Repeat Client Rate: The Math That Runs Your Business

When a company publishes a metric it invented, the securities rules require the real accounting number alongside it with equal or greater prominence, plus a quantitative reconciliation between the two. Repeat client rate is a metric you invented.

None of those rules reach a fishing guide. What they supply is the only serious published standard for a homemade number, and it is a standard almost no guide's repeat rate would meet: not because the arithmetic is hard, but because the denominator is usually undefined and the figure is quoted without the underlying counts it came from. Below the requirements are read from the regulation as a model, then the metric is built properly. Nothing described here binds you. This is not legal or financial advice. The rest of the operating numbers are indexed on the running the business hub.

What does the rule require of an invented metric?

Four things, and the first two are the substance.

Section 229.10(e)(1)(i) of Title 17 requires, whenever a non-standard financial measure is included in a filing, a presentation with equal or greater prominence of the most directly comparable measure calculated in accordance with generally accepted accounting principles.

It then requires a reconciliation, by schedule or other clearly understandable method, which must be quantitative for historical measures, of the differences between the invented measure and that comparable one.

Subparagraph (C) requires a statement of the reasons management believes the measure provides useful information, and (D) requires disclosure of any additional purposes for which management uses it.

Translated out of the filing context, that is: show the raw counts at least as prominently as the ratio, show how you got from one to the other, and say what the figure is for.

Part 229 is carried on the eCFR.

The file those counts have to come out of is the subject of the spreadsheet CRM piece.

Coastal Waters, MS

Why does the denominator decide everything?

Because three honest denominators give three different answers.

An operation running ninety trips, of which twenty-seven were taken by clients who had fished before, can say its repeat rate is thirty per cent.

The same operation, counting nineteen individuals who returned out of sixty-one who fished the previous season and could have, can say thirty-one per cent.

The same operation again, counting those nineteen against every client it has ever had, can say thirteen per cent.

All three are defensible, none is wrong, and quoting any of them without the counts underneath is the practice the disclosure rule exists to prevent.

Which is why the useful discipline is to write the fraction rather than the percentage: nineteen of sixty-one, not thirty-one per cent.

The same denominator problem wrecks margin comparisons, which the margin piece takes apart.

Shallow Water, FL

What does the rule forbid?

Smoothing, and it names the test.

Section 229.10(e)(1)(ii)(B) prohibits adjusting a measure to eliminate or smooth items identified as non-recurring, infrequent or unusual, where the nature of the item is such that it is reasonably likely to recur within two years, or there was a similar item within the prior two years.

Which is a precise and unusually honest test, and it maps directly onto the thing every guide does with a bad season.

A year where the water blew out for six weeks is not an unusual event in a river valley; it is a recurring feature that happened this time.

Excluding it to produce a cleaner repeat rate is exactly the smoothing the provision names, and the two-year recurrence test is the right question to ask yourself.

Subparagraph (A) separately prohibits excluding charges that required cash settlement from liquidity measures, other than two named exceptions.

A blown-out spring belongs in the record rather than out of it, which the fall wrap-up piece handles.

Which definition should a guide use?

People, not trips, and one season, not ever.

Counting people rather than trips removes the distortion where one client who fished six days makes the operation look retentive.

Counting against the prior season rather than against all history removes the distortion where a growing list makes the rate fall every year regardless of behaviour.

So the working definition is: distinct clients who fished this season and also fished last season, divided by distinct clients who fished last season.

Which answers the question that actually matters, being what share of last year's people came back, and which nothing else answers.

Write it as a fraction, keep the two counts beside it, and never quote the percentage alone.

Both counts fall out of a two-tab sheet, built in the spreadsheet CRM piece.

What should sit beside it?

Trips per returning client, which is the other half.

Retention and frequency are separate things and a single rate conflates them, which is why an operation can improve without the rate moving at all.

Nineteen clients returning for one day each and nineteen returning for three days each produce the same repeat rate and completely different businesses.

So the pair to track is the share of last year's clients who returned, and the average number of trips those returners took.

Together they tell you whether the work to do is bringing people back or getting the people who come back to book more.

Those are different problems with different answers, and the single number hides which one you have.

Frequency is normally the cheaper of the two to move, and the packages piece explains the mechanism.

What counts as a client?

Decide once, and the group booking is the hard case.

A four-person group booked by one organiser can be counted as one client or four, and the choice changes the rate by more than any operational improvement will.

Counting the payer is simpler and understates how many people you actually served; counting everyone aboard is more honest and requires you to have recorded their names.

Either is defensible, and switching between them from one year to the next is not, because it makes the series meaningless.

Which is the discipline the reconciliation requirement encodes: the comparison only works if the basis did not move.

Write the definition down beside the number, once, and leave it alone.

Capturing every name aboard starts at intake, laid out in the intake form piece.

What does the figure actually tell you?

Whether the product was good, with a year's delay.

Almost every other number in a guiding business measures effort or demand; this one measures whether people who experienced the thing wanted it again.

Which makes it the closest available proxy for quality, and it is unavailable until a year after the quality was delivered.

That lag is the reason it is worth tracking rather than reacting to: a rate that moved this winter reflects work done two summers ago.

It also means a change you make this season will not show in the figure until the winter after next, which is a long feedback loop and an argument for patience.

What it never tells you is why, which is what the cancellation reasons and the win-back replies are for.

The replies that supply the why are the raw material of the win-back piece.

What moves it?

Three things, in descending order of effect.

Asking at the takeout, which is the single largest intervention available and which most operations skip on most trips.

Sending the photographs the same evening, which is the strongest non-price act because it puts a permanent reminder on the client's phone.

And a message in the autumn and again in January, which reaches the people who intended to book and did not get round to it.

None of those is a discount, a campaign or a system, and together they account for most of the difference between a thirty per cent operation and a forty-two per cent one.

What does not move it is anything done in the week the client is deciding, because by then they have decided.

Phrasing for the takeout ask itself lives in the end-of-trip ask piece.

When should it be computed?

Once a year, in the autumn, from the trip records.

A monthly repeat rate is noise, because the population is seasonal and the denominator changes shape through the year.

Which makes it an annual figure computed after the last trip, alongside the rest of the wrap-up, when the records are complete and recent.

Computing it in February from a half-remembered season is how a number gets quoted for years without anybody being able to reproduce it.

Recording the two counts as well as the ratio is what makes next year's figure comparable, and it takes one line in a spreadsheet.

Three or four years of that produces a series worth reading, and one year produces a number worth nothing.

Everything else the autumn has to close out is listed in the fall wrap-up piece.

Should it be published?

Only with the counts, and probably not at all.

A repeat rate on a website is a claim, and the disclosure model above says a claim of that kind travels with its reconciliation or it misleads.

Which means publishing thirty-one per cent without nineteen of sixty-one beside it is the practice the rule prohibits, translated into marketing.

There is also very little reason to publish it, since a prospective client does not know what a good figure looks like and cannot check yours.

What does persuade is the specific: a named client on their eighth season, quoted with permission, does more than any percentage.

Keep the number for yourself and put the people on the site.

Named clients outperform percentages, which is the whole argument of the review request piece.

Where does this metric go wrong?

Five, and an undefined denominator heads the list.

Quoting a percentage without the two counts it came from, so nobody including you can reproduce it next year.

Changing the definition between years, which is the reconciliation failure the model names and which makes the series meaningless.

Counting trips rather than people, so one client fishing six days looks like retention.

Excluding a bad season as unusual, when the two-year recurrence test in the rule says a blown-out spring is a recurring feature.

And treating a single figure as a verdict on quality when it conflates retention with frequency, which are separate problems.

Where this one sits among the rest is set out in the numbers piece.

The strongest form of the behaviour it measures is described in the tradition piece.

What about the client who books every second year?

An annual window misclassifies them, and there are more of them than you think.

A destination client who fishes your water on alternate years is a loyal client by every reasonable meaning of the word, and an annual repeat rate scores them as lapsed in every year they miss.

Which matters most for operations whose water is a trip rather than a drive, because the biennial pattern is common there and the annual measure will read as a retention problem that is not one.

The correction is not to widen the window, since a two-year window makes the figure unresponsive to anything you do and delays the signal by another season.

It is to run the annual measure and to keep a second, flat count beside it: how many distinct people fished with you in the last twenty-four months.

That second number moves slowly, which is the point, and it catches the alternate-year client the first number keeps dropping.

It also has a straightforward reading: the twenty-four-month count is the size of your actual client base, and the annual rate is how much of it turned up this year.

An operation whose twenty-four-month count is growing while its annual rate is flat is doing fine and would look stagnant on the single number.

What does a falling rate usually mean?

Growth, more often than decline.

An operation that doubled its client list last year will show a lower repeat rate this year almost regardless of how good the fishing was, because the denominator filled with people who have not yet had the chance to come back twice.

Which is the single most common misreading of this metric, and it produces exactly the wrong response, being a panic about quality when the actual event was a good year of new business.

The disclosure model has a name for this problem too: the reconciliation exists because a ratio moves when either half moves, and a reader given only the ratio cannot tell which half did.

So the diagnostic is to look at the two counts before reacting to the rate.

Denominator up sharply, numerator up too, rate down: that is growth, and the rate will recover as the new cohort matures.

Denominator flat, numerator down: that is a retention problem and worth acting on.

Both down: that is a demand problem, and the repeat rate is the wrong instrument for it entirely.

Should the rate be split by trip type?

Once you have three years of records, yes, and it is where the number earns its keep.

A half-day walk-and-wade and a full-day float attract different people with different reasons to return, and a blended rate averages two behaviours that have nothing to do with each other.

Splitting it usually shows something actionable, most often that one product retains and the other does not, which is an argument about what to sell rather than an argument about service quality.

The same split by season is worth running where the calendar has distinct fisheries, because a spring client and an autumn client may be two populations that never overlap.

What makes the split possible is having recorded the trip type on the trip row at the time, which nobody does retroactively and everybody wishes they had.

The caution is sample size: a split that leaves eleven clients in a bucket produces a rate that swings twenty points on two people changing their minds.

Below roughly thirty in a bucket, look at the raw names rather than the percentage, because at that size the names are more informative than the arithmetic.

What should be written down at the end of the season?

One line, six figures, in the same place every year.

Distinct clients this season, distinct clients last season, how many of last season's returned, trips taken by those returners, total trips, and the definition of a client you used.

That line is the reconciliation, it takes two minutes with the records open, and it is the difference between a series you can read in five years and a number you will not be able to reproduce next March.

Writing the definition down each time is not redundant, because the definition is the thing that quietly drifts, and a drifted definition invalidates the comparison without ever announcing itself.

Keep the lines in one small sheet rather than scattered through annual notes, since the value is entirely in seeing them stacked.

Three years of that is the shortest series worth drawing a conclusion from, and it costs six minutes total to have built.

What is the working measure?

A fraction, a companion, and a written definition.

Count distinct clients who fished this season and also fished last season, over distinct clients who fished last season, and write it as a fraction.

Put the average number of trips those returners took beside it, because retention and frequency are different problems.

Write the definition of a client down once, decide the group-booking question, and never change either.

Compute it annually in the autumn from complete trip records, and record both counts rather than the ratio alone.

Compare only to your own prior years, since no consulted source publishes a figure for this trade and any external comparison would be invented.

The statutory framework behind the disclosure rules is at 15 U.S.C. 77g, with a parallel text on govinfo.

The full operating set this belongs to is assembled in the numbers piece.