Back to Writing

Integration Complexity: A Mental Model

A framework for evaluating and prioritizing integration work.

The Problem with Integration Prioritization

Every B2B product eventually faces the integration question: “Which third-party tools should we connect to?” The backlog fills up with requests—Salesforce, HubSpot, Slack, Zapier, Jira, the list never ends.

Without a framework, prioritization becomes political. Whoever has the loudest customer or the most persistent sales rep wins. This leads to a patchwork of integrations, some heavily used, others gathering dust.

I needed a systematic way to evaluate integration requests that balanced customer value against technical cost.

The Framework

Every integration gets scored on three dimensions:

1. Customer Value (0-10)

This isn’t just “how many customers want it”—it’s a weighted assessment:

FactorWeight
Request frequency25%
Revenue at risk (churn/expansion)30%
Segment alignment (ICP fit)25%
Competitive necessity20%

A Salesforce integration requested by 50 enterprise customers threatening to churn scores higher than a niche tool requested by 5 SMB accounts.

2. Technical Complexity (0-10, inverted)

Lower is better. Factors include:

  • API quality: Is the docs good? Is there a sandbox? Rate limits?
  • Data model alignment: How well does their object model map to ours?
  • Authentication complexity: OAuth 2.0 is straightforward; custom SAML is not.
  • Maintenance burden: Will this break every time they update their API?

Score starts at 10 (trivial) and decreases based on complexity factors.

3. Strategic Alignment (0-10)

Does this integration support where we’re going?

  • Reinforces our positioning in target segment: +3
  • Enables a new use case we’re betting on: +3
  • Creates a moat (hard for competitors to replicate): +2
  • Partner co-marketing opportunity: +2

An integration that’s technically easy but strategically irrelevant shouldn’t jump the queue.

The Prioritization Score

Priority Score = (Customer Value × 0.4) + (Technical Feasibility × 0.3) + (Strategic Alignment × 0.3)

This weights customer value highest (we’re here to serve customers) while ensuring we don’t ignore technical reality or strategic direction.

Worked Example

Integration Request: HubSpot CRM sync

DimensionScoreReasoning
Customer Value8High request volume, 3 enterprise accounts contingent on it, strong ICP fit
Technical Complexity7Good API, decent docs, but complex object model for custom properties
Strategic Alignment6Supports mid-market push, but Salesforce is bigger priority

Priority Score: (8 × 0.4) + (7 × 0.3) + (6 × 0.3) = 3.2 + 2.1 + 1.8 = 7.1

Compare this to:

Integration Request: Obscure industry-specific CRM

DimensionScoreReasoning
Customer Value3Only 2 requests, small accounts
Technical Complexity4Poor API, no sandbox, custom auth
Strategic Alignment2Not in our target segment

Priority Score: (3 × 0.4) + (4 × 0.3) + (2 × 0.3) = 1.2 + 1.2 + 0.6 = 3.0

HubSpot wins. By a lot. And now you have a rationale to share with the sales rep pushing for the niche CRM.

Using the Framework in Practice

Quarterly Integration Review

Every quarter, we:

  1. Collect all integration requests from the past 90 days
  2. Score each one using the framework
  3. Stack rank by priority score
  4. Allocate engineering capacity to top N integrations

Saying No (With Data)

The framework makes “no” easier. Instead of “we’re not doing that,” it’s “here’s the score—it didn’t make the cut this quarter, but we’ll re-evaluate as the data changes.”

Tracking Post-Launch

After shipping an integration, we track:

  • Adoption rate (% of eligible customers using it)
  • Support ticket volume
  • Revenue influenced

This feeds back into the model—if our predictions were wrong, we adjust the weighting.

Limitations

No framework is perfect. This one struggles with:

  • Truly novel integrations where there’s no historical signal
  • Platform bets (like Zapier) that enable many use cases indirectly
  • Customer concentration risk where one huge customer skews the data

For these edge cases, judgment still matters. The framework is a starting point, not a replacement for thinking.

The Outcome

Since implementing this framework:

  • Integration backlog is transparent and defensible
  • Engineering time is allocated to highest-impact work
  • Sales has a clear answer for “when are we building X integration”
  • We’ve shipped fewer integrations, but they’re more adopted

The goal was never to build more integrations—it was to build the right ones.