anastasiacrm.com
CRM Orchestration Framework

How I Think About Campaign Orchestration

10 or 1,000 campaigns a month — no longer an issue. Explore a framework for deciding what gets sent, to whom, and in what order.

← Back to home 10-part framework · ~12 min read

Most CRM teams don't have a campaign problem. They have a prioritization problem.

  • Every stakeholder has a campaign that "must" go out
  • Every campaign has an audience that overlaps with three others
  • Every customer wants personalised & valuable offers at the right time

Orchestration is the layer that decides.

Explore the framework I use: 10 pieces that go from how you classify campaigns, to how you score and sequence them, to how you measure whether the whole system is healthy — not just whether one campaign performed.

01 · Foundation

Start with a single campaign taxonomy

Campaigns shouldn't be organized by who requested them. They should be organized by the type of impact they have on the customer and the business. Once every campaign has a type, that type carries a default priority — and service or transactional messages almost always outrank promotional ones.

Priority: Highest
Critical / Service
Security alerts, important notices, document requests, changes to terms.
Priority: High
Transactional
Order, payment, delivery, appointment, and status updates.
Priority: Context-dependent
Lifecycle
Welcome, onboarding, activation, usage nudges, retention, reactivation.
Priority: Medium
Revenue / Sales
Offers, upsell, cross-sell, product launches.
Priority: Low
Engagement / Brand
Content, education, inspiration, community.
Priority: Opportunistic
Research / Feedback
Surveys, NPS, interviews, reviews.

02 · Audience model

Think in customer states, not segments

"Female, 25–35" or "owners of product X" describes who someone is — it doesn't tell you what to send them right now. A state does. States capture where a customer currently sits in their relationship with the business, so orchestration becomes a question of "what does this person need right now," not just "which campaign are they eligible for."

New customer Activated High intent Recently purchased Waiting for delivery / service Using product actively Low engagement At risk Complaint / open issue Lapsed Loyal / high value Discount-sensitive Category explorer
Why this matters If a customer has an open complaint, "discover our new collection" is the wrong message — no matter how well it targets their category affinity. The negative context has to close before anything else goes out.

03 · Category logic

Eligibility and affinity, before anyone sees a category

With more than a handful of product categories, you can't send everything to everyone. Every customer needs four questions answered for every category, before it's even considered — the goal is a ranked list of what's most relevant right now, not a menu of twenty possible campaigns.

1 Eligibility
Can this even be offered? Already purchased, wrong fit, no regional availability, wrong lifecycle stage.
2 Affinity
How likely is this to interest them? Based on views, purchases, clicks, wishlists, past categories, similar customers.
3 Business priority
How much does the business need this category pushed right now?
4 Customer priority
How genuinely useful is this to the customer at this specific moment?

04 · Scoring

The priority matrix

A campaign should win a send slot because it scores highest — not because a stakeholder pushed hardest for it. A simple weighted score turns four eligible campaigns into a clear, defensible order.

Priority score = relevance + business value + urgency fatigue risk conflict risk
OrderCandidate campaignWhy
1Complaint follow-upOpen negative context — must close before anything else lands
2Delivery updateTransactional, time-sensitive, expected by the customer
3New product launchSent only once context is clear — high business value, no conflict
4Weekend promoLowest priority — sent only if the customer isn't already overloaded

05 · Guardrails

Contact policy & frequency caps

Without caps, a thousand well-targeted campaigns still add up to noise. The policy needs to be explicit — daily and weekly limits overall, per-channel limits, and cooldowns after specific events.

Message type / rulePolicy
TransactionalAlways allowed — bypasses frequency caps
Critical / serviceAlways allowed
Sales promoMax 2 per week
Survey / NPSNot within 7 days of a complaint
Push notificationsMax 1 per day, unless critical
Loyal / high-value customersLower promo pressure, higher relevance threshold to qualify

06 · Channel strategy

Channels have roles — omnichannel isn't "send everywhere"

The question isn't "should we duplicate this across channels." It's "which channel carries this specific meaning best" — matched to the customer state and the intent of the message.

Push
Quick action, urgency, reminders.
Email
Detail, education, richer content.
SMS
High importance, short, time-sensitive.
In-app
Contextual, while the customer is already engaged.
WhatsApp / LINE / KakaoTalk
Conversational, service-oriented, high-trust markets.
Call
High-value, complex, sales or service moments.

07 · Content system

Personalization as modules, not manual work

Hand-personalizing every message doesn't scale. What scales is a library of content blocks that a CRM engine — or a human — can assemble within rules, based on the customer's state and signals.

Hero message Category block Offer block Reason-to-believe CTA Local market adaptation Lifecycle-specific copy Product recommendation Service / context block
Customer stateAt risk
Category affinityAccessories
Recent signalViewed product twice
ChannelEmail
ToneHelpful, not aggressive
CTAExplore / Save / Book a consultation

08 · Governance

A campaign control tower

At scale, orchestration needs one place where the whole system is visible — whether that's a BI dashboard, a CRM calendar, a project tool paired with a dashboard, or a CDP decisioning layer. The tool matters less than having a single view.

Active campaigns and their owners
Segments touched and audience overlaps
Where campaigns conflict
Priority and frequency impact
Suppressions in effect
Market / local adaptations
Results, in one place
A single, shared source of truth

09 · Rules

Write the conflict rules down in advance

This is where orchestration actually becomes real — not a philosophy, but a set of if/then rules the system enforces automatically.

IfThen
Customer is in a complaint flowSuppress sales promo
Customer just purchasedSuppress acquisition offers
Customer received a high-priority service messageDelay non-critical promo
Customer hasn't opened 5 messages in a rowReduce send frequency
Customer shows high intentPrioritize conversion support
Customer is loyal / high valueDon't over-index on discounting
Customer fits multiple categoriesPick the single top category by relevance score

10 · Measurement

Measure the system, not just the campaign

Open rate, CTR, and conversion on a single campaign can be misleading on their own — a campaign can convert today and quietly erode engagement next month. The system needs its own set of metrics.

Campaign-level (necessary, not sufficient)
  • Open rate
  • Click-through rate
  • Conversion rate
  • Revenue per campaign
System-level (what actually tells you it's working)
  • Incremental lift vs. holdout / control
  • Unsubscribe & opt-out impact
  • Fatigue score, engagement trend by segment
  • Revenue and LTV per customer
  • Cannibalization between campaigns
  • Retention / repeat purchase
  • Complaints after a campaign send

Want this applied to your CRM stack?

This is the framework I bring into consulting engagements and in-house roles — adapted to the platforms, markets, and stakeholders you're actually working with.

LinkedIn Profile · ↑ Back to top

© 2026 Anastasia Valiaeva