Tasks / Growth

Write a GTM plan

Can the model pick the customers who value the difference most, position against what they do today, and build a go-to-market that lasts beyond launch day?

Task type v1.1 · 2 tasksLast changed 6 Oct 2026 · ChangelogDifficulty

What AI gets right here, and what you’ll still have to catch

From 16 graded outputs by 8 models. 50% were usable with at most a quick edit.

Reliably right

Positioned against the real alternative100% pass
It names the real alternatives (accountant-run payroll with mistakes, owner spreadsheet with risk) and ties the message to confidence and accuracy that those alternatives lack.
GPT-6 Astra · ChatGPT · Payroll, for the people who fear it
Protects the self-serve engine100% pass
Uses the 70% free-to-paid stat and product-qualified deal data to argue against capping the free plan, and keeps self-serve open.
GPT-6 Astra · ChatGPT · Enterprise, without killing the engine
Does the capacity sums100% pass
Works out deals needed (69-80), qualified opportunities (134), and AE-years (8) with checkable arithmetic.
GPT-6 Astra · ChatGPT · Enterprise, without killing the engine

Where it slips

Avoids unsupported claims47% pass
Presents the incorrect 'five of six no decision' figure as established fact, which is an unsupported claim about the current situation.
Opus 5.5 · Claude · Enterprise, without killing the engine
Uses the supplied evidence correctly52% pass
The claim 'Five of our six losses on existing-user deals ended in no decision' is not supported by the data (4 no-decision, 2 Miro), so a statement about the current situation is invented.
Opus 5.5 · Claude · Enterprise, without killing the engine
Identifies material uncertainty73% pass
The plan does not name specific unknowns that could change the decision, nor does it say how they would be resolved or what would trigger a pivot.
Gemini 3.8 Flash · API · Payroll, for the people who fear it

How it’s graded

Each check is built on how world-class product and growth leaders do this work, as they’ve described it on Lenny’s Podcast, and names the leader it comes from.

Positioned against the real alternative
Does the plan name what the target customer does today without this product (including the status quo or doing nothing), and tie the message to the value that alternative lacks?
Passes when Names the real alternative and the differentiated value against it, and picks the segment that cares most about that value.
April DunfordLenny’s Podcast
Sells against a pain the buyer feels
Is the core message built on a pain or risk the target buyer already feels, with evidence from the material, rather than on generic upside like saving time?
Passes when Leads with the pain the evidence shows buyers care about, and uses that evidence to choose the message.
Jeanne DeWitt GrosserLenny’s Podcast
Diagnoses message or channel first
Does the plan work out whether the problem is the message (people reach it but don't convert) or the channels (the right people never see it), and spend accordingly?
Passes when Uses conversion and reach data to say which it is, and puts the effort there.
Emily KramerLenny’s Podcast
Plans beyond launch week
Does the plan say how demand keeps coming after launch, with the ongoing engine and what's measured over months, rather than treating launch day as the whole plan?
Passes when Gives the ongoing motion after launch and the measures that show whether it's working over time.
Arielle JacksonLenny’s Podcast

Plus our standard checks

Uses the supplied evidence correctly · Addresses the actual decision · Respects explicit constraints · Identifies material uncertainty · Avoids unsupported claims · Produces the required deliverable · and 3 written for each task, which you’ll see in the tasks below.

The tasks

Read the brief, then compare up to three outputs, each with the LLM judge’s verdict on every check. Highlights mark what a PM had to fix.

The brief

You're the VP Product at Flowboard. The board wants $5M of new enterprise ARR next year, and our CRO, Jordan Hale, has drafted the enterprise go-to-market below. The CEO has asked for your version before the board meeting: who we sell to, how the self-serve product and the sales team work together, what we say against the alternatives, the capacity it takes, and what you'd change in Jordan's plan and why. No more than 1,300 words. The files are attached.

What the model was given6 items: About Flowboard, today.md, cro_enterprise_plan.md, enterprise_deals_last_12_months.csv, large_companies_using_flowboard.csv (companies over 1,000 employees with active self-serve or free teams), enterprise_win_loss_interviews.md

About FlowboardOnline whiteboards and diagramming for teams. Free plan, self-serve paid plans, and an enterprise plan sold by a small sales team.

today.md5 lines · Download

# Today

- Self-serve: 31,000 paying teams; free plan feeds 70% of new paying teams.
- Enterprise: 46 customers, average contract $64,000.
- Sales team: 3 account executives, closing about 10 enterprise deals a year each.

cro_enterprise_plan.md8 lines · Download

# Enterprise plan (CRO, draft for the board)

Target: $5M new enterprise ARR next year.
1. Cap the free plan at 3 users (today: unlimited users, 3 boards each), so teams have to talk to us.
2. Route every team over 10 seats to sales; remove self-serve checkout above 10 seats.
3. Hire 8 account executives and 6 SDRs; SDRs work outbound lists of Fortune 2000 companies.
4. Enterprise features: SSO, SCIM, audit logs, data residency.
…

enterprise_deals_last_12_months.csv65 rows · Download

deal,source,outcome,sales_cycle_days,acv_usd,lost_to
E-200,Outbound,won,201,32700,
E-201,Product-qualified (teams already using Flowboard),won,62,97000,
E-202,Product-qualified (teams already using Flowboard),won,61,108000,
E-203,Outbound,lost,192,,Miro (company standard)
E-204,Outbound,lost,204,,Miro (company standard)
E-205,Product-qualified (teams already using Flowboard),won,77,54200,
…

large_companies_using_flowboard.csv (companies over 1,000 employees with active self-serve or free teams)company,employees,teams_using_flowboard,active_users_30d,paying_seats,viewed_security_page_30d,asked_for_sso Aldergate Bank,15000,2,18,10,yes,yes Brightfield Bank,2400,6,54,36,yes,yes Corvel Logistics,15000,5,45,22,yes,yes Dunmore Holdings,4800,3,12,4,yes,yes Eastlake Health,1200,3,30,14,yes,yes Fennick Holdings,4800,8,48,30,yes,yes Galloway Retail,9000,9,36,20,yes,no Hartwell Group,1200,4,48,21,yes,no Iverson Retail,1200,3,18,9,yes,no Juniper Retail,1200,3,24,10,yes,no Kestrel Logistics,4800,1,12,6,yes,no Linton Group,1200,2,10,4,no,no Marlowe Group,2400,2,10,7,no,no Northcote Retail,9000,2,10,6,yes,no Oakhurst Group,9000,3,36,18,yes,no Pemberton Health,1200,1,6,2,yes,no Quayside Bank,4800,2,12,4,no,no Redfern Group,15000,3,18,10,no,no Stanway Health,9000,3,30,12,no,no Thornbury Bank,15000,3,24,12,no,no Upton Bank,9000,3,33,23,yes,no Varden Bank,2400,3,12,5,no,no Whitmore Bank,4800,2,8,5,no,no Yardley Logistics,15000,2,8,4,yes,no

enterprise_win_loss_interviews.md5 lines · Download

# Enterprise win/loss interviews (12 buyers, this year)

- Won (7): every win had teams already using Flowboard. "Our designers refused to give it up." "IT approved it because 300 people were already on it."
- Lost (5): four had standardised on Miro company-wide; switching meant migrating thousands of boards. One stayed on whiteboard photos and slides.
- What buyers said Flowboard does that Miro doesn't: real-time diagramming that stays tidy, and a free plan teams can start on without IT.

What a strong answer doesThe answer key the graders mark against

Builds enterprise sales on top of the self-serve engine instead of throttling it. The deals file shows product-qualified deals (teams already using Flowboard) won 19 of 25 at an average of about $72,600 and about 63 days, against outbound winning 4 of 25 at about $36,800 and about 190 days; product-qualified wins were about $1.38M of the $1.86M won. The free plan feeds 70% of new paying teams. So capping the free plan at three users and removing self-serve checkout above ten seats would choke the pipeline the enterprise business depends on, and the interviews say IT approved Flowboard because teams were already on it. Proposes product-led sales: AEs work the large-company list (multiple teams, SSO requests, security-page visits) with SSO and admin controls as the bridge, while keeping self-serve open. Positions against Miro and the status quo, not 'visual collaboration for the enterprise': where a company has standardised on Miro, don't lead with a migration; win where teams already use Flowboard, on tidy real-time diagramming and starting without IT. Reads outbound's poor results as a message problem as well as a channel one. Works out capacity: $5M at about $72,000 is about 70 deals, far more than three AEs close (about 30 a year), so it needs AE hiring focused on product-qualified accounts rather than SDR outbound lists. Plans the ongoing motion and the measures (product-qualified accounts, conversion, cycle time) beyond the first quarter.

Critical failures (cap the score at 40)

Endorses capping the free plan at three users or removing self-serve checkout above ten seats

Task

Staff-level · v1.0

The outputs

8 setups answered this brief.

GPT-6 AstraChatGPT100Usable with a quick edit Opus 5.5Claude91Usable with a quick edit Sonnet 5.5API84Needs more work GPT-6.1 SolAPI82Needs more work GPT-6 LunaAPI80Needs more work Haiku 5.5API80Needs more work Gemini 3.8 FlashAPI71Needs more work Gemini 3.5 Flash-LiteGemini57Needs more work

Sonnet 5.5API

Needs more than a quick edit11 of 13 checks passed

The output contains a factual error ('Six of the ten accounts above 10 seats have shown no IT interest' is wrong) and an unsupported effort claim, so a PM would need to re-check the data before using it.

Output

Enterprise GTM: VP Product recommendation

Bottom line: Jordan's plan sends our new hires at the channel that works worst and puts friction on the channel that works best. I'd back the enterprise features and some added sales capacity, but build the motion around the teams already using Flowboard. The board should see $5M as a stretch and about $3.4M as the likely case.

What last year's deals show

Last year we worked 65 deals, won 29, and booked about $1.86M of new ARR.

Source	Deals	Win rate	Avg won ACV	Avg won cycle	Share of new ARR
Product-qualified (teams already using us)	25	76%	$72.6K	63 days	74%
Inbound demo	15	40%	$55.1K	86 days	18%
Outbound	25	16%	$36.8K	190 days	8%

Outbound is two-thirds of the effort and 8% of the revenue. Of its 21 losses, 17 went to Miro as the company standard.

Across all sources, Miro took 26 of our 36 losses. "No decision" took 10, including 4 of the 6 product-qualified losses. The interviews agree: every win had teams already on Flowboard, and four of the five losses were Miro-standardised companies where switching meant migrating thousands of boards.

Who we sell to

Primary customer: a company with Flowboard already in several teams, where Miro is not yet the standard and IT is starting to engage.

The usage data lets us tier the 24 large companies already on Flowboard: - Tier A (6 accounts): asked for SSO. All six also viewed the security page. Aldergate, Brightfield, Corvel, Dunmore, Eastlake and Fennick together have 27 teams and 116 paying seats. Sales owns these now. - Tier B (10 accounts): viewed the security page but haven't asked for SSO. A product-led nudge and a light sales touch. - Tier C (8 accounts): no IT signal yet. Keep nurturing in-product.

Seat count is the wrong trigger. Aldergate (10 seats) and Dunmore (4 seats) asked for SSO and would be missed by a ">10 seats" rule. Six of the ten accounts above 10 seats have shown no IT interest.

Two gaps in this data: - The file covers only companies over 1,000 employees, so it understates our real product-qualified pool. - We don't record which incumbent tool a prospect uses. We should add that field now, because "Miro is the standard" predicts the loss.

How self-serve and sales work together

Self-serve generates demand and sales converts it.

Free plan stays open. It feeds 70% of new paying teams, and buyers cite "a free plan teams can start on without IT" as something Miro doesn't offer.
Checkout stays open at every size. Sales is triggered by signals: an SSO request, security page views, three or more teams in one domain, or a domain crossing an active-user threshold. A rep reaches out while the team keeps buying normally.
Sales sells the consolidation: security review, one contract, and an admin console, backed by the customer's own usage data ("300 people already use it"). That also answers the "no decision" losses, which need a champion kit and a usage-based business case.
Inbound demos are routed to the same team. A 40% win rate is worth keeping, but half the losses are to Miro standards, so qualify for the incumbent tool early.

What we say against the alternatives

Miro: don't contest "the enterprise whiteboard." Position as the best tool for real-time diagramming that stays tidy, and one that teams adopt without waiting for IT. Don't ask anyone to migrate. Win the team first and let usage make the case. Only 2 of 25 product-qualified deals went to Miro, versus 17 of 25 outbound deals.
Status quo (photos and slides): show the cost of the current process using their own team's usage.
Other tools: the data doesn't show us losing to any, so I'm making no claims about them.

Jordan's line, "The visual collaboration platform for the enterprise," is Miro's category claim, and it leaves out our actual advantages. I'd use something like: "Diagramming your teams already love, ready for IT to approve."

Capacity

Today each AE works about 22 opportunities a year (65 across 3 AEs) and closes about 10.

To reach $5M: - 60 product-qualified wins at 76% means about 79 opportunities, or $4.4M. - 12 inbound wins at 40% means 30 demos, or $0.66M. - That's about 109 opportunities, or roughly 5 AEs at today's load. - Product-qualified cycles are 63 days against 190 for outbound, so new AEs ramp faster on this mix. - Hire 3 AEs for 6 total. That gives a buffer, a faster ramp, and no outbound headcount.

Other hires (not part of Jordan's plan): - 2 reps who work product-qualified accounts, in place of 6 outbound SDRs. - 1 solutions engineer for security reviews. - 1 RevOps/growth analyst to build account-level usage scoring.

That's 7 sales-side heads against Jordan's 14, plus the engineering squad below.

The real risk is volume, not headcount. $5M needs product-qualified opportunities to roughly triple, from 25 to about 79. At double (50 opportunities), we land near $3.4M. The 24-company list can't deliver that alone, so RevOps needs to size the full pool, including sub-1,000-employee companies, before we commit to the board.

Engineering sequence: 1. SSO and SCIM first. Six accounts have already asked, and IT approval is the gate in our win stories. 2. Audit logs next. 3. Data residency only when a named deal requires it. Nothing in the loss data points to features. We lose to incumbency and indecision, so features are a gate, not a growth driver.

What I'd change in Jordan's plan

Jordan's item	My view
1. Cap free at 3 users	Reject. It throttles the source of 70% of new paying teams and 74% of enterprise ARR, and removes the "start without IT" advantage. If we need a lever, gate admin and security features, not users.
2. Remove checkout above 10 seats	Reject. Ten of the 24 large accounts are above 10 seats, and they are the ones converting. Keep checkout and trigger sales on IT signals.
3. 8 AEs + 6 SDRs on Fortune 2000 outbound	Replace. Outbound wins 16% of the time, at a $36.8K ACV and 190-day cycle, mostly against Miro standards. Hire 3 AEs and 2 product-qualified reps instead.
4. SSO, SCIM, audit logs, data residency	Keep, resequenced. SSO/SCIM first, audit logs second, residency on demand.
5. Messaging	Replace. It fights Miro on Miro's ground. Lead with team adoption and tidy diagrams, then IT readiness.

Why Jordan's plan lands short in-year (my estimate): an outbound AE produces about $368K a year (10 deals × $36.8K). New hires starting in Q1 won't see a first outbound close until around Q3, so 8 AEs yield well under $1M this year. Adding that to last year's $1.86M gives roughly $2.6M, before counting damage from the free cap and checkout changes to our best channel.

What I'd ask the board for

Commit to about $3.4M, with $5M as the stretch if product-qualified volume triples by mid-year.
Approve the smaller hiring plan and the SSO/SCIM squad.
Review progress quarterly on product-qualified opportunity count, win rate, and the share of losses to Miro.

I'd like to walk Jordan through this before the board meeting. We agree on the target and on the need for enterprise features. We disagree on where the next dollar of sales effort should go.

Claims the judge couldn’t find in the brief

Six of the ten accounts above 10 seats have shown no IT interest.
Outbound is two-thirds of the effort and 8% of the revenue.
We don't record which incumbent tool a prospect uses.
The file covers only companies over 1,000 employees, so it understates our real product-qualified pool.

Grades and run details

Decision model 96 · LLM judge 79

Decision model checks

passUses the supplied evidence correctly18%
passAddresses the actual decision98%
passRespects explicit constraints35%
passIdentifies material uncertainty96%
partialAvoids unsupported claims34%
passProduces the required deliverable82%
passFollows the deals data100%
passProtects the self-serve engine100%
passDoes the capacity sums96%
passPositioned against the real alternative99%
passSells against a pain the buyer feels77%
passDiagnoses message or channel first99%
passPlans beyond launch week100%

Run

Run: #1
API response time: 87 s
Submitted: 6 Oct 2026

Opus 5.5Claude

Usable with a quick edit11 of 13 checks passed

The output is a complete, actionable plan; the one factual error (5 of 6 no-decision losses) is minor and can be corrected with a quick edit without redoing any reasoning.

Output

Flowboard Enterprise Go-to-Market: VP Product Version

To: CEO | cc: Jordan Hale, CRO

Summary

Our enterprise deals come from teams already using Flowboard. Last year, deals with existing teams closed 3x faster, at 2x the contract size, and at nearly 5x the win rate of outbound deals. Jordan's plan invests most heavily in outbound, our weakest motion, and caps the free plan that creates our strongest one. I recommend we build the enterprise motion on product usage instead.

On the evidence we have, that motion supports about $3.4M of new enterprise ARR, with a funded path to $5M if product-qualified pipeline grows as expected. We should give the board that range rather than commit to $5M.

What the data shows (65 enterprise deals, last 12 months)

Source	Deals	Win rate	Avg contract (won)	Days to close (won)	Main loss reason
Teams already using Flowboard	25	76% (19)	$72.6K	63	No decision (5 of 6)
Inbound demo request	15	40% (6)	$55.1K	86	Miro standard (7 of 9)
Outbound	25	16% (4)	$36.8K	190	Miro standard (17 of 21)

Where the revenue came from: existing-user deals produced $1.38M of the $1.86M won.
Where we lose: 26 of 36 losses were to companies already standardised on Miro, and outbound found most of them.
What buyers told us: all 7 interviewed winners already had teams on Flowboard. All 4 Miro losses came down to the cost of migrating thousands of boards. When buyers described what we do better, they named the free plan teams can start on without IT.

Who we sell to

Primary target: companies with 1,000+ employees where two or more teams already use Flowboard. - Champion: the team lead whose people "refused to give it up." - Approver: IT or security, who signs off once enough people are already using the product.

The current list of large companies using Flowboard (24 accounts) splits into three tiers:

Tier	Signal	Accounts	Action
1	Viewed security page and asked for SSO	Aldergate, Brightfield, Corvel, Dunmore, Eastlake, Fennick (6)	AE engaged within 2 weeks
2	Viewed security page, no SSO request	Galloway, Hartwell, Iverson, Juniper, Kestrel, Northcote, Oakhurst, Pemberton, Upton, Yardley (10)	Sales-assist outreach; in-product enterprise trial
3	Usage only	Linton, Marlowe, Quayside, Redfern, Stanway, Thornbury, Varden, Whitmore (8)	Product-led nurture until they cross a threshold

Who we don't target: companies standardised on Miro with no Flowboard usage. That describes most of the outbound pipeline, and we lost 17 of those 21 outbound deals.

Data gap to close before the board meeting: 24 accounts can't produce the roughly 50 product-qualified opportunities this plan needs. We need the full pool of qualifying accounts, including companies under 1,000 employees that look like past enterprise buyers. The capacity numbers below depend on that pool existing.

How self-serve and sales work together

1. Self-serve stays open. - No seat cap on checkout. - The free plan keeps unlimited users. - It feeds 70% of new paying teams and is the source of nearly every enterprise win.

2. Product defines a "product-qualified account" (PQA) and routes it to sales. - Trigger: a 1,000+ employee domain with any of these: - 2+ teams, - 25+ active users, - a security-page view, - an SSO request. - Product owns the scoring. Sales owns the response time. - Routing is by account signal, not by team size. A 15-seat team at a 60-person agency stays self-serve.

3. Enterprise features are the upgrade path. - Admin controls (SSO, SCIM, audit logs) appear in-product with a "talk to us / start a trial" path. - This gives IT a reason to engage without blocking the teams that are already working.

4. Fix the "no decision" problem. Five of our six losses on existing-user deals ended in no decision. The champion needs help getting IT to approve, not more prospecting. We will provide: - a security review pack, - an admin console demo, - a business case built from the account's own usage data.

5. No channel conflict. - AEs get credit for self-serve revenue that consolidates into their enterprise contract. - Self-serve keeps its own targets below the enterprise threshold.

What we say against the alternatives

Headline: "The diagramming tool your teams already chose — now ready for IT."

Against Miro: don't pitch a company-wide replacement. Buyers have told us migration cost decides those deals, and we lose.
Lead on what buyers said we do better: real-time diagramming that stays tidy.
Lead on proof of adoption: "300 people are already on it."
Where Miro is the standard, test positioning Flowboard as the diagramming tool alongside it. Treat this as a hypothesis and measure it on 5–10 deals before scaling.
Against whiteboard photos and slides (no decision): the free plan lets a team start this week without an IT project. IT controls come later, when the usage justifies them.
Against "the enterprise platform": that phrase is Miro's ground, and claiming it invites the head-to-head comparison we lose.

Capacity

Pipeline and revenue. These assume historical win rates and contract sizes hold.

Source	Opportunities	Wins	ARR
Existing users (2x last year, via PQA routing)	50	38	$2.76M
Inbound (better routing and messaging)	25	10	$0.55M
Targeted outbound (only accounts not standardised on Miro)	15	2–3	$0.10M
Total	90	~50	~$3.4M

People: - AEs: about 50 deals at roughly 10 per ramped AE needs 5–6 AE-years. - We have 3 AEs. Hire 4 more in Q1. - I'm assuming about a one-quarter ramp, which is plausible with 63-day cycles but needs validating with Jordan. - Sales engineer (1): owns security reviews and the no-decision problem. - Account manager (1): owns expansion across our 46 enterprise customers. - Product: a PQA scoring and routing squad, plus the enterprise feature roadmap. - Total: about 7 hires, versus 14 in Jordan's draft.

Path to $5M (the remaining ~$1.6M) has two options: 1. Stretch pipeline. At 76% win rate and $72.6K average, $5M needs about 90 existing-user opportunities. We should only fund this against evidence: - Checkpoint: if Q1 produces 15+ product-qualified opportunities, hire 3 more AEs in Q2. 2. Expansion. If the board counts expansion of existing enterprise customers as new ARR, those 46 accounts (~$2.9M base) are a real source. The board needs to settle that definition.

What I'd change in Jordan's plan, and why

1. Cap the free plan at 3 users. → Drop it. - The free plan feeds 70% of new paying teams and seeds almost every enterprise win. - Buyers named it as a reason they chose us over Miro. - A cap forces conversations with teams that haven't adopted yet, and that kind of deal is where we lose most often.

2. Route every team over 10 seats to sales and remove checkout. → Route by account signal; keep checkout. - With 31,000 paying teams, a seat-count rule would bury AEs in small-company teams that will never be $64K deals. - It would also add friction to the self-serve revenue we already have.

3. Hire 8 AEs and 6 SDRs on Fortune 2000 outbound. → Hire 4 AEs, 1 sales engineer, and 1 account manager; no outbound SDR team. - Outbound wins 16% of the time at $36.8K over 190 days. - Reaching $5M through outbound would take about 850 opportunities. - An AE hired in Q1 on a 190-day cycle closes little or nothing in-year, so most of that headcount pays off in the following year. - If we want SDRs at all, use 1–2 to work product-qualified accounts, not cold lists.

4. Enterprise features (SSO, SCIM, audit logs, data residency). → Agree, with sequencing. - Ship SSO, SCIM, and audit logs first. Six Tier 1 accounts are asking for SSO now. - Build data residency against named deals. The bank and health accounts on our list are the likely candidates.

5. Messaging: "The visual collaboration platform for the enterprise." → Change it. - It positions us head-on against Miro on Miro's strongest ground. - Lead with the diagramming quality and bottom-up adoption that buyers say actually win the deal.

Where Jordan and I agree: we need more sales capacity and real enterprise features. We differ on where the pipeline comes from.

I'd like to walk Jordan through these numbers before the board meeting so we present one plan. That plan would be built on how our customers actually buy.

Claims the judge couldn’t find in the brief

Five of our six losses on existing-user deals ended in no decision.

Grades and run details

Decision model 96 · LLM judge 86

Decision model checks

passUses the supplied evidence correctly19%
passAddresses the actual decision80%
passRespects explicit constraints40%
passIdentifies material uncertainty96%
partialAvoids unsupported claims35%
passProduces the required deliverable45%
passFollows the deals data100%
passProtects the self-serve engine100%
passDoes the capacity sums92%
passPositioned against the real alternative99%
passSells against a pain the buyer feels79%
passDiagnoses message or channel first98%
passPlans beyond launch week100%

Run

Run: #1
Time to output: 1.5 min
Submitted: 5 Oct 2026

Haiku 5.5API

Needs more than a quick edit11 of 13 checks passed

The output contains a factual error in the median cycle and an unsupported claim about data residency, so a product manager would need to re‑verify the data before using it.

Output

To: CEO From: VP Product Re: Enterprise plan for the board

Bottom line: Jordan is right that enterprise revenue should run through sales. But the draft caps the motion that produces most of our enterprise deals, and it spends hiring budget on the motion that loses most often. I'd keep the free plan open, route accounts on buying signals, hire against proven pipeline, and present $3.0M committed with $5M as a stretch tied to specific gates.

What the last 12 months show (65 enterprise opportunities, 29 won, $1.86M ACV)

Source	Opps	Won	Win rate	Avg won ACV	Notes
Product-qualified (teams already on Flowboard)	25	19	76%	$72.6K	74% of won ACV; median cycle 66 days
Inbound demo request	15	6	40%	$55K
Outbound	25	4	16%	$37K	Wins took ~190 days; 17 of 21 losses went to Miro

Across all 36 losses, 26 went to Miro as company standard and 10 to no decision. All seven enterprise wins in the interviews had teams already using Flowboard.

Who we sell to

Our best enterprise buyers are large companies where we already have several teams. Of the 24 companies over 1,000 employees using us, 16 have looked at our security page in the last 30 days. Six have also asked for SSO: Aldergate, Brightfield, Corvel, Dunmore, Eastlake and Fennick. Together they hold 116 paying seats. That's the first account list, and it's warm.

I would not start with Fortune 2000 cold outbound. That motion wins 16% of the time, and the companies on those lists are the ones most likely to have standardised on Miro.

How self-serve and sales work together

The free plan stays open. A 3-user cap breaks the motion that generated 19 of our 29 wins. The enterprise deals came from teams that grew to hundreds of people on free, with designers who "refused to give it up." A cap doesn't make those teams call us. It sends them to Miro, or it keeps them on the free plan without ever reaching the buyer.
Routing should follow signals, not one seat number. Route an account to sales when it has three or more teams in one company, asks for SSO or SCIM, views the security page, or passes about 25 paying seats. Self-serve checkout stays available below that line. Removing checkout at 10 seats would push paying mid-size teams into a sales cycle they don't need, and we'd lose that ARR.
Division of labour. Sales owns the contract, the security review and the SSO rollout. Product owns in-app signals, admin controls and the usage data that tells sales where to go. AEs work the warm account list, not cold lists.

What we say against the alternatives

Against Miro: Don't compete on feature parity. Buyers chose us for real-time diagrams that stay tidy and a free start that doesn't need IT. "The visual collaboration platform for the enterprise" puts us in Miro's category and invites a feature comparison we lose. I'd use: "The whiteboard your teams already chose, now with the controls IT needs."
Where Miro is already company standard, we lose most of the time. Don't spend outbound effort there. Instead, scope a Miro board import. Buyers cited migrating thousands of boards as the reason they stayed. If we can make the migration a file import, we can contest some of those standardisation decisions. Engineering needs to size this.
Against no decision (10 losses): My hypothesis is that these stall in IT or security review. SSO, SCIM and audit logs address exactly that. Sales should check lost-deal notes to confirm.

Capacity

Three AEs closed 29 deals, about ten each, or roughly $620K per AE. New AEs take around six months to ramp.
The binding constraint is pipeline, not AE headcount. At the product-qualified average of $72.6K, $5M is about 69 wins. At a 76% win rate, that needs roughly 90 product-qualified opportunities a year. We have 25.
Hiring eight AEs before that pipeline exists leaves most of them waiting. Six SDRs working outbound would be pointed at our lowest-converting motion.
My estimate: three current AEs plus four hires in Q1 (about half a year of productivity each) gives roughly five effective AEs, or about $3.1M. The large-account program could add $0.5–1M. That supports a $3.0–3.5M base. The $5M case requires roughly tripling product-qualified pipeline and having the new hires productive by Q3. Nothing in the data shows that yet.

What I'd change in Jordan's plan, and why

Keep the free plan unlimited. It produces the enterprise pipeline, and a cap would reduce that pipeline.
Replace the 10-seat rule with signal-based routing (see above) and keep checkout below about 25 seats. This sends sales the accounts most likely to buy without cutting paying self-serve revenue.
Hire four AEs now and two more on a gate: 40 or more qualified product-qualified opportunities in a quarter, and a trailing win rate above 60%. If the gate isn't met, we don't carry the cost.
Cancel the six outbound SDR roles. Put two SDRs on inbound demo follow-up (40% win rate, and most inbound losses were to Miro or no decision, so speed-to-follow-up matters). Put two on expansion of the 24 large accounts. Hold two until the gate is met.
Build SSO, SCIM and audit logs first. Six accounts have asked for SSO, and 16 of 24 large accounts have looked at the security page. Defer data residency. No deal in the last 12 months cited it, so build it only when a named deal requires it.
Change the messaging as above and drop "for the enterprise."
Scope the Miro board import as a product workstream, with a decision in 30 days.

What I'd ask the board for

A committed number of $3.0M enterprise ARR, with $5M as a stretch.
Gates the board can track quarterly: qualified product-qualified opportunities, win rate, the six SSO accounts closed, and no-decision losses as a share of the total.
Approval to hire four AEs now and two on the gate.

I'd rather present a number we can defend than one that depends on hiring eight people before we have the pipeline to keep them busy.

Claims the judge couldn’t find in the brief

Median sales cycle for product-qualified deals is 66 days.
No deal in the last 12 months cited data residency as a requirement.
New AEs take around six months to ramp.

Grades and run details

Decision model 88 · LLM judge 79

Decision model checks

failUses the supplied evidence correctly13%
passAddresses the actual decision97%
passRespects explicit constraints52%
passIdentifies material uncertainty95%
partialAvoids unsupported claims34%
passProduces the required deliverable91%
passFollows the deals data98%
passProtects the self-serve engine100%
passDoes the capacity sums97%
passPositioned against the real alternative97%
passSells against a pain the buyer feels86%
passDiagnoses message or channel first98%
passPlans beyond launch week100%

Run

Run: #1
API response time: 54 s
Submitted: 8 Oct 2026

Check by check

Where the outputs differ comes first. Open a check to read the judge’s reasons. “Mixed” means the two graders disagreed.

Where they differ 1

Uses the supplied evidence correctlyMixedMixedWrong

Sonnet 5.5 · API

Several claims about the current situation are unsupported or incorrect: 'Six of the ten accounts above 10 seats have shown no IT interest' is false (only 2), 'Outbound is two-thirds of the effort' has no basis, and 'We don't record which incumbent tool a prospect uses' is not in the supplied context.

Opus 5.5 · Claude

The claim 'Five of our six losses on existing-user deals ended in no decision' is not supported by the data (4 no-decision, 2 Miro), so a statement about the current situation is invented.

Haiku 5.5 · API

The median cycle for product‑qualified deals is stated as 66 days but the data gives 62, and the claim that no deal cited data residency is unsupported.

All got wrong 1

Avoids unsupported claimsWrongWrongWrong

Sonnet 5.5 · API

Presents 'Outbound is two-thirds of the effort' and 'Six of the ten accounts above 10 seats have shown no IT interest' as facts when the first is unsupported and the second is incorrect.

Opus 5.5 · Claude

Presents the incorrect 'five of six no decision' figure as established fact, which is an unsupported claim about the current situation.

Haiku 5.5 · API

It presents the incorrect median cycle and the ‘no deal cited data residency’ statement as fact without support.

All got right 11

Addresses the actual decisionRightRightRight

Sonnet 5.5 · API

Commits early to a clear recommendation (reject free cap and checkout removal, build on product-qualified, target $3.4M with $5M stretch) and states what would change it (product-qualified volume tripling).

Opus 5.5 · Claude

Commits to building the enterprise motion on product usage, gives a clear $3.4M target with a path to $5M, and specifies the checkpoint (15+ PQAs in Q1) that would change the call.

Haiku 5.5 · API

The output commits to a clear plan ($3.0M committed, $5M stretch with gates) and states what would change it.

Respects explicit constraintsRightRightRight

Sonnet 5.5 · API

The output is a memo within ~1,200 words, addresses all requested topics, and does not endorse capping the free plan or removing self-serve checkout.

Opus 5.5 · Claude

The output is a memo to the CEO, covers all required topics, stays within the word limit, and does not propose capping the free plan or removing self-serve checkout.

Haiku 5.5 · API

The memo is within 1,300 words, addresses all requested topics, and is addressed to the CEO.

Identifies material uncertaintyRightRightRight

Sonnet 5.5 · API

Names the key unknown (size of the full product-qualified pool), bounds the risk ($3.4M likely vs $5M stretch), and says RevOps must size the pool before committing.

Opus 5.5 · Claude

Identifies the unknown pool of qualifying accounts, the dependency on that pool for capacity, and sets a Q1 checkpoint to resolve it; also flags the board's definition of new ARR.

Haiku 5.5 · API

It identifies unknowns (e.g., why no‑decision losses stall, Miro import feasibility) and says how to resolve them.

Produces the required deliverableRightRightRight

Sonnet 5.5 · API

Delivers a complete, actionable memo to the CEO/board within the word limit, covering all requested elements.

Opus 5.5 · Claude

The memo is complete, addresses all requested elements, is within the word limit, and a CEO could act on it with light edits.

Haiku 5.5 · API

It delivers a complete, actionable memo in the requested form, suitable for the CEO with minimal edits.

Follows the deals dataRightRightRight

Sonnet 5.5 · API

Computes win rate, deal size, and cycle time by source and builds the motion around product-qualified accounts.

Opus 5.5 · Claude

Computes win rate, deal size, and cycle time by source and builds the plan around product-qualified accounts.

Haiku 5.5 · API

The plan computes win rate, size and cycle by source and builds the motion on product‑qualified accounts.

Protects the self-serve engineRightRightRight

Sonnet 5.5 · API

Uses the 70% free-plan conversion and product-qualified deal data to argue against capping the free plan and removing checkout, and keeps self-serve open.

Opus 5.5 · Claude

Uses the 70% free-plan conversion and product-qualified deal data to argue against capping the free plan and removing self-serve checkout, and keeps self-serve open.

Haiku 5.5 · API

It explains why capping the free plan and removing checkout would hurt the pipeline and keeps self‑serve open.

Does the capacity sumsRightRightRight

Sonnet 5.5 · API

Works out deals needed (60 product-qualified, 12 inbound), total opportunities (109), and AEs required (5, hire 3 to reach 6), with checkable arithmetic.

Opus 5.5 · Claude

Works out the number of deals needed (~70 for $5M), the AE capacity required (5-6 AE-years), and provides a pipeline table with checkable sums.

Haiku 5.5 · API

The output works out the deals needed (~69), pipeline required (~90 PQOs), and AE capacity to reach $5M.

Positioned against the real alternativeRightRightRight

Sonnet 5.5 · API

Positions against Miro (don't contest the enterprise whiteboard; win the team first) and the status quo (photos and slides), and picks the segment where Miro is not the standard.

Opus 5.5 · Claude

Names Miro (company standard) and whiteboard photos/slides as the real alternatives, and ties the message to diagramming quality and bottom-up adoption that those alternatives lack.

Haiku 5.5 · API

It names Miro and ‘no decision’ as the real alternatives and ties the message to design adoption and IT controls.

Sells against a pain the buyer feelsRightRightRight

Sonnet 5.5 · API

Leads with the pain of IT approval friction and the risk of losing a tool teams already love, backed by interview evidence that IT approved because people were already using it.

Opus 5.5 · Claude

Leads with the pain of IT needing to approve a tool already in use and the risk of losing it, supported by buyer interviews where IT approved because teams were already on it.

Haiku 5.5 · API

The message is built on buyer pain (IT losing control, design team refusal) evidenced by the interviews.

Diagnoses message or channel firstRightRightRight

Sonnet 5.5 · API

Diagnoses that outbound has poor conversion (16% win rate) and long cycles, and shifts effort to the product-qualified channel while also changing the message against Miro.

Opus 5.5 · Claude

Diagnoses that outbound targets Miro-standard accounts where the message can't overcome migration cost, and shifts investment to product-qualified accounts and a different message.

Haiku 5.5 · API

It diagnoses that the outbound channel is underperforming (low win rate, high loss to Miro) and shifts effort to warm product‑qualified accounts.

Plans beyond launch weekRightRightRight

Sonnet 5.5 · API

Describes the ongoing motion (product-led nudges, sales outreach on signals, RevOps scoring) and measures (opportunity count, win rate, share of losses to Miro) beyond launch.

Opus 5.5 · Claude

Describes the ongoing PQA scoring and routing, the 'no decision' fix with security packs and admin demos, and a Q1 checkpoint to decide further AE hiring.

Haiku 5.5 · API

The plan includes ongoing routing, quarterly gates, and measures (PQOs, win rate, SSO closures) beyond launch.

Results

Every setup we’ve tested on this task type, across all its tasks and repeats, graded on the current checklist. Calibrated.

#	Model · Harness	Decision model	LLM judge	Runs	Critical failures
1	GPT-6 AstrawithChatGPT	100.0	100.0	2	None
2	Opus 5.5withClaude	96.0	85.2	2	None
3	GPT-6 LunawithAPI	94.2	89.3	2	None
4	GPT-6.1 SolwithAPI	93.8	78.0	2	None
5	Sonnet 5.5withAPI	98.1	77.7	2	None
6	Haiku 5.5withAPI	94.2	77.7	2	None
7	Gemini 3.8 FlashwithAPI	73.9	81.9	2	None
8	Gemini 3.5 Flash-LitewithGemini	74.2	63.5	2	None

What you get for the cost

No setup scores within 5 points of the leader at half its cost or less here, so the leader is also the best value.

Nobody beats these on both score and cost

Task score and relative cost by setup
Setup	Task score	Relative cost (cheapest = 1)
GPT-6 Luna · API	90	1.0×
Haiku 5.5 · API	82	3.3×
Gemini 3.5 Flash-Lite · Gemini	68	3.5×
Gemini 3.8 Flash · API	77	11×
GPT-6.1 Sol · API	84	16×
Sonnet 5.5 · API	84	46×
Opus 5.5 · Claude	91	75×
GPT-6 Astra · ChatGPT	100	81×

Cost is each output’s tokens at the model’s list price, shown relative to the cheapest setup. Runs done in an app have no usage meter, so theirs are estimated from the same setup’s measured runs. How we compare cost

The PM job

Taking a product or add-on to market with marketing and sales.

Why it matters

Launch plans fail by trying to reach everyone, by selling upside nobody feels, and by treating launch day as the whole plan.

What good looks like

One primary segment: those who value the difference most
Positions against what customers do today
Sells against a pain the buyer feels, with evidence
Diagnoses whether the problem is the message or the channel
Plans beyond launch week

Deliberately not measured

Creative copywriting quality

Capability tested

Go-to-market judgement

The failure we’re looking for

Every channel, every segment, and nothing after launch day

Grading

Decision model and LLM judge, calibrated against a blind PM review

What AI gets right here, and what you’ll still have to catch

Reliably right

Where it slips

How it’s graded

Positioned against the real alternative

Sells against a pain the buyer feels

Diagnoses message or channel first

Plans beyond launch week

Plus our standard checks

The tasks

The outputs

Output

Enterprise GTM: VP Product recommendation

What last year's deals show

Who we sell to

How self-serve and sales work together

What we say against the alternatives

Capacity

What I'd change in Jordan's plan

What I'd ask the board for

Claims the judge couldn’t find in the brief

Grades and run details

Decision model checks

Run

Output

Flowboard Enterprise Go-to-Market: VP Product Version

Summary

What the data shows (65 enterprise deals, last 12 months)

Who we sell to

How self-serve and sales work together

What we say against the alternatives

Capacity

What I'd change in Jordan's plan, and why

Claims the judge couldn’t find in the brief

Grades and run details

Decision model checks

Run

Output

Claims the judge couldn’t find in the brief

Grades and run details

Decision model checks

Run

Check by check

Where they differ 1

All got wrong 1

All got right 11

Results

What you get for the cost

About the task

The PM job

Why it matters

What good looks like

Deliberately not measured