A few days ago, someone in the Make Community ran into a problem while building an automation. The task was to check the email address entered by a user against a Google Sheet when they submitted an application. If no match was found, the email had to be added to the sheet. A task as simple as it is common, right?
But there was a catch ![]()
The workflow returned zero matches from the sheet, yet the route for adding a new member did not run.
Since I first came across the problem in the Make Community, I started looking for a solution in Make as well. I built a simple workflow, created a test Google Sheet, and started running the scenario with different inputs.
What happened in Make
In the original workflow, the Router came before Search Rows. This meant it had to choose a route before it was known whether the email had been found. The search result only became available inside one branch and could not control the other.
So Search Rows had to be moved before the Router:
Webhook
│
▼
Search Rows
│
▼
Router
The key detail was how Search Rows behaved when no matches were found. Although it found no rows, Make still provided a technical output containing:
Total number of bundles: 0
This meant the Router could check that number directly: a positive value sent the request down the existing-member route, while the fallback route added the new email.
Running the same test in n8n
Once the solution was working in Make, I became curious about how the same scenario would behave in n8n. I used the same Google Sheet, the same input data, and the same logic: an existing email, a new email, and then the same new email again.
Webhook
│
▼
Get row(s) in sheet
│
▼
If: row_number exists?
├── true → Existing email
└── false → Append row in sheet
The difference became apparent when the lookup found no matches. By default, Get row(s) returned zero items, so the following If node did not run at all: it simply had nothing to process.
To keep the workflow running, I had to enable Always Output Data in the Get row(s) settings.
When no matches were found, the node then passed one empty item downstream, allowing If to route it to the branch that added a new row.
The next If node checked whether row_number existed. A real row returned by Google Sheets contained a row number and followed the true branch. The artificial empty item did not have this field, so it followed the false branch and triggered the addition of a new row.

However, the empty item no longer contained the original firstName and email. So in Append row in sheet, I referenced those values directly from the Webhook rather than taking them from the search result.
firstName ← {{String($('Webhook').first().json.body.firstName).trim()}}
email ← {{String($('Webhook').first().json.body.email).trim().toLowerCase()}}
The hidden whitespace trap
The workflow appeared to be ready, but submitting the same new email again still followed the false branch and created a duplicate.
The output of Append row revealed hidden line breaks: \n after firstName and \n\n after email. They were outside {{ ... }}, so .trim() inside the expression did not remove them. Google Sheets treated the clean email and the email with this hidden suffix as different values.
I replaced the lookup and append fields with single-line expressions containing no line breaks. After that, the repeated email was found correctly and no additional row was added.
Test results
After fixing the fields, I ran three clean tests:
- Existing email →
truebranch, no new row was added. - New email →
falsebranch, one row was added. - The same new email again →
truebranch, no duplicate was created.
The final sheet contained exactly two records: the original one and the newly added one.
Same outcome, different paths
Both platforms produced the same result, but they represented an empty lookup differently. Make exposed Total number of bundles: 0, while n8n required Always Output Data to pass an empty item to the next node.
It is a small difference, but it determines whether the no-match branch runs at all.
P.S. Two requests arriving together can still add the same email twice. In production, let a database unique key supervise the children.



