Engineering Delivery

Goals, alerts and Scout

Goals, alerts and Scout are the parts of Engineering Delivery that tell you when something needs attention and what to do about it. Goals give each measure a target, alerts ring the bell when something crosses a line your team cares about, what to improve next and the weekly note say what to look at, and Scout, Solidafy's AI assistant, answers questions about your numbers. This page explains each, and what to do when one of them speaks up.

Goals and budgets

A goal is the target a measure is held to, such as "change lead time of at most 7 days". A budget is the monthly limit for AI spend. You set both in Settings, under Goals and budgets, and every tile then shows how far there is to go.

Where do suggested goals come from?

Every goal starts from a suggestion with its source, "never a grade of your team". The suggestions come from your own numbers.

  • "Suggest goals 10% better than now" suggests, for each measure the app has read, a target 10% better than your current value, in the direction the measure improves. A suggested share never goes past 100%. It is offered once the app has measured your own numbers.
  • DORA's 2025 report publishes no benchmark and no performance levels, only how the teams it surveyed answered, and asks each team to improve against its own past. So no published figure is suggested as a goal. Performance levels and benchmarks explains what DORA publishes and how to use it well.

You can add the suggestions to your draft and then edit each goal before applying. If you already set a goal for a measure, the app asks before replacing it: you can fill only the empty goals instead. Setting the app up sets no goals.

What is the "Default" on a tile with no goal?

Until you set a goal, each measure shows a default: your own baseline, worked out the same way as the preset. A default is only a suggestion. It is never a goal your team set, so it never raises an alert and never reads as met or missed. To make it a real goal, use "Use the default" next to it in Settings.

What do I do when a goal is not met?

Treat it as a question, not a verdict.

  1. Open the measure and read its guide, especially "Where to start".
  2. Look at what to improve next at the top of the page, which names the figures behind the gap.
  3. Ask whether the goal is still right. A goal may simply not fit your system yet. Raise it step by step as the team gets there.

Warning: DORA warns against setting metrics as goals in a way that invites gaming. Use goals to start a conversation about how the team works, never to judge a person.

How do AI budgets work?

A budget is a monthly limit for AI spend, in total, by tool and by team. Each spend tile says how much is used, the days left and where the month is heading, for example "On pace for $X by month end". The states are:

  • "Under budget"
  • "Nearing the budget", once 80% or more is used
  • "Over budget"
  • "Set a budget", when there is none yet

The suggested budget is your past spend plus 10%: last month's total, or the average of the last few months.

Alerts

Alerts appear in the bell at the top of the app. Each person is alerted only about teams and services they can see. An alert is raised once and stays until what caused it clears, so the same problem does not alert again and again. You choose which alerts are on in Settings, under Alerts.

"A goal is not met"

When it fires: a measure you set a goal for is on the wrong side of it. People who see every team see it.

What to do: follow the steps under What do I do when a goal is not met? above.

"Review wait is most of the time to merge"

When it fires: changes spend more than half their time to merge waiting for a first review. People who see every team see it.

What it means: the biggest delay before merging is not the review itself but the queue before it. That is good news, because a queue is the easiest wait to shorten.

What to do: agree a review window as a team, review before starting something new, and keep changes small. See Time to first review.

"One service needs fixes more often"

When it fires: one service needs a fix after deploys clearly more often than every other. People who see that service's team see it.

What it means: a team average can hide one fragile service. This alert points straight at it.

What to do: open the service's releases that needed a fix, look for what they share (size, a missing test, a risky area), and add a check for the last failure. See Change fail rate by service.

"More changes merged without review"

When it fires: more changes were merged with no review than in the period before. People who see every team see it.

What it means: fewer changes are getting a second pair of eyes before they reach customers.

What to do: open the changes and look for a pattern, such as one repository without branch protection or late night fixes. Require a review on the main branch, and talk about the exceptions at the retro. See Changes merged without review.

Scout

Scout is Solidafy's AI assistant. In Engineering Delivery it reads the app's numbers, for the teams you can see, and never names or ranks a person. Most of what this section covers is written by the app's own rules, with no AI model: Scout only answers questions in Ask Scout, and writes the weekly note and the retro nudge if an admin chooses it to. You choose what is on, and who writes the notes, in Settings, under Scout.

Scout in the app is an operator the app installs, so you also find it under Workspace > Operators. Each person can ask it up to 10 questions a day, and it has its own monthly budget. Uninstalling the app removes it.

What is "What to improve next"?

At the top of the Overview, Speed, Review, Flow, Quality, Team health and AI pages, when a goal is missed or a measure got worse than the team's own last 90 days, a short recommendation headed "What to improve next" says:

  • Where: the part of delivery to look at, such as a service, a team, a stage or a size of change.
  • Why: what is off and why it matters, with a Source link to the published research it rests on.
  • Try: a concrete next step, with the figure behind it.

The app writes it from fixed rules, with no AI model, so the same numbers always give the same recommendation. It carries the line "From these numbers and the research cited with each. Team level only." Any goal the section is short of is listed under "Short of goal", and a figure under Try opens the items behind it. When nothing is off, the update shows its summary alone.

How should I use a recommendation?

As a starting point for a team conversation. Check the evidence, decide whether it matches what the team sees day to day, and if it does, make it the team's focus or a working agreement. If it does not, the numbers may point at a setting to fix, such as what counts as a deploy. See Setting it up.

What can I ask Scout?

Ask Scout is off until an environment admin turns it on in Settings, under Scout, with "Let people ask Scout questions". It is the one part of Scout that always uses an AI model, so each person can ask up to 10 questions a day. While it is off, the Ask Scout button is not shown. Once on, it opens from every page and suggests questions to start with:

  • "How is delivery going?"
  • "What should we look at before the retro?"
  • "Are we on track for our goals?"

As the panel says, "Answers come only from this app's numbers, for the teams you can see. No one is named." Scout cannot see data the app does not hold, and it will not compare people.

What is the weekly note?

A short note each week on delivery and review. It lists up to three things that are worth a look, with their figures, and says how goals are doing, for example "Your one goal is met." or "2 of 3 goals are met." You choose the weekday in Settings; it goes out early in the morning, European time.

By default the app writes the note itself, from the same numbers and rules as the pages, with no AI model. In Settings, under Scout, an admin can set Notes written by to "Scout, in its own words" instead. Scout then writes each note from the app's numbers, its figures are checked before it is sent, and it uses Scout's monthly budget. The same choice covers the retro nudge.

What is the retro nudge?

A reminder before the retro, saying how many changes have waited more than three days for a first review, so the team can deal with them or talk about them. With sprints it goes out before the sprint's retro; without sprints, every Thursday, the eve of a Friday retro. Like the weekly note, the app writes it from its rules unless an admin chose Scout to write the notes.

What to do: pick up the oldest waiting reviews first, then look at why they waited. See Time to first review.

Learn more