Memory Notch

Every agency retainer carries a hidden cost: the hours senior staff spend re-answering questions that were already solved on a previous client engagement. Knowledge management platforms now close that gap by detecting when documentation drifts from reality and routing fixes for human review, which converts tribal memory into a billable asset instead of a turnover liability.

Documentation Decay Half-Life is the interval between a process changing and the wiki page describing it going wrong. Every agency process has one: a client onboarding checklist, a QA rubric, a reporting template. The half-life shortens as delivery velocity rises, so a retainer running weekly campaign iterations decays faster than a quarterly strategy engagement. The framework matters because stale docs do not fail loudly; they fail as rework, repeated questions, and inconsistent client delivery when a new hire follows a page nobody has touched in eight months. Slite attacks this by monitoring Slack, GitHub, and Linear for drift and drafting fixes for human review, while GitBook runs a stale-content agent and Guru auto-verifies pages as usage increases. The practical move is to measure half-life per document class, then assign review cadence by decay rate rather than by calendar. A 30-day half-life page needs an owner and a trigger; a 12-month page needs almost nothing.

Retrieval Friction Floor is the minimum effort a team member must spend to find an answer before they give up and ask a colleague instead. Every knowledge platform has one, and it sets the ceiling on adoption regardless of how much content sits inside. If a search takes four clicks, a channel switch, or a login the person does not have open, the answer effectively does not exist for them. Agencies feel this hardest during delivery crunches and handoffs, when a project manager needs a client's approved tone rules in under a minute. Guru attacks the floor with permission-aware answers surfaced inside existing workflows, while Tettra pushes answers into Slack channels where questions already get asked. Slab's Unified Search pulls results from Slack, Google, and Asana without leaving the wiki. Lower the floor before adding more documentation, because a fast answer to a partial question beats a slow answer to a complete one.

Turnover Knowledge Half-Life is the rate at which undocumented operational knowledge loses its usable value after the person who held it leaves the agency. Unlike documentation decay, which is gradual and continuous, turnover half-life is a step function: a single departure can erase months of accumulated client context, process nuance, and relationship history in one day. The framework matters because agencies price retainers on delivery consistency, and a senior strategist or account lead walking out the door can reset onboarding timelines for every client they touched. The practical countermeasure is not more documentation volume but capture at the point of decision: recording why a deliverable was scoped a certain way, not just what was produced. Slite's approach of monitoring Slack, GitHub, and Linear to detect when documentation drifts from reality is one example of tooling that shortens the half-life window by flagging gaps before a departure exposes them.

Measure retrieval success on real client questions before adding or replacing a knowledge platform, then buy only the layer that fixes the measured failure.

Assign a named owner and a drift-detection trigger to every critical document before you evaluate, buy, or migrate a knowledge platform.

IF your agency's delivery quality depends on answers that must be provably current (client SOPs, compliance steps, pricing rules, onboarding paths), THEN invest in a governed knowledge base that owns the canonical copy and enforces verification. IF your knowledge already lives across Slack, Google Drive, Asana, and client-owned wikis that you cannot migrate, THEN buy a federated search layer that indexes those systems in place and accept that accuracy depends on the source tools.

A fixed-scope engagement that moves a client's scattered SOPs, onboarding guides, and delivery playbooks into one governed knowledge platform, then wires it into the tools the team already uses. The offer targets the two costs agencies feel most: slow onboarding and knowledge loss when a delivery lead leaves.