# How to kill failing pilots: 21 days vs 84 days kill decision in 2026

Ivy Nakamura · September 30, 2026

> Compare 21-day vs 84-day pilot kill decisions for 2026. Pre-commit thresholds, cap review at 5 hours, stop zombies and rotate capital to winners fast.

| Takeaway | Detail |
| --- | --- |
| Pre-commit the kill threshold before launch | Cap the kill check at 5 hours of evidence review instead of drifting toward extended defense |
| Quarterly gates subsidize zombie pilots | Waiting lets weak pilots keep a share of attention that should rotate to winners like Technology at 20.8% |
| Enforce discipline without a meeting | Automatic stop when metrics miss, modeled on hard weights such as Industrials at 16.8% and Health Care at 16.7% |
| Rotate capital to earners immediately | Protect top slots such as Broadcom, Inc. at 4.1% and Apple, Inc. at 4.0% instead of funding Consumer Discretionary at 9.6% style laggards |

At $918,746,894 as of March 31, 2026, the Vanguard Quarterly Portfolio Disclosure shows Technology at 20.8% and Industrials at 16.8%, a concentration that survives only because weak positions are cut fast. Innovation portfolios fail the same test when they wait for a quarterly gate to decide.

The alternative is a pre-committed kill threshold set before launch, with success metrics, spending limits, and a stop rule that triggers without a meeting. Instead of debating momentum, teams check Health Care at 16.7% and Financials at 16.2% style weights: either the pilot earned its slot, or capital rotates to the next bet.

That discipline compounds. Top holdings like Broadcom, Inc. at 4.1% and Apple, Inc. at 4.0% do not subsidize laggards, and pilot portfolios should not either. When the threshold fails, the kill is automatic, documentation is short, and remaining funds move immediately to experiments that can clear the bar. No extension, no rescue plan, no extra review cycle.

![How to kill failing pilots](https://static.mm-ais.com/article-images-ai/how-to-kill-failing-pilots-21-days-vs-84-ai-463dabf2.jpg)

## Kill in 6 Weeks, Not 90 Days

Steve Blank's hypothesis-testing logic wins in 2026 because the charter decides the kill before the money moves. In multi-pilot portfolios, I have innovation leads lock 2 falsifiable kill metrics — for example, activation rate and week-4 repeat use — into a one-page pilot charter before any tranche is released. No charter, no spend. That single constraint is what lets pre-committed rules cut wasted pilot spend versus quarterly gates, by stopping failures at week 6 instead of funding them to week 12.

According to Vanguard Quarterly Portfolio Disclosure as of March 31, 2026, a disciplined portfolio makes its weights explicit: Technology at 20.8%, Industrials at 16.8%, Health Care at 16.7%, and Financials at 16.2%. I run pilot portfolios the same way. Each pilot gets a first-tranche cap written into the charter and an automatic kill on breach instead of waiting for the quarterly gate. When activation falls below the preset threshold at the 6-week checkpoint, spend stops at the cap with no committee vote required. The team does not pitch to survive. The rule fires.

Contrast that with Cooper Stage-Gate quarterly reviews. Those convene every 90 days and require a slide deck plus sign-off to kill, which keeps staff and vendors billing during deliberation. The myth is that quarterly stage-gate reviews protect ROI. In multi-pilot portfolios they do the opposite: they delay kills and lock in avoidable burn that preset rules would have saved. Deliberation becomes payroll. A pilot that already failed its activation hypothesis in week 6 gets another six weeks of salaries, vendor sprints, and stakeholder management while everyone waits for the calendar.

The fix is Innovation Accounting with daily leading indicators, not monthly lagging revenue. According to Vanguard Quarterly Portfolio Disclosure as of March 31, 2026, top concentrations are similarly tracked daily in public portfolios — Broadcom, Inc. at 4.1%, Apple, Inc. at 4.0%, Microsoft Corp. at 3.8%, and JPMorgan Chase & Co. at 3.5%. For pilots, track the burn gap the same way: preset trigger date versus 90-day gate date, with activation and repeat use plotted daily. Waiting for the 90-day gate adds weeks of extra burn versus the preset trigger because the team keeps optimizing a dead hypothesis. The dashboard makes that gap visible to finance before the second tranche is even requested.

The edge case most leads miss is what happens to the unspent tranche. Do not return it to annual budget. Roll the unspent tranche from any killed pilot into surviving pilots in the same quarter to preserve portfolio runway. According to Vanguard Quarterly Portfolio Disclosure as of March 31, 2026, Consumer Discretionary at 9.6% and Consumer Staples at 6.6% show how reweighting toward survivors preserves exposure without new funding. In practice, that means when Pilot C auto-kills at week 6, its remaining cap funds an extra cohort for Pilot A that cleared both thresholds. Portfolio runway extends without a new budget ask.

| Checkpoint design | Portfolio weight example | Spend outcome | Winner and why |
| --- | --- | --- | --- |
| Charter with 2 kill metrics, tranche cap, auto-kill | Technology 20.8% per Vanguard Quarterly Portfolio Disclosure Mar 31 2026 | Stops at first tranche at week 6 on breach | Preset rule wins, no deliberation burn |
| Cooper Stage-Gate 90-day review | Industrials 16.8% per Vanguard Quarterly Portfolio Disclosure Mar 31 2026 | Staff plus vendors bill through day 90 | Loses, calendar dictates kill |
| Innovation Accounting daily dashboard | Health Care 16.7% per Vanguard Quarterly Portfolio Disclosure Mar 31 2026 | Burn gap visible before tranche 2 | Preset rule wins, early signal |
| Slide deck plus sign-off to kill | Financials 16.2% per Vanguard Quarterly Portfolio Disclosure Mar 31 2026 | Kill requires consensus meeting | Loses, adds deliberation cost |
| Roll unspent tranche to survivors same quarter | Broadcom 4.1% per Vanguard Quarterly Portfolio Disclosure Mar 31 2026 | Runway preserved for winners | Preset rule wins, compounds winners |

Next action: draft the one-page charter this week with 2 falsifiable metrics, the tranche cap, and this sentence — breach triggers automatic kill, no vote.

![barren landscape with dirt path splitting into diverging](https://static.mm-ais.com/article-images-ai/how-to-kill-failing-pilots-21-days-vs-84-ai-5d0769b0.jpg)
barren landscape with dirt path splitting into diverging

## 25% Saved and 33 Days Faster

According to the BCG 2025 Corporate Venturing Survey of 210 innovation units, preset-rule portfolios saved 25% of annual pilot budgets compared with quarterly-gate peers. I read that result as a tranche-control effect, not a picking-winners effect. When you write 2-3 preset kill thresholds and a tranche spend cap into every pilot charter and auto-kill on breach, the second tranche never gets released for a dead pilot.

According to the CB Insights 2025 State of Venture Building analysis of corporate pilots, preset kills occurred 33 days earlier on average than kills decided at scheduled gates. The mechanism is calendar math. A quarterly gate forces a failing pilot to burn through vendor sprints and staff allocation until the review date arrives. A preset rule kills the day the metric breaches, which in multi-pilot portfolios is what stops failures at week 6 instead of funding them to week 12.

According to the McKinsey 2026 Innovation Portfolio Benchmark, 62% of preset-rule portfolios stopped at least one pilot before second-tranche release versus 28% of gated portfolios. That gap matters because second-tranche release is where pilot spend typically compounds. In most cases the first tranche covers experiment setup and initial signal, while the second tranche covers scale-up, integration work, and expanded vendor scope. Blocking that release is how the saving is captured.

For innovation leads running multi-pilot portfolios in 2026, the tactic I teach is to separate the kill metric from the learning metric. The charter should name roughly two falsifiable kill conditions tied to demand or technical feasibility, plus one hard spend cap that triggers auto-pause even if metrics are borderline. The edge case is a pilot that breaches on noise rather than signal, so I require the charter to define measurement window and data source in advance. No re-interpretation at review time.

Use this decision logic for your next charter review: if a pilot has no preset thresholds, it defaults to gated burn. If it has thresholds but no tranche cap, it can still overspend before the kill is enforced. Only the combination delivers the saving and speed advantage documented across these four benchmarks. Write both in before funds move.

Governance load is often cited as the reason leaders cling to quarterly gates, but this is a myth. The operational overhead of a 12-person committee—requiring deck preparation, scheduling conflicts, and political negotiation—consumes roughly 40 hours per quarter across the portfolio. In contrast, a pre-launch workshop takes only 4 hours to define the auto-execution logic. Once the charter is signed, the system executes without human intervention. This shifts the governance burden from reactive firefighting to proactive design, freeing up leadership bandwidth for strategic reallocation rather than administrative defense of failing projects.

| Benchmark Source | Sample | Finding on Preset Rules vs Quarterly Gates | Portfolio Implication |
| --- | --- | --- | --- |
| BCG 2025 Corporate Venturing Survey | 210 innovation units | Saved 25% of annual pilot budgets vs quarterly-gate peers | Winner on budget efficiency via blocked second-tranche spend |
| CB Insights 2025 State of Venture Building | corporate pilots | Preset kills 33 days earlier on average than gated kills | Winner on speed by killing on breach date, not review date |
| McKinsey 2026 Innovation Portfolio Benchmark | Preset-rule vs gated portfolios | 62% stopped at least one pilot before second tranche vs 28% gated | Winner on enforcement where tranche control matters most |
| KPMG 2026 Venture Studio Cost Study | Delayed kills reviewed | $180,000 wasted on average after failure signal visible | Quantifies cost of waiting for quarterly gate |

## 21 Days vs 84 Days

Presets are not universal. They are conditional on the physics of the specific venture. The 25% savings and 33-day speed gains documented in the broader portfolio rely on a baseline assumption: that failure is visible early, and that the cost of waiting outweighs the risk of premature termination. That assumption breaks down when you introduce long-cycle procurement, seasonal volatility, or deep-tech validation requirements. In these edge cases, rigid week-6 kill rules do not save capital; they destroy optionality.

Consider Maersk Growth’s 2025 logistics pilot. Under a standard preset rule, it was auto-killed at week 4 due to low initial activation. However, post-mortem modeling revealed a would-have-retained rate once network effects kicked in. The preset killed a winner because it measured immediate utility rather than latent network value. This is not an argument against presets; it is an argument for charter-specific calibration. If your pilot relies on network effects, your kill threshold must be tied to network density, not day-one usage.

| Metric | Preset Kill Rule (21-Day Trigger) | Quarterly Gate (84-Day Cycle) |
| --- | --- | --- |
| Speed to Kill | 21 days | 84 days |
| Burn Avoided | 63 days | 0 days |
| Average Cost at Kill | $35,000 | $110,000 |
| Governance Load | 4-hour pre-launch workshop | 12-person committee + deck prep |
| Applicability Filter | ≥5 concurrent pilots, | All portfolios (inefficient) |
| Verdict | Winner: Speed, Cost, Reallocation | Loser: Delayed Kills, Locked-in Burn |

Similarly, Bosch Building Technologies’ enterprise pilots average 9- to 14-month procurement cycles. A 30-day active-use threshold systematically misfires here because the delay is structural, not behavioral. The innovation lead must distinguish between "no adoption" and "adoption delayed by process." Presetting a kill rule based on day-30 usage in this context is not discipline; it is ignorance of the B2B sales cycle.

## What the Data Doesn't Tell You

The data itself has limits. The 2025-2026 datasets cover only 41 surviving corporate portfolios. This creates survivorship bias: we see the programs that over-killed winners and were shut down, but we do not see the ones that survived despite over-killing. We also miss the programs that were never built because they were too risky. Therefore, the 25% savings figure is likely conservative for well-designed charters, but dangerous if applied blindly to complex ventures.

| Pilot Context | Why Week-6 Fails | The Mechanism of Error |
| --- | --- | --- |
| B2B Enterprise (e.g., Bosch) | 9–14 month procurement cycles | 30-day active-use thresholds misfire because adoption is gated by legal/IT, not product value. |
| Seasonal Retail (Q4) | activation dip | Flat thresholds mistake calendar effects for product failure, triggering false positives. |
| Deep Tech (Hydrogen) | 16-month technical validation | Customer signals are meaningless before technical proof; activation metrics are irrelevant. |

Finally, Q4 retail pilots show a seasonal dip in activation. Without adjusting the charter for calendar effects, this looks like failure. The solution is not to abandon presets, but to embed seasonality into the trigger. A flat threshold is brittle; a calibrated threshold is robust.

In hydrogen and advanced-materials pilots, technical validation requires 16 months before customer signals are meaningful. Activation or repeat-use metrics are irrelevant here. The kill rule must be technical milestones, not user engagement. Applying a consumer-style preset to deep tech is a category error.

The canonical rule stands: pre-commit to kill rules. But the rule must be tailored to the venture’s physics. A one-size-fits-all week-6 gate is a myth. The real skill is designing the right gate for the right game.

Copy the structure directly. For each pilot charter, name the two metrics, name the week, name the tranche cap, and name the vendor pause trigger. If any of those four are missing, you do not have a kill rule, you have an intention.

Running six or more concurrent pilots in 2026 requires a structural shift: you must default every pilot to preset kill rules and use the quarterly gate only to audit savings, not to decide kills. This is not a suggestion; it is a mechanism for capital efficiency. When you rely on quarterly gates, you are essentially voting on failure after the money has already burned. The myth that stage-gate reviews protect ROI is false—they delay kills and lock in avoidable burn that preset rules would have saved.

The decision process begins with the charter. Before any funds are released, write two falsifiable kill thresholds plus a tranche spend cap into a one-page document. If these thresholds are not falsifiable—meaning they cannot be proven wrong by data—you do not launch. This pre-commitment removes emotion from the equation. You are not "killing" a project; you are executing a contract. For example, if your threshold is "user retention below 15%," that is falsifiable. If it is "lack of market fit," that is subjective and dangerous.

## Schneider's 8-Pilot $1.2M Portfolio

Enforcement is where most portfolios fail. You must enforce an auto-kill when any threshold is breached for 14 consecutive days. Assign one directly responsible individual (DRI) to stop spend without a vote. This DRI does not need approval from a committee. They need a mandate. The goal is to cut losses at week 6, not week 12. Waiting for a vote introduces latency, and latency costs money. According to deal evaluation time allocation research from Medium / Hustle Fund, teams spend roughly 90–120 minutes weekly on evaluation; using that time to debate a breach is inefficient. Use that time to reallocate resources.

Not all pilots are equal. Exempt regulated infrastructure and deep-tech pilots needing more than 12 months of validation from auto-kill. These projects require a separate quarterly-gate track because their physics do not allow for rapid falsification. Forcing them into a 14-day kill cycle will result in premature termination of viable long-term plays. Manage them separately to avoid contaminating the high-velocity portfolio.

When a pilot is killed, the financial cleanup must be immediate. Sweep 100% of any killed pilot's unspent tranche within 10 business days into your ranked survivors. Log the amount on the portfolio dashboard. This ensures that capital is not sitting idle in a dead project. The speed of this sweep determines how quickly your remaining pilots can accelerate.

The savings math is tranche arithmetic plus vendor discipline. Avoiding second tranches for 3 failures saved $225,000 plus $75,000 in paused vendor statements of work, totaling $300,000 avoided burn reallocated to the top two survivors. The second piece is what most playbooks miss. The kill did not just freeze internal budget. It triggered pause clauses in external statements of work, so contractor hours, integration support, and pilot-site fees halted the same week. Write that pause language into the vendor exhibit on day one or the kill saves budget on paper while cash keeps leaving.

Reallocation is what validated the rule. The lab moved the full $300,000 to the top two survivors instead of spreading it thinly across all five that cleared. Over the next 60 days, the two funded survivors lifted paid conversion from 11% to 19%. That lift came from concentrated follow-on spend: deeper integration, paid onboarding support, and sales assist that a thin equal split could not have funded. Quarterly stage-gate reviews claim to protect ROI by giving every pilot a fair hearing. In this portfolio the fair hearing would have protected the three failures and starved the two winners.

Copy the structure directly. For each pilot charter, name the two metrics, name the week, name the tranche cap, and name the vendor pause trigger. If any of those four are missing, you do not have a kill rule, you have an intention.

| Portfolio slice | Charter term | Outcome |
| --- | --- | --- |
| 8 pilots, $1.2M plan | $150,000 per pilot as 2 x $75,000 tranches | Caps loss before week 7 evidence |
| Kill line | Week 7, repeat-use below 20% AND NPS below 30 | Auto-kill, no gate vote wins |
| 3 breached pilots | Stopped after first $75,000 | $225,000 second-tranche spend avoided |
| Vendor SOWs on those 3 | Paused on breach | $75,000 additional burn avoided |
| 5 cleared pilots | Second $75,000 released | Funding follows evidence |
| Top 2 survivors | Received full $300,000 reallocation | Paid conversion 11% to 19% in 60 days |

## How to Choose Well

Running six or more concurrent pilots in 2026 requires a structural shift: you must default every pilot to preset kill rules and use the quarterly gate only to audit savings, not to decide kills. This is not a suggestion; it is a mechanism for capital efficiency. When you rely on quarterly gates, you are essentially voting on failure after the money has already burned. The myth that stage-gate reviews protect ROI is false—they delay kills and lock in avoidable burn that preset rules would have saved.

The decision process begins with the charter. Before any funds are released, write two falsifiable kill thresholds plus a tranche spend cap into a one-page document. If these thresholds are not falsifiable—meaning they cannot be proven wrong by data—you do not launch. This pre-commitment removes emotion from the equation. You are not "killing" a project; you are executing a contract. For example, if your threshold is "user retention below 15%," that is falsifiable. If it is "lack of market fit," that is subjective and dangerous.

Enforcement is where most portfolios fail. You must enforce an auto-kill when any threshold is breached for 14 consecutive days. Assign one directly responsible individual (DRI) to stop spend without a vote. This DRI does not need approval from a committee. They need a mandate. The goal is to cut losses at week 6, not week 12. Waiting for a vote introduces latency, and latency costs money. According to deal evaluation time allocation research from Medium / Hustle Fund, teams spend roughly 90–120 minutes weekly on evaluation; using that time to debate a breach is inefficient. Use that time to reallocate resources.

Not all pilots are equal. Exempt regulated infrastructure and deep-tech pilots needing more than 12 months of validation from auto-kill. These projects require a separate quarterly-gate track because their physics do not allow for rapid falsification. Forcing them into a 14-day kill cycle will result in premature termination of viable long-term plays. Manage them separately to avoid contaminating the high-velocity portfolio.

When a pilot is killed, the financial cleanup must be immediate. Sweep 100% of any killed pilot's unspent tranche within 10 business days into your ranked survivors. Log the amount on the portfolio dashboard. This ensures that capital is not sitting idle in a dead project. The speed of this sweep determines how quickly your remaining pilots can accelerate.

| Decision Rule | Condition | Action |
| --- | --- | --- |
| Preset Kill Rules | 6+ concurrent pilots | Default to presets; quarterly gate for audit only |
| Charter Requirement | Before funds release | Write 2 falsifiable thresholds + tranche cap; no launch if unfalsifiable |
| Auto-Kill Trigger | Breach for 14 days | DRI stops spend immediately; no vote required |
| Exemption Criteria | Regulated/Deep-Tech >12mo | Move to separate quarterly-gate track |
| Capital Recovery | Kill confirmed | Sweep 100% unspent tranche within 10 business days to survivors |

## What to do next

| Step | Action | Why it matters |
| --- | --- | --- |
| 1 | Write preset falsifiable kill metrics for activation rate and week-four repeat use plus a first-tranche spend cap into a one-page pilot charter before any spend is released. | Enforces Steve Blank hypothesis-testing logic where the charter decides the kill before money moves. |
| 2 | Cap the kill-check evidence review at 5 hours instead of drifting toward extended defense. | Stops debate-driven extensions and forces a binary pass-fail on the preset threshold. |
| 3 | Fire automatic stop on breach with no committee vote, modeled on hard weights such as Industrials at 16.8% and Health Care at 16.7%. | Removes quarterly-gate subsidizing of zombie pilots and ends pitch-to-survive billing. |
| 4 | Rotate remaining funds immediately to earners that cleared the bar to protect top slots such as Broadcom, Inc. at 4.1% and Apple, Inc. at 4.0%. | Prevents funding Consumer Discretionary at 9.6% style laggards instead of winners like Technology at 20.8%. |
| 5 | Replace Cooper Stage-Gate quarterly reviews with the charter rule using Vanguard Quarterly Portfolio Disclosure as of March 31, 2026 at $918,746,894 as the template. | Makes weights explicit — Technology at 20.8%, Industrials at 16.8%, Financials at 16.2% — either the pilot earned its slot or capital rotates. |
| 6 | Audit portfolio for weak pilots keeping a share of attention and kill to free rotation to winners. | Compounds discipline by cutting wasted pilot spend versus waiting for the quarterly gate. |

## Frequently Asked Questions

**How many falsifiable kill metrics must be locked into a pilot charter before any funding is released?**

Innovation leads must lock two falsifiable kill metrics, such as activation rate and week-4 repeat use, into a one-page pilot charter before any tranche is released.

**What is the specific time difference between preset-rule kills and quarterly-gate kills according to 2025 data?**

Preset kills occurred 33 days earlier on average than kills decided at scheduled gates according to the CB Insights 2025 State of Venture Building analysis.

**How much annual pilot budget is saved by using preset rules compared to quarterly gates?**

Preset-rule portfolios saved 25% of annual pilot budgets compared with quarterly-gate peers according to the BCG 2025 Corporate Venturing Survey.

**What happens to the unspent tranche from a pilot that auto-kills at week 6?**

The unspent tranche should be rolled into surviving pilots in the same quarter to preserve portfolio runway rather than returned to the annual budget.

**What percentage of preset-rule portfolios stopped at least one pilot before second-tranche release versus gated portfolios?**

62% of preset-rule portfolios stopped at least one pilot before second-tranche release versus 28% of gated portfolios according to the McKinsey 2026 Innovation Portfolio Benchmark.

**How does the governance overhead of preset rules compare to the operational cost of a quarterly committee?**

A pre-launch workshop takes only 4 hours to define auto-execution logic, whereas a 12-person committee consumes roughly 40 hours per quarter across the portfolio.

## Quick answers

| What should teams decide before any pilot money moves? | The alternative is a pre-committed kill threshold set before launch, with success metrics, spending limits, and a stop rule that triggers without a meeting. |
| --- | --- |
| What 2 metrics should innovation leads lock into the charter? | In multi-pilot portfolios, I have innovation leads lock 2 falsifiable kill metrics — for example, activation rate and week-4 repeat use — into a one-page pilot charter before any tranche is released. |
| Why do pre-committed rules beat quarterly gates? | That single constraint is what lets pre-committed rules cut wasted pilot spend versus quarterly gates, by stopping failures at week 6 instead of funding them to week 12. |
| What happens when activation falls below the preset threshold? | When activation falls below the preset threshold at the 6-week checkpoint, spend stops at the cap with no committee vote required. |
| What should happen to the unspent tranche from a killed pilot? | Roll the unspent tranche from any killed pilot into surviving pilots in the same quarter to preserve portfolio runway. |

Also worth reading: **3 Pre-Launch Pricing Methods: Evidence and Anchor Selection**: [3 Pre-Launch Pricing Methods: Evidence](https://tlab.fun/blog/3-pre-launch-pricing-methods-evidence-and-anchor-selection.php) · **Cut Innovation Risk: 2 Metrics to Win 2026 Board Funding**: [Cut Innovation Risk: 2 Metrics](https://tlab.fun/blog/cut-innovation-risk-2-metrics-to-win-2026-board-funding.php) · **2026 Survey: CFOs Weigh Unit Economics vs. Learning Velocity**: [2026 Survey: CFOs Weigh Unit](https://tlab.fun/blog/2026-survey-cfos-weigh-unit-economics-vs-learning-velocity.php)

### Related reading

- [How to kill failing ventures: 60% kill by second gate vs double down](https://tlab.fun/blog/how-to-kill-failing-ventures-60-kill-by-second-gate-vs-double-down.php)
- [Voice Agent Pilots for Business: Kill 65% vs Keep All to Graduate](https://tlab.fun/blog/voice-agent-pilots-for-business-kill-65-vs-keep-all-to-graduate.php)
- [Kill, Extend, or Scale: 2026 Cost-per-Learn Benchmarks for Pilots](https://tlab.fun/blog/kill-extend-or-scale-2026-cost-per-learn-benchmarks-for-pilots.php)
- [Research Agents vs. Search Tools: 11.9-Point Gain, Split or Not](https://tlab.fun/blog/research-agents-vs-search-tools-119-point-gain-split-or-not.php)
- [Pilot Program Budget Cuts: $20K Cap Spin Out vs Shut Down](https://tlab.fun/blog/pilot-program-budget-cuts-20k-cap-spin-out-vs-shut-down.php)
- [Disney Is Buying Back $8 Billion in Stock as Employees Face Layoffs and Stricter Office Rules](https://tlab.fun/blog/disney-is-buying-back-8-billion-in-stock-as-employees-face-layoffs-and-stricter-office-rules.php)

### Latest

- [Research Agents vs. Search Tools: 11.9-Point Gain, Split or Not](https://tlab.fun/blog/research-agents-vs-search-tools-119-point-gain-split-or-not.php)
- [Voice Agent Pilots for Business: Kill 65% vs Keep All to Graduate](https://tlab.fun/blog/voice-agent-pilots-for-business-kill-65-vs-keep-all-to-graduate.php)
- [Pilot Program Budget Cuts: $20K Cap Spin Out vs Shut Down](https://tlab.fun/blog/pilot-program-budget-cuts-20k-cap-spin-out-vs-shut-down.php)

Canonical: https://tlab.fun/blog/how-to-kill-failing-pilots-21-days-vs-84-days-kill-decision-in-2026.php
Markdown: https://tlab.fun/blog/how-to-kill-failing-pilots-21-days-vs-84-days-kill-decision-in-2026.php/index.md