Most Routine questions come down to a few causes, and each has a quick fix. Find the situation that matches yours, then try the steps under it. For how Routines work, see Routines in Insights.
This article covers:
A conditional Routine did not send an email
A Routine says it cannot find a Skill
A Routine asks for a detail or says one is missing
An email is shorter or less complete than usual
An email points you to support
The same email arrives at every check
Figures differ from a dashboard
A condition returns different rows than you expected
A conditional Routine did not send an email
A conditional Routine sends an email only when its condition is met, so a quiet inbox often means the Routine is working. A few things can look like a missing email:
The condition was not met. A "results present" condition sends only when the check returns at least one row. A "results missing" condition sends only when it returns none.
It was the first check. For a "result changed" condition, the first check sets the starting point and sends nothing. Emails begin from the second check, when the result is different from the previous one.
The result did not change. "Changed" means the whole result is different, not that one row was added.
What to try:
Open the Routine and select Test condition, then Preview query. Confirm that it returns what you intended.
Check that the property, period, measure, and level are written the way you mean them. If not, rewrite Watch for and test again.
Confirm the Routine's behavior is the one you want: results present, results missing, changed, or unchanged.
A Routine says it cannot find a Skill
If a Routine runs a Skill and the email says the Skill cannot be found, the Skill was deleted or its identifier changed. The Routine finds a Skill by its identifier. For how a Routine runs a Skill, see Run a Skill Automatically with a Routine.
What to try:
Restore the Skill with the same identifier, or edit the Routine to use the new identifier.
If someone shared the Skill with you, check with them before relying on it. Their edits and deletions affect your Routine from the next run.
Test with Run now once it is fixed.
A Routine asks for a detail or says one is missing
A Routine runs without someone present, so a Skill that normally asks a question, such as which property to use, cannot ask. The email tells you which detail is missing.
What to try:
Put the answer in the Routine's Instructions. Name the property exactly as it appears in Insights.
Answer every question the Skill asks, such as dates or a threshold.
If the Skill still needs input you cannot supply, adjust the Skill so it needs none, or create a second version for scheduled use.
An email is shorter or less complete than usual
The same Routine can return a fuller or lighter analysis from one run to the next, especially for a broad scope. The headline figures stay the same, but sections or detail can differ.
What to try:
List the sections that must always appear and say what to write when a section has no data. For example: "Always include these sections: [your sections]. If a section has no data, keep its heading and write that no data was returned."
Ask it to say so when a detailed result is sampled, truncated, or only partly reviewed, instead of presenting it as complete.
Narrow the scope, for example one property or a shorter period, if you need the same depth every time.
For a scheduled Routine, use Run now more than once and compare the emails. Tighten the Instructions until the sections agree.
When you need every detailed row, use a saved query, a dashboard, or a scheduled export instead.
An email points you to support
When a Routine finds no data, it can sometimes suggest contacting support.
What to try: add a line to the Instructions such as "If there is no data, say so plainly and do not refer the reader to support."
The same email arrives at every check
A "results present" condition stays true for as long as the result is there, so the email arrives at every check. This is useful when you want a standing reminder, such as a list of values that are still unmapped.
What to try:
If you want an email only when something changes, rewrite Watch for so it sends when the result changes from the previous check.
If the standing reminder is not useful, resolve the underlying item, for example by mapping the value, or choose a less frequent check.
Figures differ from a dashboard
A Routine uses the data permissions of the person who created it, and it only sees what that person can see. A dashboard on screen may also have filters set, such as a property, currency, or comparison period. If the two cover different properties or settings, the figures differ, and neither is wrong.
What to try:
In the Instructions, name the property, the currency, the dates, and the comparison.
Open the dashboard and compare its filters with what the Routine's email says it used.
Check who created the Routine, because everyone who receives the email gets that person's view of the data.
A condition returns different rows than you expected
A condition is written from your own words, so small differences in wording change the result. Two common cases:
Level of detail. "Any stay date" checks the property as a whole. An Alert built on a dashboard tile might check each room type. Say which level you mean.
"Latest price" and "updated today". These return different rows. "Latest price" checks every stay date and inventory, while "updated today" checks only prices that changed today.
What to try:
Say the property, the period, the measure, and the level of detail in Watch for.
Use Test condition and Preview query, and compare the result with what you expected.
For ready-to-use wording, see Routine Examples and Templates in Insights.
Have product questions or see a result you did not expect? Select ? Help in the lower-left corner of the side navigation panel to contact our support team.