Is it cheaper to fix the agency you have or replace it?

Take a UK accountancy practice paying its agency £2,000 a month. Suppose the new agency needs three months before its work buys anything and notice overlaps by one month. Suppose also that the managing partner spends £1,500 of time choosing and briefing it. That is £9,500 spent before the practice sees a single better enquiry, and it appears on no proposal. Every number in that sentence is an assumption, which is the point: the cost of switching is a sum the firm can run on its own figures, and most advice on the subject comes from agencies that sell the switch. This piece gives the sum, then the one extra number that tells a partner whether to try a fix first. We cite no general cost of switching, because none exists that a seller of the service did not publish.

What a switch costs before it delivers anything

A switch has four lines, and three of them are paid whether or not the new agency is better. The first is the learning period: the months the new agency is paid while its work cannot yet be judged. The second is the overlap, because the old agency's notice can run while the new one starts. The third is any exit charge the contract sets. The fourth is the partner's time to select and brief. Run them on the firm's own fee.

Where the learning months come from

Google says the learning period for paid search depends mainly on conversions, conversion cycles and bid strategy. Its help page on the duration of the learning period names those three factors. A practice that logs a handful of enquiries a month collects conversions slowly, so its learning period may sit at the longer end, and a firm with more volume may need less. That is a reason to put the question to the incoming agency in writing, with the firm's own monthly enquiry count attached, before any months are entered on the worksheet. An agency that will not estimate its own learning period has told the partner something too.

What a fix costs

A fix is a trial with a date, and its cost is the shortfall that carries on while it runs. Choose a cure period, c, in months. Estimate the shortfall, g, as what the gap costs the firm each month in its own terms, such as enquiries missing against what the practice needs. Add the partner time, P, to run the trial. If the trial fails, the firm has paid c × g + P and still has to switch, so the switch cost is not saved. With c at 2 months, g at £1,500 and P at £800, a failed trial costs £3,800.

The number that decides it

Trying the fix first is the cheaper bet when its chance of working is above (c × g + P) divided by S. The algebra is short. Switching now costs S. Trying first costs c × g + P whether or not the fix works, plus S if it fails. The two are equal when the chance of success, p, equals (c × g + P) ÷ S. On the worksheet above that is £3,800 ÷ £9,500, which is 40%. If the partner believes the agency is more likely than that to put the problem right once it has been shown the evidence, the trial is cheaper. If not, switching is.

Which input moves the answer most

A long learning period makes a fix more attractive, and a long cure period makes it less so. Each extra learning month adds a month of fee to S, which lowers the chance a fix needs. Each extra month of cure period adds a month of shortfall to the cost of failure, which raises it. The partner controls the second. A trial agreed for one month rather than three costs a third as much in shortfall, so the first decision is to set the shortest period in which the change could honestly be seen. For paid search the length is partly set by the conversions the firm generates, so a very short trial may show nothing. In that case write a longer period and accept the higher threshold.

Writing the trial down

A trial only counts if both sides can later say whether it succeeded. Send the agency one email that names the problem, the measure the firm will read for itself, the cure period and the review date. Ask it to confirm the plan or to say what it would change. Reaching the review date with the measure unmet is then a fact on paper, and the worksheet takes over. If the measure was met, the firm has saved S and has a supplier that has shown it will respond to evidence, which a UK practice can reasonably weigh at the next renewal.

Estimating the chance without inventing it

No published figure says how often an agency fixes a problem it is shown, so the chance is the partner's judgement, and it can be tested. We make no claim about how often fixes work. What a managing partner can do is ask the agency what it will change and by when, and see whether the answer comes back dated and specific. An agency that can name the cause is more likely to correct it than one offering reassurance, and that is a reading of the answer, not a statistic. Our piece on running a review from the firm's own records shows how to find the cause first, and a paid audit is worth buying only if it beats that.

Which figures to enter

Use the cost the firm actually bears, which for VAT depends on its supplies. GOV.UK says a business that makes only exempt supplies cannot recover VAT on its purchases, and that one making both kinds is treated as partly exempt. A practice that reclaims the VAT on the agency's invoice should enter the net fee. One that cannot should enter the gross. Ask the firm's accountant which applies, because the answer changes the fee lines of the worksheet by 20%.

Do the sum with the contract open. Our guide to what to ask for before renewing covers reading the notice clause, and the exit charge belongs on the worksheet before any other line is argued over.

When not to switch, and when not to wait

Do not switch while the cause of the shortfall sits inside the firm. If enquiries arrive and the practice does not convert them, or the tracking that counts them was never set up, a new agency inherits the same gap and the worksheet has been paid for nothing. Check that before entering any shortfall. Our process for introducing a specialist starts from the same point.

Do not wait either when the threshold is above 100% or when the agency will not agree a dated measure. In either case the trial is not a trial. A partner who has run the numbers has a better reason to stay than loyalty, and a better reason to leave than irritation. For how we introduce accountancy practices, recruitment agencies, tax advisory and financial advice firms to one agency, see how we work.