Minutes, not the next sync
The webhook starts the run and the run finishes it. There is no batch window between the form and the Slack message.
Lead routing
A form submission arrives, the flow reads it, decides what it is worth, writes it to your CRM and tells the right person — and keeps the trace, so a lead that went nowhere has an explanation.
The problem
Speed to first reply is the number that moves conversion, and it is almost never lost in the reply. It is lost between the form and the person: a webhook into a spreadsheet, a hourly sync into the CRM, a round-robin rule inside a tool nobody on the revenue team can open.
The other cost is silence. When a lead does not get followed up, a linear automation tells you the run succeeded — it delivered the row it was asked to deliver. It cannot tell you which branch the lead took, what the enrichment came back with, or why it went to nurture instead of sales.
What the flow looks like
Numbered because the order is the execution order, not because it is a list. However many steps it has, the whole thing is one run.
Give your form, your ad platform or your website a URL. The payload arrives as data and the run starts on it — no polling, no intermediate spreadsheet.
Trim the free-text company name, split the full name, drop the tracking parameters you do not keep. Sandboxed Python, for the part that is genuinely fiddly.
An AI step turns the message box into fields you can branch on — intent, product area, company size — with a required output shape, so the next node never receives a paragraph where it expected a value.
An existing customer, a repeat form fill or a lead your team already owns should not be routed as new. Look it up before deciding, not after.
Company size, region, product and intent meet in one grid your revenue team can edit — the routing rule stops living in three tools and starts living in a table.
Sales-ready, nurture, or ignore. Each branch is explicit on the canvas, so the answer to "why did this go to nurture" is a path you can point at.
Create the HubSpot contact and deal, add the nurture leads to a SendGrid sequence, and post the sales-ready ones to the Slack channel that owns them.
Return the classification and the record link in the response, so your own site can confirm what happened instead of guessing.
What the flow can reach
A Connection node calls one of 52 packaged providers, or any HTTPS or JSON API you have credentials for. Six databases connect directly. Every call happens inside the run, with your credentials, and lands on the same trace as the rest of it.
What changes
The webhook starts the run and the run finishes it. There is no batch window between the form and the Slack message.
Whoever owns the pipeline edits the grid, tests it against real submissions in Sandbox and publishes a version. No ticket, no deploy, and the live version does not change until they say so.
Open a lead from six weeks ago and see the enrichment it got, the branch it took and what was written where — including the ones that quietly went nowhere.
Questions
Three ways. An Inbound Webhook node gives you a URL to paste into your form tool, your ad platform or your site. Your own backend can POST to the flow’s endpoint and read the classification in the response. Or a scheduled job can sweep rows from a database or an uploaded file and run the flow over each one.
No. One run is one execution of a published version, from the trigger to the last step. Eight steps, two branches and three outbound calls count as one run — and a branch that is not taken costs nothing.
The delivery is retried six times on a widening gap — immediately, then after 5 seconds, 30 seconds, 5 minutes, 1 hour and 23 hours. Retries are free; a lead that took three attempts to land is still one run. If the last attempt fails, your workspace admins are emailed with the endpoint and the response code.
Ready when you are
Build the flow in Sandbox, run your own cases through it, and read the trace before anything reaches a customer.