> For the complete documentation index, see [llms.txt](https://guides.tability.io/docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://guides.tability.io/docs/playbooks/forecast-whether-youll-hit-your-key-results.md).

# Forecast whether you'll hit your key results

Find out mid-quarter which key results are on course to land, which are at risk, and what to do about the ones that aren't.

To start, share this page with your AI assistant and say:

> Run this playbook

The instructions below are for your assistant. You don't need to read or copy them.

### Instructions for AI assistants <a href="#instructions-for-ai-assistants" id="instructions-for-ai-assistants"></a>

If the user shares this page (or an article linking to it) and asks you to run it, follow the steps below from start to finish. If they only ask you to read, summarize, or edit it, do that instead.

#### Ground rules <a href="#ground-rules" id="ground-rules"></a>

* Ask one short question at a time. Deliver a first forecast within three questions. Reuse anything the user has already told you.
* Confirm before touching a Tability workspace. Only use a workspace the user gives you by URL. Show its name and ID and get a yes before reading from it or writing to it.
* Show your working. For every forecast, give the numbers behind it: progress, time elapsed, trend, and confidence. Don't make up numbers, and mark forecasts with too little data as "can't tell".
* Treat a bad forecast as useful. Report it plainly and pair it with options. Never soften a red to protect anyone.
* Judge goals, not people. Describe missing updates as gaps to close, not failures. Praise honest red statuses and consistent updates.
* Keep the user in charge. You recommend; the user decides whether to push, adjust, or deprioritize. Ask before writing anything to Tability. Draft messages for the user to send; don't message anyone else.
* Never change targets or confidence on your own. An owner's confidence is theirs; flag disagreements instead.
* Verify what you write. Read records and scheduled jobs back before saying they worked.
* If something is blocked, keep the work done so far, tell the user the exact next step, and carry on with what you can.

#### 1. Choose the plan <a href="#id-1-choose-the-plan" id="id-1-choose-the-plan"></a>

Start with a brief intro and ask:

> Which plan should I forecast? Share the Tability workspace URL and the plan's name or link.

If they share only a workspace, list its current plans and ask which ones. Confirm the workspace and plans, then ask only what you still need, such as:

* Which key results are known to be non-linear (for example, revenue that lands at quarter end, or a launch late in the cycle).
* Whether there's an OKR meeting coming up this forecast should feed.

#### 2. Gather the data <a href="#id-2-gather-the-data" id="id-2-gather-the-data"></a>

For each plan, record the start date, end date, today's date, and the share of the cycle elapsed.

For each key result, collect:

* Start value, target, current value, and owner.
* Every check-in, with date, value, confidence, and comment.
* Any milestone or expected-progress targets set on the key result.
* Its initiatives and their status.

If the plan has many key results, read objective by objective and keep a running table.

#### 3. Forecast each key result <a href="#id-3-forecast-each-key-result" id="id-3-forecast-each-key-result"></a>

For each key result:

1. **Progress vs time.** Compare progress (share of the way from start to target) with time elapsed. Where the key result has milestone or expected-progress targets, compare against those instead of a straight line.
2. **Trend.** From the check-in history, project the current rate of progress to the end date. Note whether it's speeding up or slowing down.
3. **Confidence.** Read the latest confidence and how it has changed.
4. **Forecast.** Call it On course, At risk, Off course, or Can't tell, and give the projected end value.

Flag these patterns:

* Progress well behind time elapsed (for example, 70% through the cycle and 30% to target).
* Green confidence with progress or trend that says otherwise.
* Yellow (At risk) three or more check-ins in a row.
* A run of green late in the cycle with little progress, which tends to turn red at the end.
* Blocked or unstarted initiatives under a key result that's behind.

Also give each plan's average progress against time elapsed.

#### 4. Find the missing data <a href="#id-4-find-the-missing-data" id="id-4-find-the-missing-data"></a>

List every gap that makes the forecast unreliable:

* Key results with a check-in due but not done.
* Owners who missed two or more check-ins in a row on the same key result.
* Key results with no target, an unmeasurable target, or a 0% to 100% target with no metric behind it.

Mark the forecast for each affected key result as Can't tell unless other evidence supports a call. For each gap, name the owner and the missing information.

#### 5. Recommend what to do <a href="#id-5-recommend-what-to-do" id="id-5-recommend-what-to-do"></a>

For each key result forecast as At risk or Off course, lay out three options, and recommend one with a reason:

* **Push harder:** what extra focus, people, or help would get it back on course, and by when it must recover.
* **Adjust the target:** a realistic target based on the trend.
* **Deprioritize:** what the freed-up effort could go to instead.

List the questions to put to each owner: what they've done so far, what's blocking them, and what they need to get back on course.

Save everything as {plan-name}-forecast-{date}.md, marked Draft, using the outline at the end of this page. Share it with a short summary: how many key results are on course, at risk, off course, or can't tell; the three that most need attention; and the biggest data gaps. Ask whether they approve or want changes.

#### 6. Prepare the follow-up <a href="#id-6-prepare-the-follow-up" id="id-6-prepare-the-follow-up"></a>

With the user's approval:

* Draft an agenda for the next OKR meeting that starts with red and yellow key results, then data gaps.
* Draft a short note for each owner with a missing check-in, asking for the latest update in Tability. The user sends them.
* Draft a short message praising anyone who marked a goal red honestly and anyone with no missed updates.
* Apply any target changes or deprioritizations the user decides on, and read them back to confirm.

#### 7. Repeat and wrap up <a href="#id-7-repeat-and-wrap-up" id="id-7-repeat-and-wrap-up"></a>

Ask whether they want the forecast to repeat. If yes and this tool can run scheduled tasks, create one that reruns steps 2–5 every two weeks until the end of the cycle, and give it the forecast file and everything it needs, since it won't remember this conversation. It should only notify the user when a forecast changes, a new gap appears, or a key result crosses into At risk or Off course. If scheduling isn't available, give them a prompt to run themselves.

Mark the forecast file Final and share it with a short summary: the forecasts, the decisions made, the drafted messages, any changes applied in Tability with links, the schedule, and what the user still needs to do.

#### Forecast file outline <a href="#forecast-file-outline" id="forecast-file-outline"></a>

```markdown
# {Plan}: forecast for {date}

Status · Cycle dates · Time elapsed · Workspace · Plan link

## Summary

## Forecast by key result

| Key result | Owner | Progress | Time elapsed | Trend | Confidence | Forecast | Projected end value |

## Key results that need attention

## Missing data

## Options and recommendations

## Meeting agenda and drafted messages

## Decisions and Tability changes (links)

## Schedule
```

END OF PLAYBOOK
