BeginnerLesson 3 of 305 min read

No-Code Platforms: Steps, Fields and Order

See what any no-code automation platform does underneath by writing a workflow as data and checking that every step has the fields it needs.

In this lesson you will

  • Describe a workflow as steps and fields
  • Find a step with a missing field
  • Test a workflow before switching it on
  • Connect apps with care

About 20 minutes

Meeting the platform for the first time

Ravi opens a no-code automation platform for the first time and sees boxes joined by arrows. One box says ‘new email’, another says ‘add a row’, a third says ‘send message’. He drags them around and wonders what the platform is really doing. The honest answer is that every such tool does the same thing underneath: it stores your boxes as a list of steps, runs them one after another and hands the results from each step to the next. This lesson speaks about platforms in general terms, because the buttons, names, plans and limits differ between products and change often. Check any tool’s own help pages for those details.

Anatomy of a workflow

  • Trigger. The first step. It starts the run and supplies the first pieces of data, such as the subject, body and sender of an email.
  • Actions. Later steps that do something: look up, write, send, label.
  • Fields. Named pieces of data. A step may need some fields and may produce new ones.
  • Connections. Permission you give the platform to act in another app on your behalf.
  • Run history. A log of what happened each time, which is how you find faults.

Most beginner mistakes come from the third item. A step asks for a field that nothing earlier has produced, so it either fails or quietly uses an empty value. The order of steps matters.

A checker that reads a workflow

The program below writes a workflow as plain data: each step has a name, a list of fields it needs and a list of fields it gives. A short function walks down the steps keeping a set of fields that exist so far. If a step needs a field that is not in that set, it is reported. This is a simplified version of the warning many platforms show when a step is not connected properly. The workflow has a deliberate fault: the last step sends a reply but no earlier step wrote a draft.

# a workflow as data: each step says what it needs and what it gives
def check(flow):
    have = set()
    problems = []
    for name, needs, gives in flow:
        missing = [n for n in needs if n not in have]
        if missing:
            problems.append((name, missing))
        have |= set(gives)
    return problems

flow = [
    ("new email arrives", [], ["subject", "body", "sender"]),
    ("classify email", ["subject", "body"], ["label"]),
    ("add a row to the sheet", ["label", "sender"], ["row"]),
    ("send the reply", ["label", "sender", "draft"], ["sent"]),
]
print("steps:", len(flow))
print("problems:", check(flow))

fixed = flow[:3] + [("write a draft reply", ["label", "body"], ["draft"])] + flow[3:]
print("after the fix:", check(fixed))
print("steps now:", len(fixed))
print("order:", [s[0] for s in fixed])
steps: 4
problems: [('send the reply', ['draft'])]
after the fix: []
steps now: 5
order: ['new email arrives', 'classify email', 'add a row to the sheet', 'write a draft reply', 'send the reply']

The workflow has 4 steps and the checker found one problem, which it printed as [(‘send the reply’, [‘draft’])]. In plain words, the step that sends the reply needs a draft, and nothing has produced one. The fix, shown in the second half, is to insert a step that writes a draft reply, after classifying the email and before sending. After that the checker reports no problems, and the workflow has 5 steps. Notice that the fix is about order and about supplying data, not about changing the sending step at all.

When you build in a real platform, you can do this check by eye. Click each step and ask where its fields come from. Most tools show a list of available fields from earlier steps. If a field is missing from that list, an earlier step is missing or in the wrong place.

Test with sample data

Almost every platform lets you run a single test with sample data before turning the workflow on. Use it. Send yourself a test email, run the workflow, and read the history of each step: what went in and what came out. Then try a strange email, with an empty subject or a very long body, to see how the workflow behaves.

Staying safe with connections

  • Connect with an account that has only the access the workflow needs.
  • Keep a record of which apps the platform may act in.
  • Remove connections you no longer use.
  • Never paste passwords or secret keys into a step’s text fields where others can read them.
Common mistake: skipping the test run

Turning a workflow on without a test run. A missing field can send blank messages to real customers.

Platforms differ

This lesson describes platforms in general. Names of features, plans, limits and prices vary and change, so none are quoted here. Check your chosen tool’s current documentation, and note that features may depend on your plan or region.

Sketch a workflow for one of your own tasks as a list of steps. Under each step write the fields it needs and the fields it gives. Then check, step by step, that every need has an earlier give.

Practice for this lesson

Check what you learned

Five quick questions. Pick an answer and you will see straight away whether it is right and why.

What is a field in a workflow?

  1. A type of trigger
  2. A named piece of data passed between steps
  3. A paid plan
  4. A folder of emails

Steps need some fields and produce others.

What did the checker find in the sample workflow?

  1. The trigger was missing
  2. Too many steps
  3. A wrong password
  4. The reply step needed a draft that no step made

The send step asked for a field that nothing had produced.

How was the sample workflow fixed?

  1. By deleting the classify step
  2. By adding a step that writes a draft before sending
  3. By removing the trigger
  4. By sending twice

The missing field was supplied by a new step placed earlier.

What should you do before turning a workflow on?

  1. Share it with customers
  2. Delete the run history
  3. Give it full account access
  4. Run a test with sample data and read each step

A test run shows what goes into and comes out of each step.

Why are product names and limits not quoted in this lesson?

  1. They differ between tools and change often
  2. They are secret
  3. They are always free
  4. They never matter

Check your tool’s current documentation for details.

Summary

  • A workflow is a list of steps that pass named fields along.
  • A step must not need a field that comes later or never.
  • Test with sample data and read each step.
  • Connect with the least access needed.

Next lesson

Lesson 4 shows how to move and tidy data between steps.

Questions

Which platform should I choose?

This course does not pick one. Compare current features, plans and support for the apps you use, and try a free or trial option if one is offered.

What if my platform shows different words from the lesson?

Names vary. Look for the same ideas: a trigger, actions, fields passed between steps and a history of runs.

Is it safe to give a platform access to my email?

Give only the access the workflow needs, review it regularly and read the platform's own security information first.

Progress is saved in this browser.