# The same request rules in Zapier and Make

Tested 25 September 2026 using Zapier’s Pro and Make Core. AI operated both products. Neither account was upgraded for this test.

Both products handled all 40 submissions correctly across two valid rounds. Each round had 20 inputs. We sent them one at a time through native webhooks into separate Google Sheets tabs. The saved outputs accumulated between rounds.

| Case | Inputs per product | Required outcome | Zapier correct | Make correct |
| --- | ---: | --- | ---: | ---: |
| Standard request | 24 | Request saved | 24 | 24 |
| Priority request | 8 | Request saved and added to priority queue | 8 | 8 |
| Empty email | 4 | Rejected without a request row | 4 | 4 |
| Repeated business request | 4 | Duplicate recorded without another request row | 4 | 4 |

Each product produced 32 request rows, 40 outcome rows and eight priority queue rows. No real messages were sent. The queue contained synthetic records.

Zapier used 42 tasks in each valid round: 84 total. Make used 78 credits in each valid round: 156 total. These counts include lookups, outcome records, rejected inputs and duplicates. Setup usage is excluded. We reconciled native run histories with account usage meters.

The measured rate is 2.1 Zapier tasks or 3.9 Make credits per submission. Scaling the same mix to 1,000 submissions gives 2,100 tasks or 3,900 credits. That is arithmetic, not a measured high-volume run. Tasks and credits are different units; neither number establishes a dollar winner.

## Excluded round

We excluded the third pair. One Zapier input’s first matching observation arrived after the 180-second cutoff. The observer incorrectly continued. Make’s third round stopped after six inputs when the observation connector failed. These are measurement failures; they do not establish product failures or comparative speed. Original records remain retained.

## What remains unknown

Saved Sheet rows prove the surviving results at observation time. They do not prove that no earlier write was repeated or overwritten. Native histories were reconciled by round; each run ID was not directly joined to its input ID.

The frozen manifest also retained a contradictory permission flag. We documented that error after scoring began and preserved the original file. Scoring rules did not change.

This test does not measure beginner ease, speed, scale, concurrency, recovery or team handover. No original video was captured during these scored runs. Any later walkthrough must identify itself as a later recording of retained results.

## Per-input observations

[Download the 80 accepted observations](zapier-make-native-request-observations-2026-09-25.csv). This export contains one row per submitted input per product. It omits account details and does not report observation time as product speed.

Source audit SHA-256: `7e6ec5b95683da2fa430fa481b68a45cb5af8d0a9519e7db8ca01a8f194e90f7`. The private source audit also retains the excluded third round.