Seven fixed phases from discovery to hypercare, run by Salesforce implementation specialists inside an AI-first engineering firm. A clean data model, documented automation and a CRM that LLMs and agents can actually use.
Salesforce rewards teams that decide what the system is for before they switch it on. Most implementation requests arrive in one of three shapes.
In each case the job is the same. Understand how you sell and serve, design the smallest org that supports it, then migrate data you can trust. Everything else is sequencing.
You see a document, a demo or a report at the end of every phase. Nothing stays "in progress" for more than a fortnight without something to look at.
Process maps, a target object model and a backlog ranked by value, wrapped in a scope statement with a price. You know the number before build starts.
Data model, security and sharing design, automation inventory and integration contracts. Reviewed with your admin before a single field is created.
Built in a sandbox and promoted through a release pipeline, never clicked straight into production. Flow first; Apex only where Flow cannot do the job.
When we write codeProfiling, cleansing rules, deduplication and reconciled trial loads before a rehearsed cutover. Legacy systems go read-only, not off.
Data migration serviceERP, telephony, marketing and data warehouse connected through documented APIs, with monitoring and a named owner. LLM services join through the same layer.
Integration patternsRole-based training, admin runbooks and 2–4 weeks of hypercare with daily triage. Then your admin or our managed services take over.
Managed servicesThe weeks below describe a Sales or Service Cloud implementation with one or two integrations. Multi-cloud programmes follow the same phases over 4–6 months.
Stakeholder interviews, process walkthroughs and profiling of the source data. Deliverables: current-state map, prioritised backlog, scope statement and risk register.
Object and field model, page layouts, security and sharing, automation inventory and integration contracts. Deliverable: the solution design document, signed off at gate one.
Configuration in a developer sandbox, promoted through a release pipeline, with a demo every week. Deliverables: configured org, test scripts and admin documentation.
Field mapping, cleansing rules, deduplication and trial loads into a full sandbox with reconciliation reports. Deliverable: a rehearsed cutover plan with timings and owners.
Your users run real scenarios against migrated data while defects are triaged daily. Deliverable: UAT exit report and the go/no-go decision at gate two.
Final load over a weekend or quiet period, access switched on, legacy systems set to read-only. Deliverable: live org and a cutover log showing every record count reconciled.
Daily triage, quick fixes, adoption nudges and dashboard tuning. Deliverable: handover pack and open-items list at gate three, then managed services or your admin takes over.
Three gates stop the project moving forward on assumptions. Each one is a short meeting with a document on the table and a yes or no at the end.
An org that has been live for a few years usually carries debt. The question is whether that debt is specific enough to fix, or structural enough to start again.
| Signal | Fix in place when… | Re-implement when… |
|---|---|---|
| Data model | Standard objects are used as intended and custom objects have owners | Custom objects duplicate standard ones and nobody can explain half the fields |
| Automation | Flows are documented and one tool handles each process | Workflow Rules, Process Builder, Flow and Apex all fire on the same object |
| Security | Permission sets and groups control access | Profiles were cloned per user and field-level security is unknown |
| Data quality | Duplicates are under control and required fields are populated | The real pipeline lives in a spreadsheet next to the CRM |
| Integrations | Documented, monitored and owned | Undocumented middleware and nightly CSV jobs nobody dares to stop |
| Cost of change | A new field or Flow ships in days | Every change breaks something else |
| Our usual recommendation | Health check, then a ranked fix backlog delivered in sprints | Health check, then a re-implementation with migration run in parallel |
The org health check scores technical debt, security, performance and adoption in two to three weeks and ends with this exact recommendation.
Implementation rarely travels alone. These are the pages our implementation clients read next, plus how we run projects in general. See all Salesforce services.
ETL steps, deduplication, validation and a cutover plan you rehearse before the real weekend.
ReadA scored audit of technical debt, security, performance and adoption, with a fix-or-rebuild verdict.
ReadAdmin, release management and monitoring after go-live, in tiers without surprises.
ReadOur workshop-first process, sprint cadence and the communication habits behind every engagement.
ReadFixed scope, retainers and team extension, and when each one makes sense for a Salesforce programme.
ReadHow to sequence AI work so that a CRM implementation becomes the foundation rather than a blocker.
ReadA single cloud with one or two integrations usually takes 8–16 weeks from discovery to go-live, plus 2–4 weeks of hypercare. Multi-cloud programmes run 4–6 months. The variables that move the range most are data quality, the number of integrations and how quickly your side can make decisions.
Fix in place when the data model is sound and the problems are specific. Re-implement when automation tools overlap, nobody can explain the fields, and the real pipeline lives in a spreadsheet. Our org health check gives you the evidence to decide in two to three weeks.
Yes, and we prefer it. Your admin sits in design reviews, sees every Flow before it is promoted and receives the runbooks. The aim is that your team can change the org after we leave, with managed services as an option rather than a dependency.
A product owner with decision rights, a data owner and sandbox access. Subject-matter experts need roughly four to six hours a week during discovery and UAT. We set this out in week one so nobody is surprised later.
Yes, as long as the data model gives them something to read. We design fields and picklists with lead scoring, case summarisation and agent actions in mind. The AI work is then sequenced using the Salesforce AI roadmap.
Discovery is a fixed fee. Build is fixed-scope wherever the design has been signed off, with a time-and-materials option for genuinely open scope. See our engagement models.
Tell us about your systems, your data and your constraints. We reply within one business day with a straight answer and a suggested next step.