AiAdvisors
Back to the index
//Weekly

Week 45 / 2026.

A short record of changes between issues. Full answers and charts live on the question pages.

Data window
Nov 2–8, 2026
Last updated
Nov 20, 2026
Author
Romuald Członkowski

Commentary

Between 2 and 8 November, n8n-mcp produced 76,320 workflows, 10,903 a day on average, across 1,971,275 tool calls with a 0.14% error rate. That is 4% more workflows a day than in the previous issue and 8% more calls, against 162,111 active installs, up 2%; 8% of them create workflows. The busiest day was 3 November with 13,300 workflows, the quietest 7 November with 6,653. Week-one retention stands at 31% of the 6,936 installs that started in the week of 26 October. 303 installs hold 31 workflows or more, 2.4% of the ones that build.

Every new workflow comes with 7.3 updates. The window holds 150,163 edits: all of them went through the API, 54% produced a verifiable improvement, at an average of 2.59 operations and 1.38 changed properties with a median of 843 ms. What usually changes is a node that already stands there: updateNode 47,436 times, patchNodeField 41,506, and among the known edit purposes a configuration change leads (56,299) ahead of adding functionality (18,790). Of the calls, 52% are reads, 12% writes, 6% documentation and validation, 4% tests, and fetching executions alone accounts for 33% of the total. Price an implementation on building the flow alone and you have priced the smaller part of the work.

The window added 28,675 workflows to the corpus, which now holds 1,307,312, with a median of 7 nodes and 51% running to 11 or more. Code appears in 73% of them, HTTP Request in 62%, Data Table in 12.9%, an LLM model node in 6% and an AI Agent in 3%, while 14% call a sub-workflow. Since March I have been seeing the same direction: the Data Table share has climbed from 3.3% to 13.2% in the November data, and the model-node share has fallen from 17% to 5%. Data table is also the week's most searched term, with 165 installs behind it, just ahead of webhook (164). Among model nodes, OpenAI holds 34%, Anthropic 24%, OpenRouter 15% and Google Gemini 15%. A webhook is the trigger in 45% of workflows, a schedule in 21%.

Validation turned away 0.8% of workflows, and again it is the smallest ones that trip over it: 1.56% at 2–3 nodes, against 0.12% at 31–100. Day by day the rate stayed between 0.38% and 1.34%, and since this metric swings between roughly 0.4% and 3%, I do not read a single week as a trend. 10% of connection tests end in an error, and 21% of those come down to an address or the network. This issue's pitfall is marked critical: SplitInBatches combined with a Wait node loses the batch iterator state, because it does not survive the pause, so items can be skipped or processed twice. Process batches synchronously or through a queue instead.

Changes versus the previous issue

figurevaluechange
Workflows created76,320—
Workflows per day10,903+4%
Tool calls1,971,275+8%
Calls ending in error0.14%+38%
Failed validation0.8%+21%
Active installs162,111+2%
Week-1 retention30.7%-7%
Workflows added to corpus (week)28,675+3%

This week's answers

Most searched nodes

data table (165) · Webhook (164) · Telegram (59) · Slack (45) · http request (44) · anthropic (30) · gemini (29) · whatsapp (29) · Gmail (22) · mcp client (22)

n8n pitfalls · critical

SplitInBatches + Wait nodes cause state loss

Combining SplitInBatches with Wait nodes causes state loss. The batch iterator state is not preserved across wait periods, leading to items being skipped or reprocessed. Process batches synchronously or use a queue-based approach instead.

//Contact

Let's talk about AI in your business

Tell Alicja what takes up time in your business. She'll book you a free strategy consultation with me.

Prefer a calendar? Pick a time

or write: romuald@aiadvisors.pl · +48 695 263 884