The fastest way to waste a subscription audit is to export 90 days of card charges and call the list complete. A 90-day window misses annual renewals by design: a charge billed once a year simply does not appear in a single quarter of statements, however carefully that quarter is read. December is precisely when annual charges matter, because a tool that renews in January will otherwise roll into a new term before anyone has voted on it. Pull 12 months. That single change is the audit.
There is a second trap underneath the first. Most audits decide keep or cancel on the subscription fee alone. Research cited in the packet puts the fee at only 25–40% of what a tool actually costs once operating labour is counted (s1). A $40-a-month tool that nobody administers, secures, or troubleshoots cleanly is not a $480-a-year decision. On the sourced range, its true annual cost sits somewhere between about $1,200 and $1,920 — the fee divided by 0.40 at the low end and by 0.25 at the high end. That is a labelled calculation from the sourced range, not a quote for any specific product. Decide on the larger number.
Pull 12 months, not 90 days: building the complete list
Follow the money through every channel it can hide in. The sourced method is to check company cards across 12 months of statements, bank direct debits, expense reports, and app stores (s1). Work through the list in order, and give every charge a row in a simple register: tool, owner, renewal date, billing cycle, number of paid seats, and monthly or annual cost.
- Export the full year. Download 12 months from each company card and bank account, not the default recent-activity view. Sort by payee name, because the same tool often appears under two or three billing names.
- Sweep the side doors. Check expense reports and app-store receipts for tools bought personally and reimbursed, or bought on a phone and never registered centrally (s1).
- Name an owner for every row. A tool with no named owner is already a cancel candidate. Ownership here means one person who can say who uses it and what breaks if it disappears.
- Record the renewal date, not just the price. Annual tools need a decision date at least a month before renewal. That date, not December itself, is the real deadline for each row.
What success looks like: every recurring charge from the last 12 months appears once in the register, with an owner, a seat count, and a renewal date. Nothing is left as a mystery line on a card statement.
Seats vs humans: the fastest saving in the building
The most common finding in a first-time audit is not a forgotten tool. It is unused seats on tools the business actively uses (s3). The platform is valued, the work is real — and the business is paying for people who left, changed roles, or never activated an account.
Do this seat by seat. Export the user list from each per-seat tool and compare it against the current team list. Flag three groups: people who no longer work at the business, duplicate accounts for the same person, and accounts with no activity in the review period. A sourced rule from the packet is blunt: if a tool has had no active users in ninety days, cancel it at the next renewal (s2). Apply the same ninety-day lens inside a kept tool — a seat with no active user in ninety days is a downgrade or removal candidate at renewal, unless the owner can name a seasonal reason in writing.
| Finding | Test | Verdict |
|---|---|---|
| Tool unused entirely | No active users in 90 days (s2) | Cancel at next renewal |
| Tool used, seats wasted | Paid seats exceed active users; under 30–40% of paid seats active means investigate immediately (s1) | Remove or downgrade seats |
| Two tools doing one job | Owners of both tools describe the same task | Consolidate to one, cancel the other |
| Tool used and fully seated | Active users match paid seats | Keep — and renegotiate before renewal |
True cost, then the verdict: keep, downgrade, consolidate, cancel
With the register complete and the seats counted, price each tool honestly before judging it. Take the annual fee and apply the sourced total-cost range: if the fee is 25–40% of true cost (s1), a $1,200 annual fee implies a true cost of roughly $3,000 to $4,800 once administration, onboarding, support, and workarounds are counted. The reader can redo that math — fee divided by 0.40, then fee divided by 0.25 — for any row in the register. Tools that looked cheap on fee alone often fail on true cost; tools with a higher fee and no admin burden sometimes pass.
Then assign every tool one outcome, with no undecided rows left behind: Keep, Renegotiate, Downgrade, or Cancel (s1). Two timing rules sharpen the verdicts. Use the thirty-day window before renewal to renegotiate tools you are keeping (s2) — a vendor facing a dated cancellation alternative negotiates differently from a vendor facing an auto-renewal. And add a purchasing rule for next year: new subscriptions above a threshold the business sets need one person's sign-off and a register entry (s2). The threshold amount is the business's own choice; the sourced point is that a threshold and a single sign-off exist at all.
Expect the work to take real time. For a business running 20–50 tools, a prepared audit takes 2–4 hours, and the first one takes longer (s3). Budget a half day for the first pass. That time cost is also the honest limit on the exercise: chasing a tiny monthly charge through three support chats can cost more in labour than the charge itself, which is exactly why the true-cost lens cuts both ways.
Locking it in: the register and the renewal calendar
Close the audit by turning findings into dates. Enter every kept tool's renewal date into a calendar with a reminder thirty days ahead — the renegotiation window from the packet (s2) only works if someone is reminded to use it. Record each cancellation with its effective date and the person who confirmed that no data export was needed first; cancelling a tool that holds the only copy of client files or invoices is not a saving. Finally, re-run the seat count whenever someone joins or leaves, not once a year. Seat waste is a hiring-and-leaving problem that happens to be billed monthly, and a register updated at each change keeps next December's audit to the 2–4 hour prepared version (s3) instead of another archaeological dig.
— Penny Hughes, Money Writer — Saving Desk