Deck: A step-by-step workflow for integration architects who need every decision tied to the call, ticket, or document that produced it.
What this article solves
A source-linked delivery record connects each CRM implementation decision to its origin: the discovery call, the migration assumption, the configuration change. This guide walks through how to build one from day one of a rollout, so no decision lives only in someone's memory or a private Slack channel.
Who this is for: Integration architects, technical leads, and delivery managers at agencies and CRM consultancies who own the implementation record across multiple systems and stakeholders.
It was 9:14 on a Tuesday morning when the client's operations lead sent the same question for the third time in six weeks. The message landed in #crm-rollout-general: "Can someone confirm whether the Salesforce account hierarchy is syncing from the ERP or being managed manually in CRM?"
The integration architect on the engagement, Priya, knew this had been decided in workshop three. She remembered the conversation. She remembered the whiteboard. What she could not find, forty minutes later, was any written record that would survive a steering committee question. The decision had been made in a Zoom call, referenced once in a Slack thread that was now buried under two hundred other messages, and never written into the integration spec.
The client was not being difficult. The answer simply did not exist anywhere a person could point to.
Why CRM rollouts lose their decisions
CRM implementations generate more decision volume than most teams expect. Integration mappings change after workshops. Data migration assumptions get made verbally and never confirmed in writing. Custom development rationale ends up in a private channel between two people who are no longer on the engagement.
By the time the project reaches UAT or handoff, the delivery team is rebuilding context from memory. The client asks a question that was answered in week two. Someone has to go find the Fathom summary from that call, the Jira ticket that referenced it, and the Confluence page that was supposed to document the outcome but was never updated after the scope changed.
This is not a discipline problem. It is a structural one. The tools that run the engagement (Jira, Slack, Fathom, Confluence, Google Drive) each hold a fragment of the record. Nothing connects them.
Step 1: Define the record before kickoff, not after
The delivery record is not a document you write at the end. It is a structure you set up on day one so that decisions have a place to land as they happen.
Before the first client call, establish four categories that will hold every decision made during the engagement:
- Integration decisions: which systems sync, which fields map where, who owns conflicts
- Data migration assumptions: what gets migrated, what gets excluded, what the client confirmed in writing
- Configuration choices: any setting that deviates from default, with the reason recorded
- Custom development rationale: what was built, why it was built that way, which requirement drove it
These categories become the skeleton. Every ticket, call summary, and document produced during the rollout should map to one of them.
Step 2: Link every decision to its source artifact
A decision without a source is a rumor. When the integration mapping changes after workshop three, the record needs to show: what changed, when, and what produced the change. That means linking the decision to the artifact that contains the evidence.
In practice, this looks like:
- Fathom call summary: the call where the mapping was discussed. Link it directly to the decision entry.
- Jira ticket or Linear issue: the work item that tracked the change. Attach it to the same entry.
- Confluence page or Google Drive doc: the spec or design doc that reflects the updated state.
- Slack thread permalink: if the final confirmation happened in chat, capture the link before it scrolls away.
The goal is not to duplicate content across four tools. The goal is to make the decision findable by following one link back to its source.
Step 3: Capture configuration and custom work at the moment of change
The hardest decisions to recover are the ones that happened quietly. A field mapping adjusted during development. A workflow rule changed to accommodate a client edge case. A custom object added because the standard data model did not fit the client's account structure.
These changes rarely generate a formal document. They generate a PR, a comment in a ticket, or a message to a colleague.
The discipline here is small but specific: when a configuration or custom development decision is made, write one sentence describing the rationale and attach it to the work item before closing it. Not a paragraph. One sentence with a link to whatever drove the decision.
That sentence, repeated across every ticket, becomes the configuration decision log by the time the project closes.
Step 4: Keep the record current through steering meetings
Steering meetings are where delivery records die. The team prepares a status update, the client asks a question, the answer gets given verbally, and the record never reflects what was said or decided.
Instead of rebuilding context for every steering meeting, ScopeDocs makes the implementation record traceable to the call, ticket, document, or change behind it, so the answer the client heard in the meeting is the same answer the next consultant finds when they join the engagement.
Before each steering meeting, spend ten minutes reviewing the four decision categories and flagging anything that changed since the last session. After the meeting, add any new decisions or scope changes to the record with the call summary linked as the source. This takes less time than answering the same question again three weeks later.
In practice
A mid-size CRM consultancy is six weeks into a Salesforce rollout for a manufacturing client. The integration architect opens the delivery record on Monday morning before the weekly sync. She checks the integration decisions category and sees that the account hierarchy question, the one that surfaced in #crm-rollout-general, is listed but has no source link. She pulls the Fathom summary from workshop three, finds the timestamp where the decision was made, and attaches the link. She adds one sentence: "Account hierarchy managed in ERP; Salesforce receives read-only sync via middleware. Confirmed by client ops lead in Workshop 3 (Fathom, 42:17)." The next time the question comes in, the answer takes thirty seconds to find. The steering committee gets a cited response, not a reconstructed one.
Delivery record setup checklist
- Define the four decision categories before the first client call
- Create a home for the record that all delivery team members can access and edit
- Link every integration mapping decision to the call or document where it was confirmed
- Attach Jira or Linear ticket numbers to every configuration change at close
- Capture custom development rationale in one sentence per ticket before marking done
- Add a ten-minute record review to the pre-steering-meeting agenda
- After each steering meeting, log any new decisions with the Fathom summary linked as source
Related ScopeDocs resources
- What is source-linked documentation?
- ScopeDocs product overview
- ERP Implementation Handoff Checklist: Requirements, Decisions, Configuration, and Cutover
- What Code Reviews Actually Capture (And Why CRM and ERP Pods Keep Losing It)
- All posts tagged: knowledge-management
- Integrations: what to connect first
The answer to the client's question was always there. It lived in a call summary, a ticket, a thread. The work is building a structure that connects those artifacts before someone has to go looking. Capture the decision once, link it to its source, and keep it current. That is what a delivery record is for, and it is what ScopeDocs is built to support.