What information do I need before I start the annual return?

England and Wales: researched. Scotland: not applicable to this step — OSCR’s own annual return has a different question set; see Step 8, Complete the OSCR annual return. Northern Ireland: not applicable to this step — CCNI’s own monitoring return has a different question set; see Step 9, Complete the CCNI annual monitoring return.

The question list is published separately from the form

Which is why so many charities start the return, get three sections in, and discover they are missing something that will take a week to reconstruct. The Commission publishes a question guide. Read it first, gather against it, then log in once.

The questions that need data you probably do not keep

Overseas income and spending. Questions 3.1 and 3.3 want money in and out of the UK broken down by country, by source type and by transfer method — including how much moved outside the regulated banking system. Almost no charity’s accounts hold that shape of data. If you have any overseas activity at all, this is the section to start on.

Highest single donations. Question 1.4 asks for the single largest donation from a corporate donor, from an individual, and from a related party, separately. That means someone has to sort the donation ledger three ways.

Headcount and payroll. Section 6 wants headcount by employment type, how many work outside the UK, total payroll spend, and employees in benefit bands. The bands do not line up with anything your accounts already produce.

Which policies were in place at the period end. Not now — at the period end. If you adopted a policy in March for a year ending in December, the answer is no.

Gross income here is a computation, not a line in your accounts

The return’s gross income figure strips out endowment received, adds back endowment converted into income funds, and excludes revaluation gains. It looks like the total on the face of your statement of financial activities and it is not the same number.

Work it out deliberately, write down how you got there, and use the same figure everywhere. It drives the threshold gates below, so getting it wrong changes which questions you are asked.

The gates run in the direction you do not expect

Question 1.3, the income breakdown, applies where gross income is at or below £500,000 and grants and contracts are under 70%% of income. Both limbs have to hold.

So growing past that ceiling removes a question rather than adding one. A charity that crosses it and diligently prepares the breakdown anyway has done work nobody asked for.

Question 1.4 works the ordinary way round: it applies where gross income is over £100,000.

Section 5, on property held by holding or custodian trustees, applies only to unincorporated charities.

Build the pack

One document, one row per question, with the answer and where it came from. Gaps stay visible as gaps until someone fills them. When it is complete, the return itself takes an afternoon rather than a fortnight.

Keep the pack afterwards. Next year it becomes the template, and the year after it becomes evidence of how a figure was arrived at if anyone asks.

Common mistake: starting the return online and discovering mid-way that data is missing

The question list is published separately from the form. Questions 3.1 and 3.3 want overseas income and spending broken down by country, by source type and by transfer method — that is not in most charities’ accounts.

Common mistake: using the accounts’ income figure without the return’s own adjustments

It looks like the same number. The return’s gross income strips endowment received and adds endowment converted into income funds, and excludes revaluation gains. It is a defined computation, not the SoFA total.

Worked example

Saltmarsh Trust has no grants, and its income sits at or below the ceiling in £500,000, with grants and contracts under 70%% of income — so the income breakdown applies, and it must answer question 1.3. A charity above that ceiling would not have to. The gate runs backwards from what most people expect.