Slack vs Glitch Behavior Fit #

Jason Hreha· Updated September 5, 2026

Key Result (company-reported): Slack’s S-1 reports more than 1 billion messages during the week ended January 31, 2019 and 143% net dollar retention as of that date. These measure aggregate activity and revenue retention, respectively. BS-0006

Background #

Slack followed the team’s unsuccessful online game, Glitch. That sequence can prompt questions about the next target behavior, but it is not a natural experiment. The population, product, market, timing, and commercial model changed. A shared founding team does not hold those factors constant.

Slack’s filing describes an internal communication tool that became a public product in 2014. Butterfield’s earlier interview describes testing with other companies and revising the product as larger teams used it. These are participant accounts, not a causal comparison with the game. The sources do not document use of DRIVE, BMF, or the Behavior Fit Assessment.

Behavior and delivery chain #

The action considered here is sending and reading work messages in shared channels. Its familiarity is a reason to investigate it, not evidence that every team can adopt it without new learning or coordination.

A useful analyst reconstruction includes more than the sender. Someone establishes the workspace and its access rules. Colleagues join the relevant conversation, interpret the message, and respond or act. An administrator or operator supplies permissions, support, and any integration required for the task. These dependencies should be observed in the intended workplace rather than assumed from the aggregate message count.

A team can send many messages while still duplicating work, missing important information, or creating interruptions. A proposed field test should therefore attach useful coordination outcomes and the cost of maintaining the workflow to its activity measures. These are research recommendations, not results reported by the case.

Case facts
Company / system	Slack (Tiny Speck)
Industry	Enterprise Software
Population	Workplace teams
Target behavior	Send and read team messages in channels as part of daily coordination
Window	Week ended January 31, 2019 for messages; January 31, 2019 revenue-retention snapshot against the same paid-customer cohort twelve months earlier
Denominator	Aggregate message count across Slack; existing paid-customer revenue cohort for net dollar retention
Key metric	1B+ messages in the week ended Jan 31, 2019; 143% net dollar retention as of Jan 31, 2019 (SEC S-1)
BFA version	2.0 (case-summary-categorical-v1)
Behavior fit	Dispositional Fit: High (illustrative analyst label: existing coordination needs are plausible, but preference for this workflow must be tested) Capability Fit: High (illustrative analyst label: basic messaging can be familiar, while channels, search, integrations and accessibility still require assessment) Context Fit: High (illustrative analyst label: team participation, employer approval and communication norms determine whether the workflow is usable) High, Medium, and Low are categorical analyst labels for case comparison, not numeric scores or direct measurements.
Confidence	Working
Evidence	BS-0006

Behavior Fit Assessment #

The categorical labels in the facts box are illustrative analyst judgments with incomplete evidence, not measured scores. The public record does not support a defensible low-fit profile for every potential Glitch player or a high-fit profile for every workplace team.

Dimension	Hypothesis to investigate	Evidence needed in a new setting
Dispositional Fit	Some teams may prefer persistent shared communication for recurring work	Observed choices between plausible communication routes, with reasons and contrary cases
Capability Fit	Familiar message entry may make a first action feasible	Completion of the actual workflow, including channel selection, retrieval, access needs, and training
Context Fit	The behavior depends on colleagues and organizational conditions	Team participation, permissions, time, competing tools, and support demands

Do not infer that a framework applied before the game launched would have predicted the outcome. To practice a decision with stated alternatives and unknowns, use the M-PESA teaching packet or Instagram packet.

Results and denominators #

The filing’s message count is not a completion or retention rate. Its net dollar retention compares revenue from the same paid-customer cohort over time; it is not the percentage of people or teams retained.

In the founder interview, the often-cited 93% continued-use figure applies to teams that had already exchanged 2,000 messages. The interview does not supply a common fixed follow-up period for that group. This selected-cohort association cannot show that causing a new team to reach the milestone will cause it to stay. BS-0006

Limitations #

Company reporting and founder recollection can help identify hypotheses, but they leave selection, operational support, and competing explanations unresolved. The case does not estimate the isolated contribution of behavior selection, product quality, or market conditions. Failure of the earlier game also cannot establish that its users were incapable of the required actions.

Practical lesson #

Treat a familiar behavior as a candidate. Compare the whole delivery chain with other ways to achieve the outcome, including reducing the work users perform. Test whether the people and supporting actors can complete it under realistic conditions. Keep usage milestones, useful outcomes, and commercial results separate when deciding what to do next.