costs/) .

How long

## How long does it take?

Fourteen days from the first session to the written recommendation. The build that may follow runs in stages, so a first useful part is in real use early rather than everything landing at the end, which is also how you find out whether the shape is right while it is still cheap to change. Data migration is the stage that most often sets the pace, because the time it takes is decided by the state of your records rather than by us.

1. **Day 1** First working session
2. **Day 14** Written recommendation, yours to keep
3. **Then** First stage of the build, in real use
4. **Then** Next stage, only once the first is working

Not in the price

## What this does not include

- ### Licences for the platforms you keep

   We do not resell seats and we take no margin on any platform. If keeping HubSpot, Salesforce or your accounts package for part of the job is the right answer, we will say so and you will carry on buying it directly.
- ### Support and hosting, inside the build price

   Software is an asset with running costs rather than a purchase. Hosting, updates when the platforms underneath it change, and somebody to call are priced separately so you can compare year two on equal terms.
- ### Data migration beyond what the audit scoped

   Migrating badly is how a new system inherits the old one’s problems, so cleaning and mapping is real work with a real price. What is being moved is established in the audit; anything found afterwards is re-quoted rather than absorbed without comment.
- ### A guarantee of savings or of pipeline

   No conversion uplift, no hours saved, no figure we cannot source. You decide what you are measuring before anything is built, and the numbers stay yours.

What you can check

## What can you check before you believe any of this?

The nearest thing you can inspect today is how we run our own business. One of our knowledge bases holds 4,535 notes and 5.9 million words, and the customer and sales records that feed it are drawn on Inside AGMM, department by department. That shows we build systems that stay accurate in daily use. It proves nothing about a CRM in your business and we will not stretch it to. What we cannot show you yet is a named CRM client with a published before and after.

[See the systems that run our own business](https://agmm.co.uk/evidence/)

People and machines

## Where does a person stay in the loop?

A CRM should record decisions rather than take them. Where AI is used it drafts, classifies or extracts, and a person still approves anything that leaves the building or changes a commercial position. Automatic scoring and prediction are easy to add and easy to regret, because a number with no working behind it gets treated as fact by whoever inherits it. If a model suggests something, the system shows why, and the person deciding can disagree without filing a support ticket. That constraint costs a little build time and it is the difference between a system your team uses and one they route around.

- **Reply approved** *Enquiries* AI drafts, classifies or extracts. A person approves anything that leaves the building or changes a commercial position.
- **Won or lost, with a reason** *Pipeline* The decision is a person’s. The record keeps the reason.

Proof it worked

## How do we know it worked?

Agree the measures before the build, not after it. The ones that work here are usually how long it takes to answer a question about a customer, how many enquiries are sitting with no next action, and how much hand reconciliation happens before a report is trusted. All three are countable today, and all three are countable in the same way afterwards. We do not measure adoption by logins, because a team can log in daily and still keep the spreadsheet. The test is whether the private spreadsheet disappears, and that is the one to watch in month three rather than week one.

*A worksheet we fill in with you, from your own records*

1. How long it takes to answer a question about a customer
2. Enquiries sitting with no next action
3. Hand reconciliation before a report is trusted

Same definition both times. If the number has not moved, that is a finding, not a failure.

Known risks

## What tends to go wrong on this kind of build?

Three, named before you commit. Data migration is the first and the biggest: what looks like a tidy export is usually five years of fields used for three different purposes, and that reconciliation is real work. The second is scope creep dressed as sensible additions, which is why a build runs in stages with something usable at the end of each. The third is the reporting trap, where a system grows dashboards faster than it grows accurate data, and everyone ends up looking at charts drawn from records nobody maintains.

Fit, stated plainly

## What do we need from you?

- An export of what you hold now, however untidy, so the migration is scoped against reality rather than a description of it.
- The people who live in the current system, including whoever keeps the private spreadsheet.
- The questions you ask of your numbers, so reporting is built around those rather than around everything measurable.
- Someone who can decide what the pipeline stages are, and make that decision stick.

## Who this is not for

- A business whose objection to its current CRM is the subscription fee, because building is an expensive way to avoid one.
- A sales process nobody has written down, where a new system would set concrete around something still moving.
- A team looking for a cheaper HubSpot rather than a different shape of system.
- A brief already specified in full and needing implementation hands only.

Questions

## What do people ask before they start?

**Why build rather than buy?**: Buy when a standard tool fits. Build when the mismatch is costing you: when the team works around the software, when the data cannot be trusted, or when the one capability you need is the one thing the product does not do. The first conversation is about which of those you are in.

**Can we move our existing data across?**: Usually yes, and it is worth doing properly. Migrating badly is how a new system inherits the old one's problems, so cleaning and mapping the data is part of the work rather than an afterthought.

**What happens if we grow?**: It is built to be changed. A custom system that cannot be extended is just a bespoke version of the problem you started with.

**How much does a custom CRM cost?**: It depends on how many kinds of user it has to serve, how many other systems it has to stay in step with, and the state of the data being moved into it. Data migration is routinely the largest line and it is almost never in the original estimate. Our guide on what custom software costs explains what moves the figure, so you can compare two quotes that look nothing alike.

**How long before the team is using it?**: We build in stages so a first useful part is in real use early rather than everything landing at the end. That also means you find out whether the shape is right while it is still cheap to change.

**What happens after it goes live?**: Software is an asset with running costs, not a purchase. Hosting, updates when the platforms underneath it change, and somebody to call. Ask us what year two looks like before you compare our figure with anyone else's, because a quote covering only the build is incomplete rather than cheap.

**Do you migrate data out of Salesforce, HubSpot or a spreadsheet?**: Yes, and it is treated as its own piece of work rather than a footnote. What takes the time is deciding what is worth keeping, reconciling duplicates, and mapping fields that were used for three different things over five years. We would rather quote that up front than discover it halfway through.

**Who owns the system, the code and the data?**: You do. It runs on your accounts, your hosting and your data, and it is documented so it does not depend on us being available. If you take it to another developer, everything they need is already written down.

**When should we not build a custom CRM?**: When a standard tool already fits and the complaint is really about price. When the sales process changes every few months, because a custom system freezes decisions you have not finished making. When nobody internally owns the data, since software makes a defined process faster but will not define it for you. We would rather say one of those on the fit call than after you have spent the money.

Free to read first

## Would you rather work it out yourself?

Everything below is free, with no account and no paywall. If it solves the problem without us, that is a good outcome.

- [Signs you have outgrown your CRM](https://agmm.co.uk/learn/outgrown-your-crm/)
- [Why your AI needs a brain](https://agmm.co.uk/learn/why-your-ai-needs-a-brain/)
- [Shared tools and context for teams](https://agmm.co.uk/learn/knowledge-base-vs-fine-tuning-vs-prompting/)
- [What custom software costs, and what moves the number](https://agmm.co.uk/learn/what-custom-software-costs/)
- [Keeping customer data safe when you automate](https://agmm.co.uk/learn/keeping-customer-data-safe-when-you-automate/)
- [Free course: give your AI a brain](https://agmm.co.uk/learn/courses/give-your-ai-a-brain/)

## What else is usually needed alongside this?

- [**Business process automation** Finance, People and Operations: the retyping, chasing and checking between your systems.](https://agmm.co.uk/services/business-process-automation/)
- [**Scheduling and booking software** Operations and Customer service: bookings, availability and customer updates in one place.](https://agmm.co.uk/services/scheduling-and-booking-software/)

[Everything we do](https://agmm.co.uk/services/)

## Tell us where the business is getting stuck.

A short call to map the constraint and give you a clear next step. That might be software, automation, training or no build at all.

[Book the free fit call](https://agmm.co.uk/ai-constraint-audit/#book)

1. Before day 1**Fit call** Free, 30 minutes
2. Day 01**Map the business** Constraints and weak points
3. Day 07**Price the opportunities** What could save or earn
4. Day 14**Rank the investment** Four outputs you keep

**Then the proposal.** The fuller audit is always part of it.

*Three 60-minute working sessions with the founders who would scope any build. Between them, we map, measure and price.*

---

Source: https://agmm.co.uk/services/custom-crm-development/ · AGMM, a UK AI consultancy · company facts: https://agmm.co.uk/about/agmm.json · every page: https://agmm.co.uk/llms.txt
