Vehicle livery, fleet graphics and building signage for businesses across the East Midlands. 30 staff.
Verdict: 1,690 pieces across 4 segments, with 290 accounts held back to measure the lift. Mailing, calls and design cost £8,406. Expected: 23 extra meetings (£359 each) and 7.7 extra wins, £12,544 of gross profit; after the meetings’ selling time, a net £1,330. The best segment: fleet operators, 20+ vehicles: door-opener pack. 2 segments don’t pay at the response expected: test them small or drop them; mailing only the segments that pay nets £3,856.
| Segment | Meetings per account | Break-even | Cost | Extra meetings | Gross profit | Net |
|---|---|---|---|---|---|---|
| Fleet operators, 20+ vehicles: door-opener pack 40 mailed, 10 held back | 15.0% without mail 3.0%; estimated | 7.0% | £1,833 | 4.8 | £6,048 less £576 of meetings | £3,639 Roll out |
| Lapsed customers: letter with a refresh offer 150 mailed, 30 held back | 6.0% without mail 2.0%; measured | 3.5% | £863 | 6.0 | £3,000 less £720 of meetings | £1,418 Roll out |
| Mid-size firms with vans: letter and vinyl sample 300 mailed, 50 held back | 4.0% without mail 1.0%; estimated | 5.4% | £3,070 | 9.0 | £3,150 less £1,080 of meetings | −£1,000 Test small or drop |
| Local businesses: shopfront signage postcard 1,200 mailed, 200 held back | 0.6% without mail 0.3%; estimated | never | £1,440 | 3.6 | £346 less £432 of meetings | −£1,526 Test small or drop |
Break-even is the meetings per account mailed at which the segment pays for its pieces, calls and meetings. The shopfront postcard never pays: a £1,200 first order at 20% conversion doesn’t cover three hours of selling time per meeting, however many respond. The mid-size van letter needs about 5.4% to pay against the 4% expected: test 100 first before mailing the rest. Mailing only the two segments that pay nets {gbp(K[23])}, against {gbp(K[17])} for all four.
Every segment has accounts held back and not mailed. After eight weeks, meetings booked per account in the mailed group less the holdout group is the real lift. It replaces the estimate, and decides whether the next wave goes out. At these response rates, though, each holdout should book fewer than one meeting, far too few to read segment by segment: Harbridge will read the holdout across the whole campaign and against the same weeks last year. Lapsed customers are the one segment measured from a past mailing; the rest are estimates until the test batches report.
Extra meetings are the meetings per account mailed less those expected without mail, times the accounts mailed; the holdout group shows the real figure afterwards. Response and win rates are Measured or Estimated as marked. Gross profit is extra wins × first-year order × margin; the selling time for each extra meeting, follow-up calls, print, postage and design are all costed. Repeat orders after the first year are not counted. Check postage at current Royal Mail business rates.
Fictional, continuing the example. Harbridge chose the guided route and, in the final session, set up an agent in an AI assistant the business already uses, connected read-only to the CRM and the wave 1 list file (its IT confirmed this before it committed). Following the plan, wave 1 went out on 2 November 2026: the door-opener packs and the lapsed-customer letters in full, the letter with a vinyl sample as a test batch of 100 (20 held out), and no shopfront postcards. Every Monday through the 8-week follow-up window the agent matched the meetings and wins in the CRM to the accounts mailed and held out, and flagged any follow-up call not made within 10 days of posting. It never posts, calls or emails anyone, and never changes a segment or a rate. Nothing left the business, and no decision changed, without one of its own people approving it.
| Date | What the agent reported | What was decided |
|---|---|---|
| 12 Oct No flag | Test run at set-up: the wave 1 list (290 to mail, 60 held out at random) matched to the CRM, every account found once. | Nothing needed. |
| 16 Nov Flag | Six of the 40 door-opener follow-up calls had not been made 10 days after posting. | The sales manager made them by 18 November; two meetings were booked. |
| 23 Nov Flag | Two recipients asked not to be mailed again, one a sole trader registered with the Mailing Preference Service. | The office manager added both to the suppression list; wave 2's lists will be checked again first. |
| 7 Dec No flag | Week 5 of 8: 14 meetings from the 290 mailed accounts (4.8%), 1 from the 60 held out (1.7%). | Nothing needed. |
| 4 Jan Flag | Window closed: the letter with a vinyl sample booked 3 meetings from 100 (3.0%), below the 5.5% it needs to break even. | Not rolled out; a plain letter without the sample will be tested on 100 accounts in February. |
| 4 Jan No flag | Window closed: the door-opener packs (17.5%) and the lapsed-customer letters (6.7%) are above break-even: 15 extra meetings and 3 extra wins against the holdout. | Wave 2 of both in March; the rates stay Estimated until a second wave confirms them. |
At the 30-day check-in on 11 November, Geoff and the managing director read the log together; no rule needed changing. When the window closed, the wave had booked 15 extra meetings against the holdout, 13.8 planned at these volumes, and 3 extra wins so far against 5.4 planned: about £5,600 of first-year gross profit at the plan's order and margin, with quotations still open. Each segment's holdout was too small to read alone, so the lift was read across the wave (6.9% of mailed accounts booked a meeting, against 1.7% of those held out). What the agent added: the counting happened every week without anyone matching lists by hand, six late calls were made while the packs were still remembered, and the test batch was read against its break-even before anything was rolled out. The months after are measured results of the mailing, not a claim for the agent.