A CRM rollout for 30 people fails for one of two reasons: nobody configured it, or somebody configured everything. The second failure is more common and more expensive, because the portal looks impressive in a demo and gets abandoned within a quarter.
The teams that stick with HubSpot start narrow. One pipeline. A handful of properties people actually fill in. Email logging that works without anyone thinking about it. Dashboards that answer the three questions leadership asks in a Monday meeting. Automation comes later, once there is real data to automate against.
This is the sequence we use on HubSpot CRM setup and automation projects, written out in the order the work should happen.
1. Decide seats and roles before you buy anything
HubSpot bills per paid seat, and the number you need is rarely the number of employees. Map every person to a role first: who runs the portal, who needs to own deals, who only reads records and logs email.
In a 30-person company that usually breaks into an administrator, one or two directors, a group of account managers who own pipeline, and support or admin staff who work records but never own revenue. Only some of those need paid seats.
Buying at the start of a billing month also matters more than people expect. If you are trialling monthly, align the upgrade with the first of the month so the subscription period matches your finance cycle.
| Role | Typical count | Needs to own deals? | Permission approach |
|---|---|---|---|
| Administrator | 1 | Rarely | Super admin — portal settings, integrations, billing |
| Director | 2 | Sometimes | Full CRM view, all reports, no settings access |
| Key account manager | 5-8 | Yes | Own deals, edit all contacts and companies |
| Sales administrator | 2 | No | Edit records, create contracts, no deletion rights |
| Customer service agent | 1-3 | No | View everything, log activity, edit contact details |
2. Choose your visibility model early
HubSpot lets you restrict records to their owner, to a team, or leave everything visible to everyone. Changing this after go-live is disruptive, so decide it in week one.
For most companies under 50 people, full visibility wins. Account managers cover for each other, support needs the deal history, and directors want to look at anything without asking. Restricting records makes sense when you have competing sales teams or regulatory separation — not because it feels tidier.
Full visibility does not mean full permissions. Everyone can see a record while only the owner and admins can delete it, and only admins can touch settings, properties and integrations.
3. Connect email properly — logging and contacts are two jobs
This is where most rollouts stall. Connecting a Microsoft 365 or Google Workspace mailbox in HubSpot enables email logging and tracking going forward. It does not import the address book that already exists in that mailbox.
The distinction is invisible until the second week, when the admin's contacts are in the CRM (they came in during initial setup) and everyone else's are not. The fix is a separate integration, not a mailbox reconnection — we wrote the full diagnosis in why your team's Outlook contacts are not syncing to HubSpot.
Do the mailbox connection as a supervised group exercise, not an email instruction. Fifteen minutes with everyone in a room beats three weeks of chasing individuals.
| Setup step | Logs future email | Imports existing contacts | Who does it |
|---|---|---|---|
| Connect personal inbox (M365 / Gmail) | Yes | No | Each user |
| Outlook Contacts / Google Contacts sync app | No | Yes | Each user, after admin enables it |
| Connect a shared/team inbox | Yes, to the shared inbox | No | Admin |
| CSV contact import | No | Yes, one-off | Admin |
4. Build one pipeline, not five
A single pipeline that reflects how a deal actually moves is worth more than a set of theoretical pipelines per product line. Name the stages after events that either happened or did not — quote sent, site visit completed, contract signed — never after feelings like "warm" or "engaged".
Six or seven stages is the practical ceiling. Beyond that, people stop updating and your forecast becomes fiction.
Set the company currency before the first deal is created. If you trade in South African Rand, Australian Dollars or anything other than USD, changing it later leaves historic deal values converted at rates nobody remembers agreeing to.
5. Keep properties to what people will actually fill in
Every required property is a tax on data entry. Start with the fields you genuinely report on: value, close or renewal date, owner, status, associated company, and the one or two qualifiers specific to your business.
Use dropdowns instead of free text wherever you plan to report on the field. Free text cannot be grouped, filtered or charted, so a text field is a field you will rebuild later.
Verify automatic contact-to-company association against real records rather than trusting the setting. Domain-based association is reliable for corporate customers and unreliable for anyone using a personal email address.
6. Handle the internal-thread problem before someone complains about it
Teams assume HubSpot understands context. If two colleagues discuss a customer by name and the customer is not on the thread, HubSpot has no reliable way to attach that email to the customer's record — it associates on participants, not on subject matter.
The workable answer is a documented manual step: from the logged email, associate the activity with the relevant company record. It takes seconds and keeps internal discussion in the customer timeline where it is useful six months later.
Agree the convention during training rather than leaving each person to invent one. Some teams standardise on always associating to the company; others only for decisions and pricing. Either works — inconsistency does not.
7. Dashboards last, and only three questions deep
Build reporting after the pipeline and properties are settled, because reports built on fields that later change have to be rebuilt anyway.
Start with pipeline value by stage, contracts closing this quarter, and activity by owner. Add anything else only when a named person asks for it in a meeting twice.
Then simplify the navigation. Hiding the objects and tools a 30-person company will never use removes most of the intimidation that stops people logging in at all.
8. Train once, live, and record it
A written handover document is read by the person who commissioned the project and nobody else. A 45-minute live session where each role does their own daily task in the real portal is what changes behaviour.
Cover four things: find a customer, log or associate an email, move a deal stage, and where to look for your own numbers. Everything else can be learned later.
Record it. New starters in month three get the same onboarding without booking anyone's time.
What we deliberately leave out of version one
No automated email sequences, no lead scoring, no lifecycle stage automation, no custom objects. Every one of those is defensible in year two and destructive in month one, because they encode assumptions about a process you have not yet observed in the CRM.
The one exception is anything that reduces manual entry — association rules, required-field logic, deal defaults. Automation that saves keystrokes gets adopted. Automation that makes decisions gets distrusted.
At $30/hour, a full 30-seat configuration of this shape is typically a $600 to $900 engagement delivered inside a week. Complexity is what makes CRM projects expensive, and most of it is optional.