What a CRM and automation build actually costs

CRM shopping almost always starts with the per-seat price, because it is the number vendors publish and the only one that can be compared on a page. It is also, in nearly every implementation we have seen, the smallest line in the project.

Where the money actually goes

Data preparation is the line that most often doubles a project, and it is the one nobody quotes because it is not the supplier's work to do. Somebody has to decide what to keep, reconcile the same person entered four different ways, and standardise fields that were never standard. That work exists whether or not it appears in the estimate.

Integrations are not a feature list

Connect it to our email, our accounting and our website form reads as one requirement and is three projects. Each connection has its own authentication, its own idea of what a contact is, its own rate limits and its own behaviour when something goes wrong at the far end.

The specific cost driver is not the happy path, which is usually straightforward, but the disagreements. When your accounting system and your CRM hold different addresses for the same customer, something has to decide which one wins, and that decision has to be made by a person before it can be encoded. Multiply by every field that exists in both systems.

The part that gets cut, and should not

Training and adoption is the line most often removed when a quote needs to come down, on the reasonable-sounding grounds that the software is intuitive and people will pick it up. What actually happens is that everybody develops their own way of using it within the first fortnight, those ways conflict, and the data becomes unreliable in ways that are very hard to unwind later.

The specific thing training buys is agreement — not on how to click things, but on what a stage means, when a record moves, and what counts as done. Two people with different definitions of qualified will produce a pipeline report that is quietly wrong, and the decisions made on it will be too.

The four-week revision

Every CRM implementation needs a revision after about a month of real use, and it is worth putting in the plan rather than treating it as a fault.

Whatever process was designed at the start was designed by people imagining how they work. Four weeks of actually working reveals the stage everyone skips, the field nobody fills in, and the case that does not fit the pipeline at all. Those are cheap to fix at week four and expensive to fix at month six, once there is data shaped around them.

A supplier who has built this into the plan is telling you they have done this before. One who treats it as scope creep has not.

Keeping the total down honestly

• Import less. Every record you do not migrate is preparation you do not pay for and clutter you never see.
• Start with one pipeline. Extra pipelines can be added in a fortnight once the first one is proven.
• Sequence the integrations. Connect the one that removes the most manual work first, then reassess — the second one often turns out to be unnecessary.
• Do not automate at the same time as implementing. Get the process recorded accurately first; automating a process nobody has validated encodes the mistakes.