Support is the department where a customer decides whether the sale was a good idea. Service teams are rarely short of effort. What they are short of is one place where a customer’s history, entitlements and open problems sit together.
Support still running out of a shared inbox
Customers repeating themselves at every handoff
No view of which accounts are expensive to serve
Adding service to an org that already runs sales
You cannot staff your way out of a context problem.
Service Cloud implementation, case routing and SLA design, knowledge, entitlements and customer self-service for teams outgrowing a shared inbox.
When support lives in an inbox and a ticketing tool nobody else can see, the cost shows up everywhere else. Sales walks into a renewal blind, finance cannot tell which accounts are expensive to serve, and the customer repeats their story to every person they reach.
We put the case, the history and the entitlement on the same record as the relationship. Agents stop hunting for context, managers can see where the time actually goes, and the rest of the business can finally see what support already knows.
Where it breaks
Email, phone, chat and the form on your website all land somewhere different, so nobody can say how much is actually coming in.
The agent who picks up an escalation cannot see what was promised during the sale or what was fixed last month, so they start from zero.
Who is under contract, what they are owed and when it expires live in somebody’s spreadsheet, so support gets given away by accident.
The answer exists, in the head of the one person who has seen it before. When they are out, the case waits.
Time is spent rather than measured, so the accounts that consume the most support look identical to the ones that do not.
A portal that cannot see the customer’s own cases or orders sends them straight back to the phone.
What we build
Nine areas of a service operation we have rebuilt often enough to know which ones break first.
Email-to-Case across multiple routing addresses, web forms, phone entry, portal submission and cases opened by monitoring systems, with known-sender matching and deduplication of the notification storm that usually exists first.
A deliberately small set of record types, with per-business-line difference carried by a product field driving dependent categories, so the fourth line onboards by configuration rather than by another build.
Functional queues named for the team that works them, category-to-queue mapping, assignment by department or inbound address, and portal routing driven by the choice the customer already made.
Team-to-team escalation that transfers ownership in one validated step, and leadership escalation that notifies a manager and flags the case without moving it, backed by scheduled re-checks.
Working and waiting intervals computed from status transitions and locked against editing, thresholds that color-code list views, and an indicator for the customer who has replied since the owner last looked.
Article types, a category model that separates internal from customer-visible, approval to publish restricted to a named role, and candidate articles surfaced on the case from its own subject line.
Console apps per team, Dynamic Forms that keep low-frequency fields collapsed, highlights panels and curated list views, built on one rule: the intake layout shows only what intake needs.
Public and internal comments that notify differently, a required public comment before a case waits, cause code and resolution required to close, and threading rules that reopen a recent case or start a new one past a cutoff.
Role-mapped permission set groups, sharing that tiers reps who see their own cases against supervisors who see all, and, where the industry demands it, platform encryption with a logging standard that keeps those values out of logs entirely.
Beyond phase one
Each one came out of delivery work rather than a product roadmap. In most cases the client did not know to ask for it.
Most orgs find out an SLA was missed by running a report on Monday. Milestones that escalate at sixty percent of the clock move the conversation to before the miss, which is the only point at which anyone can do something about it.
A phone agent, a web form and a portal article should not disagree. Attaching knowledge to case type rather than to a folder is what keeps them honest, and it is almost never in a phase one scope.
Customers say things that are not support requests and still predict churn. A lightweight record for the aside, the complaint, the offhand remark gives you a signal that no case volume chart contains.
Queues usually mirror the org chart because that is what existed on day one. Skill-based assignment reflects who resolves a thing without a second hop, which is a different list and a shorter one.
A case closed twice reads as two closures and one happy customer. Counting the reopen separately is a small change that usually rewrites what leadership believes about first-contact resolution.
Integrations
Service Cloud earns its place when the agent stops switching windows. These are the categories that most often need joining.
Calls, recordings and screen pops against the case rather than beside it.
Orders, shipments and invoices visible to the agent without a second login.
What the customer actually owns, and which version of it.
A portal customers can get into, and stay in, without a password reset call.
The second-line conversation happening where the case can see it.
Cost to serve, by account, as a number rather than a feeling.
Service Cloud · Sales Cloud · Experience Cloud
Sales Cloud · Service Cloud
Sales Cloud · Service Cloud · Conga
Email, the phone, chat, the form on your site, the rep who got texted directly. If the answer is more than two and they do not land in one queue, you already know what the project is.