Quick-Commerce Graveyard: What Gorillas, JOKR, Getir and Flink Show

Quick-commerce, the ten-minute grocery delivery model built on dark stores and venture money, produced one of the fastest sector-wide shakeouts in recent memory. WillItSurvive's assessment of 10,525 companies assessed in July 2026 contains multiple quick-commerce entries in its graveyard set, and the sector plays an unusual role in our blind backtest: it is the only category where companies we selected as healthy survivors kept turning out to be casualties themselves.

This article documents what the data actually shows, using only pre-collapse facts and dated assessments. We never declare an outcome for any operating company; forward-looking material appears only as dated watch-items.

The quick-commerce entries in the backtest

Our blind backtest works like this: dead companies are judged on facts knowable 18 months before their terminal event, with identity masked, by production judges under a time-boxed context. Each dead company is paired with a matched surviving peer judged at the same cutoff date with the same masking, and the survivor arm runs two independent replicates.

The matched survivor arm behaved very differently: only 17% of the 58 survivor twins in replicate one drew negative verdicts, and the two survivor replicates agreed on 49 of 53 shared records (92%), per the July 2026 readout.

Four quick-commerce pairs appear in the blind-pairs file:

Gorillas (dead) paired with Flink (survivor twin), cutoff June 2021. Candidate twins considered included Getir, Flink, Zapp and Jokr.
JOKR (dead) paired with Getir (survivor twin), cutoff June 2022. Candidates considered: Getir, Gorillas, Flink, Zapp.
Getir US operations (dead) paired with Instacart (survivor twin), cutoff October 2022. Candidates considered: Gopuff, Instacart, DoorDash, Gorillas.
Buyk (dead) paired with Gopuff (survivor twin), cutoff September 2020. Candidates considered: Gopuff, Getir, Gorillas, Fridge No More.

Notice the problem embedded in that list. When we went looking for healthy peers of dead quick-commerce companies, the candidate pools kept recycling the same names: Getir, Gorillas, Flink, Zapp, Jokr. The sector was so uniformly stressed that the graveyard and the control group overlapped.

The twins that turned: Getir and Flink

The most telling result in the quick-commerce data is not the dead-arm verdicts. It is what happened on the survivor arm.

Getir, judged blind as JOKR's survivor twin at a June 2022 cutoff, received a not called safe verdict in both independent replicates of the July 2026 backtest. Flink, judged blind as Gorillas' survivor twin at a June 2021 cutoff, likewise received not called safe in both replicates of the July 2026 backtest. Both were recognized by the identifiability probe, so we do not claim recognition-independence for these two specific records. But both verdicts were consistent across replicates, and both companies had been seeded into the study as the healthy half of their pair.

Getir's story goes further. The backtest's seeding-error audit found that 10 companies selected as healthy survivor twins and flagged negative by the blind judges, including Getir, Chegg, Fisker and Nikola, each turned out to be genuinely failed or gutted companies mis-selected as healthy peers, per the July 2026 readout. In other words, the blind assessment caught our own selection mistakes, and quick-commerce contributed to that list. The blind judges identified all 10, in both replicates.

The supported claim, stated the way we publish it: given only pre-collapse facts, the assessment separated the doomed from matched survivors and never once said safe-to-join on a company that was dead within 18 months. We do not phrase this as having predicted anything; the backtest is retrospective by design.

Why the sector reads this way in the data

The assessment records themselves do not carry a narrative field, but the pair structure tells the story. Buyk's cutoff is September 2020, Gorillas' is June 2021, JOKR's and the Getir-as-twin record's are June 2022, Getir US's is October 2022. The graveyard entries cluster in a roughly two-year window, and each dead company's proposed peer set was drawn almost entirely from other companies that themselves ended up in the graveyard or on the flagged-twin list. Gopuff and Instacart are the only quick-commerce-adjacent names in these pairs that were not flagged negative in the readout.

That clustering is what a sector-wide failure looks like inside a durability dataset: the category shrinks until the only defensible "survivor" comparisons come from adjacent business models (Instacart's marketplace model rather than dark-store ownership) rather than from direct clones.

For context on how unusual this is: the dead-arm zero-safe-calls result held even on the unrecognized subset, where the 17 masked dead companies the identifiability probe could not name were 88% negative with zero safe calls, per the July 2026 backtest readout. Quick-commerce entries were consistent with that overall pattern.

Watch-items, dated

Because we publish no outcome declarations, here is what a reader tracking the remaining quick-commerce and adjacent players should verify independently, as of July 2026:

Getir: flagged negative in both blind replicates at a June 2022 cutoff and listed in the seeding-error audit as mis-selected as a healthy peer. Verify current corporate status and market footprint directly.
Flink: flagged negative in both blind replicates at a June 2021 cutoff. Verify funding status and market coverage since that date.
Gopuff and Instacart: seeded as survivor twins (September 2020 and October 2022 cutoffs respectively) and not among the flagged records in the July 2026 readout. Check current unit economics disclosures yourself.

Questions this data answers

Q: Which quick-commerce companies appear in the WillItSurvive graveyard set? Four quick-commerce entries appear in the blind backtest's dead arm as of July 2026: Gorillas (cutoff June 2021), JOKR (June 2022), Buyk (September 2020), and Getir's US operations (October 2022). Each was paired with a matched surviving peer judged blind at the same cutoff.

Q: Did the blind assessment ever call a soon-dead quick-commerce company safe? No. The quick-commerce entries followed that pattern.

Q: Why were Getir and Flink judged negatively if they were selected as survivors? Both were seeded into the backtest as the healthy halves of their pairs, and both drew not called safe verdicts in both independent replicates. The subsequent audit found Getir was one of 10 twins mis-selected as healthy peers that turned out to be genuinely failed or gutted, per the July 2026 readout.

Q: Does WillItSurvive predict which delivery companies fold next? No. We publish dated assessments and watch-items only, and we never declare an outcome for an operating company. The backtest is retrospective: it shows how blind judgments on pre-collapse facts sorted dead companies from matched survivors, nothing more.

Q: How reliable are the survivor-arm verdicts? The two independent survivor replicates in the July 2026 backtest agreed on 49 of 53 shared records, a 92% agreement rate. The survivor arm drew 17% negative and 24% positive verdicts in replicate one, versus 90% negative and 0% positive on the dead arm.

Assessments reflect evidence available July 2026. Companies change; treat watch-items as things to verify, not conclusions.