Set Workflow Entry Criteria: Keep the Door Narrow
How to set workflow entry criteria: add checks to the trigger so only the right people enrol, branch on fields you record, and define both paths of every fork.
To set workflow entry criteria, add conditions to your trigger so that only the right people are enrolled, then branch only on fields you have checked are reliably filled in, and define both paths of every fork. Keep the door narrow and widen it later.
A trigger alone usually catches too many people. The event is real, but it is broader than what you meant. Entry criteria are the checks that sit between the event and the workflow, and they are the cheapest protection you have against a rule acting on the wrong person.
How to set workflow entry criteria: start narrow
The guiding rule is that it is far easier to keep the wrong people out at the door than to handle them gracefully once they are inside. Start with the narrowest entry that still catches the people you intended. You can widen it later, when you have read what the workflow did and know who you missed.
The asymmetry is the reason. A workflow that misses a few people is an inconvenience. One that catches the wrong people is an incident. The first can be fixed quietly. The second tells someone, very precisely, that nobody is paying attention.
| Entry too wide | Entry too narrow |
|---|---|
| Wrong people receive tasks and messages | Some people are missed |
| Mistakes happen at scale and are visible | The gap is quiet and easy to correct |
| Hard to untangle afterwards | Easy to widen once you have evidence |
| Risk: an incident | Risk: an inconvenience |
What the trigger lets through
Take the guide's own example: a deal moved stage. That event includes deals moving backwards, deals moving into stages that do not need this workflow, and deals belonging to customers who are already handled elsewhere. All three are legitimate occurrences of the trigger, and none is who you meant.
The way to find your own list is to ask what else could cause the event. If you have already chosen the event with care, as described in how to choose a workflow trigger, you will have a list of extra cases ready. Each one becomes either a condition on entry or a reason to build a separate workflow.
Conditions are forks
A condition is a fork: if this is true, go one way, otherwise go the other. Entry criteria are a fork at the door, and the same idea applies inside the workflow wherever it needs to treat people differently.
Two rules keep forks honest.
- Branch only on facts you actually record. A condition is only as good as the field it reads.
- Define both paths. A condition with only a yes path is a place where people silently fall out of your process.
The second rule is the one most often broken, and it is exactly the dropped handoff you were trying to fix. Whoever answers no still needs somewhere to go, even if that somewhere is to stop on purpose.
Check the field before you trust it
The trap is branching on information that is not reliably captured. Suppose you branch on company size, but half your contacts have no company size. Half of your contacts will take the no path for the wrong reason, and you will not notice. The workflow runs, the numbers look plausible, and the error is invisible.
Before you build a condition, look at how often the field it depends on is actually filled in. Open a handful of real records and see. Then ask two further questions: what does an empty value do to the fork, and is that what you want?
| If the field is | Then the condition | What to do |
|---|---|---|
| Filled in for almost every record | Is safe to branch on | Build the fork with both paths |
| Filled in for about half of the records | Sends the rest down the wrong path | Fix the capture, or branch on something else |
| Often empty or inconsistent | Hides the problem | Do not branch on it yet |
| Only recently added | Will not apply to older records | Decide what older records should do |
This check is also what makes the weekly review useful later. When numbers look odd, a half-empty field is one of the first places to look.
One job per action, and keep the chain short
Once people are inside, what happens to them should be easy to read. An action is a single thing the workflow does: create a task, apply a tag, assign an owner, send a message, notify a teammate, update a record. Keep each step to one job so that when something goes wrong you can see which step did it.
Keep the whole workflow short, too. A long chain is hard to review, hard to change and easy to break, because a mistake early on propagates through everything after it. When a process has several phases, build several small workflows that hand off to each other, rather than one long one that tries to do everything. Short workflows are also easier to switch off, as described in when to switch off a workflow.
Personalisation is part of entry too
Inserting a name or company into a message makes it feel personal when the data is right and absurd when it is wrong. An empty first name produces "Hi ," and an all-capitals name copied from a form produces a greeting that shouts. Treat this as another entry question: who has the data to receive this message gracefully?
Check what a field looks like across your real records before you use it, and prefer phrasing that still reads naturally without it. The safest personalisation is the kind nobody notices when it fails.
A worked example in the guide's words
Here is the guide's example sentence: when someone submits the enquiry form, if they are not already a customer, assign them to me, create a task to reply within the working day, and send them a short confirmation that a person will be in touch.
| Part of the sentence | Role | Question to ask |
|---|---|---|
| When someone submits the enquiry form | Trigger | Does the form still do what it did? |
| If they are not already a customer | Entry criterion | Is customer status reliably recorded? |
| Assign them to me | Action, internal | Is there one owner? |
| Create a task to reply within the working day | Action, internal | Does the task reach the owner? |
| Send a short confirmation | Action, outward | Does it read naturally if a name is missing? |
Notice the second row. The check on customer status decides the whole workflow, and if that status is poorly recorded, existing customers will be treated as new enquiries. That is the half-empty field problem in miniature.
The cost of getting entry wrong, in plain terms
It helps to picture a single wrong enrolment. A customer who already paid is treated as a new enquiry, and receives a confirmation that a person will be in touch, plus a task is created for someone to reply to a person who needs no reply. Nothing crashes. Nobody sees an error. The rule did exactly what it was told.
That is what makes entry mistakes so persistent: the workflow is working and the people are wrong. Only a regular look at who actually came in, and why, will reveal it. This is why the review described in how to spot a silently failing workflow reads who entered as well as what happened next.
Entry and exit are a pair
Deciding who comes in is only half of the door. The other half is deciding when a person leaves, because founders reliably design the start carefully and forget the end. A narrow entry that has no goal will still chase someone who already said yes. Write the goal that removes the person at the same moment as you write the criteria that admit them.
The same pair of decisions also underlies the order in which to build, which how to decide what to automate first sets out.
A checklist
- I listed what else could cause my trigger to fire.
- Each extra case is handled by a condition or by a separate workflow.
- Every condition reads a field I have checked in real records.
- Every fork has both paths defined.
- Each action does one job.
- The chain is short enough to read in one sitting.
- The goal that removes a person is written down.
Frequently asked questions
How narrow is too narrow?
If the workflow misses people you wanted, widen it. That is a better problem to have than the opposite, and reading what the workflow did will tell you who was missed.
Should I use entry criteria or branches inside the workflow?
Use entry criteria for who should be in at all. Use branches for people who should be treated differently once inside. If the two groups need entirely different treatment, separate workflows are often easier to trust.
What if I do not know how often a field is filled in?
Then find out before you build. Open real records and look. If many are empty, fix how the field is captured, or choose a different basis for the fork.
Can I add criteria after the workflow is live?
You can, but treat the change like a new workflow: test it on a contact you control and read the result before you rely on it.
The full method, with the checklists, is in The Automation Playbook, a free 29-page playbook.