Service Cloud · implementation and rebuild
Service Cloud implementation,
priced per channel.
A Service Cloud implementation starts at €9,425 for 145 hours: case model, queues, email to case and reporting for up to 15 agents, with thirty days of support, up to eight hours, and a sixty day warranty. The estimator below shows eight weeks for that base, and each addition as its own line. The implementation guide sets out the seven phases.
Elapsed
-
Estimate
-
Scale
-
Engagement
-
What makes up the price
Phases
Every line above is a published price, so the total is one figure, not a range. It is an estimate rather than a quote until the written scope is signed; once that is signed, its price cannot move upward. Everything is built inside your own org and documented as the work happens, so nothing here depends on any one person staying available.
What actually takes the time
The case object is easy. Routing is not.
A case record takes an afternoon. Deciding who a case goes to, what happens when they do not answer, and how that changes at two in the morning takes weeks, because it is a business decision wearing a technical costume.
The case model
Record types, statuses, reasons, queues, and the difference between a case that is closed and a customer whose problem is solved. Agreed in writing before anything is built.
Channels
Email-to-case is an afternoon. Telephony, chat, WhatsApp and a customer portal are each a separate integration with a separate vendor, separate credentials and a separate go-live.
Routing and SLAs
Omni-Channel, skills, capacity, entitlements, milestones and business hours. Straightforward to configure and slow to agree, which is the reverse of what most plans assume.
Knowledge
Article types, approval flow, and the uncomfortable part: someone has to write the articles. This is also the prerequisite for any agent answering customers later, which is worth knowing now.
Fit
Worth reading before you book.
This is for you if
- From 5 to 200 support agents with a real SLA to meet
- You want routing rules written down rather than living in one person's head
- Agentforce is on the roadmap and you would rather build the knowledge base once
- Someone in-house will own the queues after handover
- You want routing documented as it is built, not reconstructed later
This is not for you if
- You want a contact centre platform rather than Service Cloud
- The requirement is to replicate Zendesk field for field
- Nobody will commit to writing knowledge articles
- The go-live date is fixed before the scope is
What this looked like on a real build: an insurance agency that moved client service from shared inboxes to Service Cloud, and the same build priced stage by stage for an agency.
The base package
What €9,425 already contains.
Service Cloud is priced per channel because that is how it consumes time. The base covers the part every rollout needs regardless of how customers reach you.
A case model that survives contact with reality
Record types, statuses, reasons and the distinction between a case that is closed and a customer whose problem is solved. Written down before anything is configured, because these definitions decide what every report afterwards is capable of saying.
Queues, assignment and escalation
Who a case goes to, what happens when nobody picks it up, and how that changes outside business hours. Configured in days, agreed in weeks, which is the reverse of what most plans assume.
Email-to-case
The channel every rollout starts with. Threading, auto-response, attachments and the rules that stop an out-of-office reply reopening a solved case at three in the morning. When one email still becomes three cases, five causes and two queries that tell them apart.
Reporting managers actually open
Backlog by age, first response and resolution against target, volume by reason, and agent workload. Built from the questions asked in your operations meeting rather than from a template.
Training, documentation, thirty days after
Sessions for administrators and for agents, both recorded. Then thirty days of questions answered and small adjustments made, up to eight hours in total, while the queues meet real volume, because the routing that looked right in a workshop rarely survives the first busy Monday unchanged. For sixty days after acceptance, anything built that does not meet the signed scope is fixed at no charge. How the handover support works.
Sequencing
Which channel to launch first, and which to launch second.
The most common Service Cloud mistake is launching every channel on the same morning. When something breaks, and something always breaks, nobody can tell which channel caused it.
Email first, always
It is asynchronous, so a routing mistake costs minutes rather than a dropped call. It carries the highest volume in most B2B operations, which means the case model gets stress-tested properly. And it is the cheapest channel to unpick if the model turns out wrong.
Then chat, if you have the staffing
Chat exposes capacity honestly. An agent handling three concurrent chats is not handling three cases well, and the routing has to know that. Launch it once you have a month of email data telling you what your real concurrency looks like.
Telephony when the vendor is ready, not when you are
This is the channel with an external dependency you do not control. The integration is straightforward; getting the telephony vendor's API credentials and a test window is not. Start that conversation in discovery, not in the channel phase.
The portal last, and only with knowledge behind it
A self-service portal without a populated knowledge base is a contact form with extra steps, and customers learn that within a week. Deflection is a promise the knowledge base keeps, not the portal. Launch them together or launch neither.
Knowledge
The part everyone postpones and later pays for twice.
Knowledge is boring, unglamorous, and returns more than anything else in a service implementation. It is also the prerequisite for any customer-facing agent you deploy afterwards, which makes postponing it an expensive decision rather than a neutral one.
What I build
Article types, the approval workflow, templates, categorisation and search tuning. Data categories mapped to how customers describe problems, not to how your product is organised internally. Those two are rarely the same thing.
What you do
Write the articles. There is no way around this and no consultant should pretend otherwise. The realistic starting point is your twenty most repeated questions, which your agents can list from memory in one meeting.
Why it matters more in 2026
A customer-facing agent answers from whatever it is grounded in, and knowledge articles are the source you control best. An agent with no knowledge base does not stay silent, it improvises, and improvisation in front of a customer is the failure mode that ends agent programmes. The knowledge base is the guardrail.
Questions
Asked often enough to answer here.
How much does a Service Cloud implementation cost?
A Service Cloud implementation starts at €9,425 for 145 hours, fixed against a written scope. The base covers the case model, queues, email to case, reports and dashboards for up to 15 agents, training, thirty days of support and a sixty day warranty. Each further channel, SLA tier, knowledge base or migration is its own line in the estimator above.
Is moving from Zendesk to Service Cloud worth it?
Only if your service and revenue data need to be in the same place. Zendesk is a better help desk in isolation. Service Cloud wins when a support conversation has to inform a renewal, an escalation has to reach an account owner, or an agent needs to reason across both. If none of that applies, staying put is the cheaper answer and I will say so.
Will everything from an existing help desk map to Service Cloud?
No, and planning for that is the difference between a smooth migration and a bad quarter. Some custom fields have no equivalent, some automations were workarounds for a limitation that does not exist here. Deciding what to drop is a business call, best made before migration rather than during.
Is Field Service included in a Service Cloud implementation?
No. It is a separate product with its own licences, its own mobile application and its own scheduling engine, and it is not work I take on. People who build Field Service every month know it better, so no estimate on this page includes it.
How many support agents justify Service Cloud?
Around five. Below that the routing rules cost more to maintain than they save, and a shared inbox with discipline outperforms a configured queue without it. Above fifteen, the absence of a proper case model starts showing up in your resolution times.
Can channels be added after go-live?
Yes, and I would rather you did. Each channel added later is priced at the same figure shown in the estimator, and a channel added onto a working case model takes less time than one launched alongside four others.
Before you scope it
The order the phases go in decides most of the number.
Seven phases of a Service Cloud build, what each one has to produce before the next starts, and the two that are routinely built too early and then paid for twice. And what the Salesforce licences themselves cost, which is a separate bill from this one and usually the larger of the two over three years.
Bring the estimate to the call.
Fifteen minutes. Read out what the estimator gave you and I will tell you where it is optimistic, where it is generous, and which channel I would launch second rather than first.
Serhii Skrypnyk · Senior Salesforce Administrator · 7 Salesforce certifications · on the platform since 2018. Reviewed 10 November 2026.