ethos
← Back to Projects
📊

Drift

Contract vs reality monitoring

Thesis

Contracts say one thing, reality drifts another. Document parsing + email integration can surface mismatches automatically.

The Problem

Scope creep is a silent killer: The contract says 5 pages, you're on page 12. The timeline said 6 weeks, you're in month 3. The client asks for "one small change" for the 47th time. By the time you notice, you've already eaten the margin. Nobody tracks this in real-time because it's tedious. You only realize how bad it got during the post-mortem.

Implementation Approaches

Approach 1

Email + Document Monitor (Recommended)

Recommended

Watch communication channels and flag drift automatically

⏱️2-3 weeks
📊High complexity

Implementation

  • Connect email, Slack, and project tools
  • Parse original contract/SOW for scope boundaries
  • AI monitors communication for scope expansion signals
  • Alerts: 'Client requested X which is outside original scope'
  • Weekly drift report with quantified impact

Pros

  • +Catches drift as it happens, not after
  • +Quantifies the problem for client conversations
  • +Flywheel: learns what 'drift signals' look like across projects
  • +High value, directly protects margin

Cons

  • Requires broad data access (email, Slack, docs)
  • Privacy and security concerns
  • False positives could be annoying
  • Complex integration requirements
Approach 2

Manual Check-In Tool

Periodic scope review with AI assistance

⏱️1 week
📊Low complexity

Implementation

  • Upload contract + current deliverables list
  • AI compares and highlights mismatches
  • Manual trigger, not continuous monitoring
  • Generates scope change documentation

Pros

  • +Much simpler to build
  • +No ongoing integration maintenance
  • +User controls when to check
  • +Lower privacy concerns

Cons

  • Reactive, not proactive
  • Depends on user remembering to check
  • Misses real-time drift signals
Approach 3

Time Tracking Integration

Compare hours logged vs hours scoped

⏱️1-2 weeks
📊Medium complexity

Implementation

  • Connect to Harvest, Toggl, Clockify, etc.
  • Map time entries to contract line items
  • Alert when actuals exceed estimates by threshold
  • Show burn rate and projected overrun

Pros

  • +Concrete, quantifiable signal
  • +Time tracking data already exists
  • +Clear ROI story
  • +Simpler than full communication monitoring

Cons

  • Only catches time drift, not scope drift
  • Depends on accurate time tracking
  • Lagging indicator, damage already done

Validation Plan

Hypothesis to Test

Agencies will pay $99/mo to catch scope creep before it kills their margins

Validation Phases

1

Retrospective Analysis

1 week
  • Get 3 completed projects with known scope creep
  • Analyze: contract vs final deliverables vs communication
  • Identify: when did drift signals first appear?
  • Show agencies: 'here's where you could have caught it'
Clear pattern of early warning signals that were missed
2

Time-Based MVP

2 weeks
  • Build time tracking integration (Harvest first)
  • Compare actuals to contract estimates
  • Alert on overruns with context
  • Pilot with 3 agencies on active projects
At least one 'caught it early' moment per pilot
3

Full Monitoring Beta

3 weeks
  • Add email/Slack monitoring for one pilot
  • Test drift detection accuracy
  • Refine alert thresholds to minimize noise
  • Validate $99/mo pricing
Pilot agency renews, false positive rate under 20%

Kill Criteria

Stop and move on if any of these become true:

  • Agencies don't grant needed data access
  • Too many false positives make it noisy
  • Drift detection isn't accurate enough to be useful
  • Problem isn't painful enough to justify $99/mo