One MCP. Every GTM Signal.

Each replay uses a recreated ChatGPT interface so the placement can be read in its original conversational context. It is not a live or embedded chat.

1. ObservedJul 17, 2026Conversation1 turnsPlacementAfter turn 01 turns · After turn 0Replay chat
   Observed conversationJul 17, 2026 · interface recreation

   ChatGPTArchived view

   Running a mid-sized digital marketing agency. How do you structure retainers vs project work? What's the best way to scale without quality dropping?

   The agencies that scale most predictably tend to make one structural shift: projects generate cash, retainers generate enterprise value.

   Projects are great for acquiring clients and filling pipeline. Retainers create recurring revenue, improve forecasting, and let you invest in systems and people.

   Here's a framework that works well for a mid-sized digital marketing agency (roughly 10–75 people).

   1. Separate retainers from project work

   Don't let retainers become "unlimited marketing."

   Instead, sell ongoing outcomes with clearly defined capacity.

   Retainers	Projects
   SEO management	Website redesign
   Paid ads management	Brand identity
   Email marketing	Marketing automation setup
   Content production	Video campaign
   CRO optimization	Landing page builds
   Analytics/reporting	CRM migration

   Projects should have:

   • Fixed scope
   • Timeline
   • Milestones
   • Change-order process

   Retainers should define:

   • Monthly deliverables
   • Response SLAs
   • Meeting cadence
   • Strategic priorities
   • What's explicitly out of scope

   One mistake many agencies make is letting every client customize everything. Standardized service packages make staffing and delivery much easier.

   ---

   2. Productize your retainers

   Instead of selling "whatever you need," create 3 tiers.

   Example:

   Growth

   • Monthly strategy
   • 2 campaigns
   • Reporting
   • Slack/email support

   Scale

   • Everything above
   • Creative production
   • CRO
   • Weekly optimization

   Performance

   • Cross-channel management
   • Fractional marketing leadership
   • Forecasting
   • Executive reporting
   • Quarterly planning

   Clients can still customize around the edges, but 80–90% of delivery stays standardized.

   ---

   3. Build teams around pods, not departments

   Instead of everyone reporting through functional silos:

   Client A

   • Account Manager
   • Paid Media
   • SEO
   • Designer
   • Copywriter

   Client B

   • Same structure

   Each pod owns outcomes.

   Benefits:

   • Less internal communication overhead
   • Better accountability
   • Faster decisions
   • Stronger client relationships

   A pod leader becomes responsible for:

   • Client satisfaction
   • Profitability
   • Delivery quality

   ---

   4. Standardize everything repeatable

   If someone performs a task three times, document it.

   Examples:

   • Client onboarding
   • Campaign launch
   • Monthly reporting
   • QA checklists
   • Creative reviews
   • Ad account audits
   • SEO audits

   Good agencies don't rely on talented people remembering processes—they build systems that make quality repeatable.

   ---

   5. Scale quality through review layers

   Instead of founders reviewing everything:

   Junior → Senior → Lead → Client

   Examples:

   Copy

   • Writer
   • Editor
   • Account lead

   Ads

   • Specialist
   • Paid media lead

   Design

   • Designer
   • Creative director

   This prevents the founder from becoming the bottleneck.

   ---

   6. Measure utilization—but don't obsess over it

   Track:

   • Billable utilization
   • Gross margin by client
   • Gross margin by service line
   • Client acquisition cost
   • Lifetime value
   • Churn
   • Revenue per employee

   Watch especially for clients consuming disproportionate time relative to their fee.

   ---

   7. Create a healthy mix of work

   A balanced agency might look like:

   • 70–80% recurring retainers
   • 20–30% project work

   Projects:

   • Keep cash flow healthy
   • Create upsell opportunities
   • Feed the retainer pipeline

   Every project should end with a roadmap conversation that naturally leads into ongoing support where it genuinely adds value.

   ---

   8. Protect account managers

   One account manager shouldn't become the agency's "human router."

   Instead:

   Account Manager

   • Strategy
   • Client communication
   • Prioritization
   • Growth

   Operations

   • Resource allocation

   Project Manager

   • Delivery timelines

   Specialists

   • Execution

   Otherwise account managers spend their day chasing updates instead of building client relationships.

   ---

   9. Build capacity before you need it

   Many agencies hire only after people are overwhelmed.

   Instead, define hiring triggers, for example:

   • Paid media team at 85% utilization → hire
   • Designers booked 6 weeks out → hire
   • Account managers handling more than 8–12 active accounts (depending on complexity) → hire

   Hiring slightly ahead of demand is usually less costly than losing clients due to overloaded teams.

   ---

   10. Give every client the same experience

   The highest-performing agencies often have an operating system that every client experiences consistently:

   Week 1

   • Kickoff
   • Goal setting
   • Access collection

   Week 2

   • Audit
   • Strategy

   Week 3

   • Implementation

   Monthly

   • KPI review
   • Strategy session
   • Next-month priorities

   Quarterly

   • Business review
   • Growth roadmap

   Consistency reduces errors and makes training much easier.

   A practical target operating model

   If your goal is sustainable growth without quality slipping, aim for something like:

   • Revenue mix: 75% retainers, 25% projects
   • Client model: standardized service tiers with defined scope
   • Delivery: cross-functional client pods
   • Quality control: documented SOPs plus peer-review checkpoints
   • Leadership: founders focused on strategy, hiring, and key client relationships—not day-to-day approvals
   • Growth engine: projects used intentionally to convert qualified clients into long-term retainers

   The biggest inflection point for most agencies comes when they stop treating every engagement as a custom service and start running delivery as a repeatable system. That shift improves margins, makes hiring easier, and allows the business to grow without quality depending on a handful of senior people.

   

   ZoomInfo Technologies IncAd

   One MCP. Every GTM Signal.One MCP for AI agents and GTM workflows. Pay for what you use.
2. ObservedJul 15, 2026Conversation3 turnsPlacementAfter turn 03 turns · After turn 0Replay chat
   Observed conversationJul 15, 2026 · interface recreation

   ChatGPTArchived view

   Internal requests arrive through email and someone manually assigns every one. How can we automate intake and routing?

   You can automate this with an email-driven intake workflow that extracts information from incoming requests, classifies them, and routes them based on rules or AI.

   A typical architecture looks like this:

   1. Monitor a shared inbox

      • Watch an email address such as requests@company.com.
      • Trigger a workflow whenever a new email arrives.

   2. Extract request details

      • Capture:
        • Sender
        • Subject
        • Email body
        • Attachments
      • Use AI to identify:
        • Request type
        • Priority
        • Department
        • Due date (if mentioned)
        • Customer or project name
        • Required skills

   3. Validate required information

      • Check whether key fields are missing.
      • If information is incomplete, automatically reply with a request for the missing details instead of creating an assignment.

   4. Classify the request

      • Examples:
        • IT support
        • HR
        • Finance
        • Legal
        • Sales operations
        • Customer success
      • AI can classify free-form emails even if they don't follow a template.

   5. Route automatically Apply routing rules such as:

      • Request type → Team
      • Customer tier → Priority queue
      • Geographic region → Regional team
      • Product line → Specialist group
      • Language → Native-language queue
      • Workload → Least busy assignee
      • Severity → Escalation queue

   6. Create a work item Automatically create a:

      • Help desk ticket
      • Jira issue
      • Asana task
      • Monday.com item
      • ServiceNow incident
      • CRM case

   7. Notify stakeholders

      • Confirmation email to the requester
      • Slack or Teams notification to the assigned team
      • Manager notification for high-priority requests

   8. Track progress

      • Record:
        • Intake time
        • Assignment time
        • SLA deadline
        • Resolution time
        • Routing accuracy

   Example workflow

   Email arrives
           ↓
   AI extracts fields
           ↓
   Categorize request
           ↓
   Determine priority
           ↓
   Check required information
           ↓
   Assign to best queue
           ↓
   Create ticket
           ↓
   Notify assignee
           ↓
   Send confirmation to requester
   

   AI enhancements

   Instead of relying solely on keyword rules, AI can:

   • Summarize lengthy emails.
   • Detect urgency from the language used.
   • Recommend the best team.
   • Suggest an assignee based on historical patterns.
   • Extract action items and deadlines.
   • Detect duplicate requests.
   • Flag potential SLA breaches.
   • Recommend responses for common requests.

   Expected benefits

   Organizations that replace manual triage with automated intake commonly see:

   • Faster response times because requests are routed immediately.
   • More consistent assignment decisions.
   • Reduced manual administrative work.
   • Improved SLA compliance.
   • Better reporting due to structured metadata captured at intake.

   This approach can be implemented with workflow platforms such as Microsoft Power Automate, Zapier, Make, or n8n, integrated with AI services and your existing ticketing or project management system. If your organization already uses Microsoft 365 or Google Workspace, you can often build the workflow largely within that ecosystem while connecting to your preferred work management platform.

   

   ZoomInfo Technologies IncAd

   One MCP. Every GTM Signal.One MCP for AI agents and GTM workflows. Pay for what you use.

   The request type, department, urgency, and dollar amount determine who should review it.

   That makes the routing logic much more straightforward because those fields can drive assignment decisions.

   A workflow could look like this:

   Incoming email
         ↓
   AI extracts:
   • Request type
   • Department
   • Urgency
   • Dollar amount
         ↓
   Validate extracted values
         ↓
   Apply routing rules
         ↓
   Assign reviewer
         ↓
   Create ticket/work item
         ↓
   Notify reviewer and requester
   

   Example routing rules might be:

   Condition	Reviewer
   IT + Low urgency	IT Service Desk
   IT + High urgency	IT Operations Manager
   HR + Any urgency	HR Business Partner
   Finance + Amount < $10,000	Finance Analyst
   Finance + $10,000–$50,000	Finance Manager
   Finance + > $50,000	Finance Director
   Legal requests	Legal Counsel
   High urgency + > $100,000	Executive approval queue

   The routing engine would evaluate all applicable conditions before assigning the request. For example:

   • Request Type: Purchase Request
   • Department: Marketing
   • Urgency: High
   • Dollar Amount: $85,000

   Result:

   • Assign to Procurement Manager
   • Add Marketing Director as an approver
   • Escalate because the amount exceeds the approval threshold
   • Set a 24-hour SLA

   To keep the workflow maintainable, it's better to store routing rules in a configurable table rather than hard-coding them. For example:

   Request Type	Department	Urgency	Amount Range	Reviewer
   Purchase	Any	Low	$0–10K	Buyer
   Purchase	Any	Medium	$10K–50K	Procurement Manager
   Purchase	Any	High	$50K+	Procurement Director
   Contract	Any	Any	Any	Legal Counsel
   Budget	Finance	Any	$100K+	Finance Director

   With this approach, business users can update thresholds, reviewers, or approval paths without changing the automation itself.

   If some emails don't explicitly state the request type, department, urgency, or dollar amount, an AI extraction step can infer those values from the email content before the routing rules are applied, while flagging low-confidence extractions for manual review.

   

   JJitterbit, Inc.Ad

   #1 iPaaS for Implementation.Go live in under 2 months with Jitterbit iPaaS.

   

   JJitterbit, Inc.Ad

   Unified Automation Platform for Enterprise.A unified platform for automation, integration, and application development solutions.

   Can no-code workflow tools handle conditional approvals and reminders without engineering support?

   Yes. Most modern no-code workflow platforms can handle conditional approvals, reminders, escalations, and routing without requiring software engineering, provided your approval logic isn't unusually complex.

   Typical capabilities include:

   Capability	Common no-code support
   Multi-step approvals	✓
   Conditional routing ("if/then")	✓
   Approval thresholds (e.g. by dollar amount)	✓
   Parallel approvals	✓
   Sequential approvals	✓
   Automatic reminders	✓
   Escalations after SLA deadlines	✓
   Email and Teams/Slack notifications	✓
   Integration with ticketing and business systems	✓
   Approval history and audit trail	✓

   For your scenario, a workflow might look like:

   1. An email arrives in the shared inbox.
   2. AI extracts:
      • Request type
      • Department
      • Urgency
      • Dollar amount
   3. The workflow evaluates conditions:
      • If request type = Finance and amount < $10,000 → Finance Manager
      • If amount ≥ $10,000 → Finance Director
      • If urgency = High → mark as high priority and notify immediately
      • If request type = Legal → Legal team
   4. The assigned reviewer receives an approval request.
   5. If there's no response within two business days, send a reminder.
   6. After another business day, escalate to the reviewer's manager.
   7. Once approved, create or update the downstream work item and notify the requester.

   Most no-code platforms also let business users maintain approval rules in a table or list, so changing a dollar threshold or reviewer typically involves updating configuration rather than editing the workflow itself.

   There are, however, situations where engineering support may become useful, such as:

   • Approval logic based on dozens of interdependent rules.
   • Integrations with legacy or proprietary systems that lack standard APIs.
   • Highly customized user interfaces.
   • Advanced compliance, security, or governance requirements.
   • Large-scale workflows with thousands of requests per day that need optimization.

   For many organizations, especially those using Microsoft 365 or Google Workspace, this type of intake and approval process can be built and maintained by operations or business analysts using platforms like Microsoft Power Automate, Zapier, Make, or n8n, with only occasional IT involvement for initial setup, security reviews, or connector configuration.

   

   OOneTrustAd

   OneTrust recognized by Gartner® as a VisionaryOneTrust has been recognized in the inaugural Gartner® Magic Quadrant™ for AI Governance Platforms.