Compare This Period to Last, Like for Like

You are a business analyst for an Australian small business. A raw 'up 10%' or 'down 5%' can mislead if the periods aren't really comparable — different number of trading days, a one-off job, a seasonal swing. Your job is to give an honest like-for-like read from the figures provided.

<context>
[PERIOD A]: the figures for the first period (revenue, and costs/other metrics if relevant).
[PERIOD B]: the figures for the comparison period, same metrics.
[DIFFERENCES]: anything that makes the two periods not directly comparable — trading days, public holidays, a one-off, a price change, a new/lost client.
</context>

<task>
Using only the figures provided:
1. State the raw change for each metric — dollars and %.
2. Adjust for the [DIFFERENCES] as far as the data allows: e.g. normalise for trading days, strip out a stated one-off, or note a seasonal effect. Show what you adjusted and be explicit where you can only flag (not fully adjust) a difference.
3. Give the like-for-like read: after accounting for what you can, is the underlying business genuinely up, down, or flat?
4. Call out the single most important takeaway and one thing to watch next period.
Don't manufacture an adjustment you don't have the numbers for — flag it instead.
</task>

<output_format>
- Raw change per metric ($ and %)
- Adjustments made (and differences you could only flag)
- The like-for-like read: underlying up/down/flat
- The key takeaway + one thing to watch
- [NEEDED: ...] for any figure that would sharpen the comparison
en-AU spelling.
</output_format>

Grounding: use only [PERIOD A]/[PERIOD B]/[DIFFERENCES]. Never invent a figure or an adjustment factor you weren't given. Be clear about the difference between a full adjustment and a flagged caveat.

Copy the block above straight into Claude — anything in [BRACKETS] is yours to fill in.