Tracking gaps
Tracking gaps are missing pieces of instrumentation: behaviors users do that nothing in the analytics is recording. They surface as a byproduct of running analysis, when the practitioner keeps noticing behaviors that would have been worth examining (mentioned in marketing copy, asked about during onboarding, described in reviews) but aren't in the event logs because no one instrumented them. The team gets a short list of what to start tracking alongside the main analysis, so the next round of analysis can examine what this round couldn't.
Definition
Behaviors the offering supports but doesn't instrument, surfaced as a side effect of running multiplier analysis.
Where gaps surface
Tracking gaps fall out of the practitioner's normal sourcing work. While pulling candidate behaviors from marketing copy, onboarding flows, feature lists, reviews, and retracing the event logs, the practitioner routinely encounters behaviors that belong in the analysis but can't be analyzed because the offering doesn't track them. The marketing site promises "connect your team in one click"; nothing in the event catalog represents a connection event. Onboarding asks users to set up their first integration; the integration_connected event doesn't exist. Reviews describe a workflow the offering supports; the steps of that workflow aren't instrumented.
Retracing produces gaps in a complementary way. Reading patterns in the event logs surfaces what's behaving interestingly, but it also surfaces what's absent — behaviors the offering supports that would have shown up if instrumented. Substituting events is the same observation from a third angle: when the ideal commercial-outcome proxy isn't tracked and the analysis falls back to a pre-conversion event, the missing proxy goes on the gap list rather than getting absorbed silently into the workaround.
What each entry contains
Each gap entry names four things:
- The behavior itself — what would be tracked if instrumented.
- Why it's plausibly worth tracking — usually that the behavior is named in marketing copy, asked about in onboarding, described in reviews, or implied by retracing patterns.
- What multiplier analysis the tracking would enable — what signup-cohort question becomes answerable once the event exists.
- What downstream analyses the new tracking would support — cohort comparisons, slicing investigations, path-investment decisions that the event makes possible.
The entries are structured for actionability. The client receives a list a product team can hand to engineering with enough rationale to defend the instrumentation work in a roadmap conversation.
Compounding value over time
The annotation compounds as gaps get filled. Each newly tracked event makes a set of analyses possible that the current engagement couldn't run — multipliers against the new behavior, slicing of the new behavior, composition with other behaviors. A 1I diagnostic delivered against a sparsely-instrumented offering produces a modest analysis and a substantial gap list; six months later, after the offering has shipped the recommended tracking, the next diagnostic has more analytical surface to work with. The annotation converts incidental observations made during this engagement into capability that pays off in the next one.
Commercial honesty
Naming gaps explicitly keeps the engagement's framing honest. A diagnostic that pretends the available analysis is complete — when it's actually bounded by what's tracked — misrepresents what was found. The framing that survives client scrutiny is "here's what we could analyze given what's tracked; here's what we'd need tracked to analyze further." Bounding the analysis explicitly is what stops the client from reading multiplier output as exhaustive.
Example
A B2B SaaS offering has an onboarding flow that asks "What integration would you like to set up first?" and a marketing site that prominently features integrations as a value proposition. The event catalog tracks signup, paywall_viewed, trial_started, and first_paid_conversion — but no integration_connected event. Multiplier analysis against signups → first paid conversion produces a clean ranked list of candidate behaviors, but the tracking-gaps section flags: behavior is "user connected at least one integration"; worth tracking because integration setup is the offering's stated activation move and prominent in onboarding and marketing; would enable multiplier analysis on integration-connected vs. non-connected signups; would support sliced analysis across integration types (Slack, GitHub, Figma) and cohort comparisons of single- vs. multi-integration users. The client walks away with the multiplier analysis they paid for plus a roadmap item that pays off in subsequent diagnostic work.