Value the company and find the three biggest levers

When to use

Yearly, or when someone outside the company needs a number. Not for the annual plan itself — the range feeds it; build the plan in growth.annual-plan. Not for the annual report — that is the statutory filing in money.annual-report-lv; this playbook reads from it, it does not produce it. If a buyer has appeared, run this first, then take advice — this playbook is not a sale mandate.

Before you start

• The last closed year's figures are final (or the annual report is filed — check on ur.gov.lv).
• Owner salary and owner-only costs are known; if not, stop and get them first — they are the biggest single error.
• The purpose is stated. It changes the output: a bank wants the conservative end, an internal plan wants the levers.

What a run requires1

The trail7 steps

1. S1

   Assemble the inputsagent

   Pull the last closed year and current year to date from the money module; pull the filed annual report from ur.gov.lv as a cross-check.

   Done when turnover, gross margin, EBITDA, net result, debt list and owner costs each have one number and a source, and the two sources agree or the difference is written down.

2. S2

   Normalise the earningsagent

   Start from EBITDA; add back or remove: owner salary above/below market, owner-only costs, one-off income or expenses (a sold asset, a lawsuit, a lost customer's final invoice), and anything the new owner would not pay.

   Done when a normalised EBITDA figure exists with every adjustment on its own line with its amount.

   ⛔ Adding back "potential" income the company does not earn today is not normalisation — it is fiction.

3. S3

   Compute the range with two methodsagent

   Method A: multiple on normalised EBITDA — state the multiple used and where it came from (⚠ verify sector multiple; if no priced source, use a range and say it is an assumption). Method B: a second angle — asset value (what the assets would fetch) or a discounted cash view from the 13-week forecast annualised.

   Done when two ranges exist, each with its inputs, and an overlapping low–high range is written down.

4. S4

   Stress the rangeagent

   Recompute with the largest customer removed, and with the top line 20% down.

   Done when the low end of the range reflects the worst stated case, and the customer-concentration fact is written next to it.

   ⛔ A range whose low end assumes losing nothing is not a low end.

5. S5

   Draft the three leversagent

   For each: the metric it moves (gross margin %, customer concentration, recurring share of revenue, key-person dependence), the current value, the target, and the estimated effect on the range from S3 — as a number, even a rough one. Rank by effect ÷ effort.

   Done when three levers exist, each with a number, and the rest are listed as "considered, not top three, because …".

6. S6

   Review and acceptownerneeds approval · owner

   Approval · S6 · owner — the run stops until a named person records the decision

   Walk the owner through range, normalisation, stress cases and levers in 30 minutes. The owner accepts the range or corrects an input, and names which levers become work for the year.

   Done when the owner has said, in the run record: "range accepted as X–Y" or "corrected, because …", and which levers are in.

7. S7

   Record and hand backagent

   Save the valuation sheet in the value module; link each accepted lever to a task (bc tasks add) or a goal (bc goals add) with its number; note the run against last year's valuation if one exists.

   Done when the sheet is findable from the company record and each accepted lever has a task with a number in it.

Checks — how we know it worked

• Read the normalised EBITDA back to the owner line by line; the owner confirms each adjustment is real.
• The range's low end equals the worst stress case from S4, not the plain multiple.
• Every lever has a current value and a target value — "improve margin" is not a lever.
• The filed annual report on ur.gov.lv matches the figures used in S1.

If it goes wrong

What each step leaves behind

1. S1turnover, gross margin, EBITDA, net result, debt list and owner costs each have one number and a source, and the two sources agree or the difference is written down.
2. S2a normalised EBITDA figure exists with every adjustment on its own line with its amount.
3. S3two ranges exist, each with its inputs, and an overlapping low–high range is written down.
4. S4the low end of the range reflects the worst stated case, and the customer-concentration fact is written next to it.
5. S5three levers exist, each with a number, and the rest are listed as "considered, not top three, because …".
6. S6the owner has said, in the run record: "range accepted as X–Y" or "corrected, because …", and which levers are in.
7. S7the sheet is findable from the company record and each accepted lever has a task with a number in it.

Evidence to keep

The valuation sheet (inputs, adjustments, both methods, stress cases) · the filed report link from ur.gov.lv · the owner's acceptance or correction in S6 (who, when, what changed) · the lever list with numbers · link to last year's valuation.

How this playbook improves

After every run ask: did next year's actuals land inside this year's range — if not, which input was wrong? Did any lever get no task, and why? How long did S1 take — if the data was not ready, the fix is upstream in money.month-close, not here. A new version changes the step that produced the miss and says so in its change note.