Last week we scored the 2008 Airbnb seed deck blind, and the test was resisting the flattering hindsight bias: not over-scoring a winner because you know how it ends. This week is the harder direction. A deck the market funded, enthusiastically, at a $5 billion valuation, weeks after the documents we scored were circulated. If a rubric only ever agrees with the checks that got written, it is a mirror, not an instrument.
The documents: WeWork’s October 2014 Series D fundraising set. Thirty-eight slides plus a five-year forecast and company overview, seeking roughly $335 million. We scored them exactly as they read in October 2014, ignoring everything since.
I personally was following WeWork very closely because I was about to launch StartMart, a 35,000 sq. ft. startup accelerator in Cleveland’s landmark Terminal Tower. We were technically competitors, albeit in vastly different markets. My own take at the time was “If you can make it in New York, you can’t make it anywhere else.” I launched StartMart knowing it would break even at best, while hoping it would help our startup accelerator companies. Our five-year lease was based entirely on profit sharing.
The read, before the ending
Verdict: Borderline, capped by a red-flag gate. The weighted score alone came out 3.68 out of 5, a Lean yes. One gate, “Numbers don’t reconcile,” capped it. Predicted decision: pass.
Understand what the gate was up against. The traction in this document is the best we have scored in this series: $74.6 million in 2014 revenue, $4.2 million in operating profit, 16,279 members, growing 122% in nine months, and occupancy just under 100% across 23 cities. Traction is the heaviest dimension in our rubric at this stage, 28%, and it earned a perfect 5.
| Dimension | Weight | Score | The one fix that moves it most |
|---|---|---|---|
| Team & founder-market fit | 16% | 4 | 23 cities in 4 years is demonstrated operating. Add the gap plan: who runs real-estate risk at 376 locations? |
| Traction & evidence of demand | 28% | 5 | Reconcile the “40,000+ members” marketing figure with the forecast’s 16,279. Two member counts, one document. |
| Problem & urgency | 12% | 4 | Substantiate the 25% savings claim; per square foot this is premium-priced, so sell flexibility, not cheapness. |
| Market & timing | 15% | 3 | The TAM counts every company under 100 employees. Build it bottom-up from desks, cities, and market rents. |
| Solution & differentiation | 10% | 3 | The 44% vs 16% margin comparison to Regus rests on non-comparable accounting. Restate it like-for-like. |
| Business model & unit economics | 13% | 2 | Show GAAP unit economics by location vintage, including build-out. This is the load-bearing weakness. |
| Competition & defensibility | 3% | 2 | Only Regus is mapped, dismissively. Name the moat and evidence it as retention. |
| The ask & use of funds | 3% | 3 | State what the round buys and to which proof point. The expansion plan implies it; nothing states it. |
So why does a 3.68 with monster traction predict a pass? Because the gate fired on what the document does with its own numbers, all on its face in 2014. Unit profitability is shown only “at maturity,” in fine print, while nearly half the locations opened that same year. Rent-free concessions are booked up front rather than spread across the lease, inflating income precisely during the forecast window. WeLive, a product that had not launched, accounts for 21% of forecast 2018 revenue, and its own savings claim contradicts its per-bed pricing. And the downturn case assumes 85% occupancy, which was the incumbent’s good-year number, while margins are projected to rise through a 15x expansion. A great business is visible in these documents. The documents inflate it anyway.
WeLive, to my mind, was also an “Only in New York” proposition. Housing costs in most other cities were not remotely comparable to those in Manhattan.
Now the tape
The round closed in December 2014 at roughly a $5 billion valuation, with Goldman Sachs, T. Rowe Price, Wellington, JPMorgan, and Harvard’s endowment participating. So our predicted pass was a miss, and we log it as one.
Then the rest of the tape. The valuation peaked near $47 billion in early 2019. The IPO was withdrawn that September after the S-1 brought the accounting into the light, including the invented “community-adjusted EBITDA” metric. Chapter 11 followed in November 2023. What the gate flagged in 2014 terms (at-maturity unit economics, front-loaded concessions, a fantasy downturn case, and a duration mismatch between 15-year lease obligations and monthly memberships) is a fair description of what actually played out.
Our own StartMart and Flashstarts Accelerator was closed in December 2019 at the end of our five-year lease, right before Covid would have closed it for us.
The miss tells you about the market, not the model. Late-stage investors in a hot 2014 funded a document our gate rejected, and the objection was eventually vindicated at a cost of about $47 billion in marked-down valuation. The calibration lesson we took: when a gate fires against strong traction, log the disagreement. Do not tune the gate away because the round closed.
What this has to do with your plan
Alex, the managed agent we build at Coworkers.Global, reads business plans the way we just read this one. The WeWork case is why the red-flag gates exist and why traction cannot buy them back. Impressive numbers that do not reconcile are a worse signal than modest numbers that do, because the inflation is a choice. Alex reads your plan for exactly that: where the numbers hold, where a screener would stop trusting them, and the single weakest part to fix first.
If you are applying to YC this cycle, the on-time deadline is Monday, July 27, at 8pm Pacific. The full read is free until then, and $295 after. Drop your plan at coworkers.global/ai-business-plan-review. We are early and pre-revenue, so we lead with the quality of the read rather than a customer list we do not yet have. If you think the read is wrong, tell us.
Current view, subject to change
Two data points do not prove a rubric, and a funded-then-failed case is the easiest kind to over-claim from. We considered whether the harsher gate, material dishonesty, should have fired here and decided no: the at-maturity treatment was disclosed in fine print, and non-GAAP presentation was standard private-market practice at the time. Borderline, but disclosed. What would change how I read these documents is a GAAP restatement of the 2014 location base showing the mature-location margins held up with build-out included. Nothing I can find suggests it would have. Next week: LinkedIn, 2004, back before anyone knew what a social network was worth.
Regards,
Charles Stack