Pick the Number That Matters

Turn a goal into five numbers or fewer, tell leading from lagging, and leave the vanity numbers on the shelf.

The easiest way to ruin a good night is to look at too many numbers about it. You can count the door, the bar, the tabs, the followers, the reach, the saves, the opens, the clicks, the reviews, the stars, and by Monday morning you’ll have a page of figures and no idea what to do. The numbers will all be true, and none of them will be a decision. That’s the problem this chapter solves. It isn’t that the figures are wrong. It’s that nobody decided in advance which of them would change what the room does next week, so each one gets a glance and none of them gets a decision.

The cure is a short list chosen on purpose. Start with the goal you wrote on the Chapter 4 page, work out which stage of the Chapter 3 funnel it lives in, and pick the few numbers that show whether that stage is moving. Five is the ceiling. Most rooms do better with three. Everything else you can still look at, the way you glance at the weather, but it doesn’t get to run the meeting. The discipline is choosing, and choosing means leaving most numbers off the page on purpose, even ones you like.

You’ll also learn two distinctions that sort numbers quickly. Leading numbers show up early and predict what’s coming. Lagging numbers show up late and confirm what already happened. Vanity numbers feel good and change nothing, while actionable numbers point at a choice. By the end of the chapter you’ll fill in a metric tree, one page that links a goal to a handful of numbers, each with an owner, a source, and a decision attached. The chapter moves from goals to numbers, explains leading and lagging numbers, separates vanity from actionable ones, builds the tree by funnel stage, and ends with how to keep the list short over time.

PR TACTIC: The So-What Test

What it is: For any number you want to track, finish the sentence: “If this goes up or down, I will ___.”

Why it works: A number with no action behind it is decoration, and a number that triggers a clear action earns its place on the list.

Try this: Run the test on every number you looked at this week and cross out the ones where you can’t finish the sentence.

22.1 From a Goal to a Number

A goal is a change you want in the world, and a number is the way you’ll know it happened. The connection between them is often loose, which is why so many rooms track something and still can’t say whether they’re succeeding. “Build a bigger Thursday” can be measured as total attendance, first-timers, regulars, bar sales, or the number of people who stay past ten. Each measures a different thing, and each would lead you to a different tactic. The first job is to say, in plain words, which one you mean. A quick way to find the right number is to ask what a successful month would look like to a regular, not to the owner. A regular wouldn’t say “we got more reach.” They’d say “the place is fuller on Thursdays,” “there are more new faces who stick around,” or “the lesson actually gets people dancing.” Each of those points to a number you can count.

Start from the funnel. If your goal is about strangers finding you, the number lives at the First-Timer stage, such as new faces per week. If it’s about turning visitors into returners, it lives at the First Win or Regular stage, such as the second-visit rate. If it’s about people bringing others, it lives at Ambassador, such as guests brought per regular. The stage tells you what kind of number to look for. A goal about reach measured with a loyalty number, or a goal about loyalty measured with reach, will give you results that look unrelated to the work. Each funnel stage also has a natural window. Discovery numbers move week to week. Return numbers need thirty to sixty days. Regular and Ambassador numbers need a quarter. Choosing a number means accepting its clock, and planning reviews that match it. If a goal sits between two stages, write two numbers and say which one is in charge. The other is a supporting player.

Then write the number as a sentence with a unit, a window, and a place to find it. “Second-visit rate: first-timers who came back within thirty days, divided by first-timers, from the door tally, each month.” That sentence is a definition, and it protects you from changing the rules halfway. If the definition shifts, the old and new numbers can’t be compared, and you’ve lost your baseline. Write the definition on the metric tree and don’t edit it without noting the date. A number is only as honest as its definition. Definitions also protect against quiet drift. If one bartender counts a returning guest as anyone who has been before and another counts only people they recognize, the same night will produce two different second-visit rates. Write the rule down and teach it to everyone who counts.

QUESTIONS PEOPLE ASK

“What if I don’t know my goal yet?” Then the first number is the funnel snapshot from Chapter 3. It will show you where the room leaks, and that leak is usually your goal.

“Can two roles share a number?” Yes, if they agree on its definition and on who collects it. Shared numbers, like the second-visit rate, are often the most useful ones in the room.

22.2 Leading and Lagging

Some numbers tell you what is about to happen and some tell you what already did. A lagging number is the result you care about: the count of regulars, the margin on a night, the rebooking rate. By the time it moves, the cause is weeks old. A leading number is an early sign that the result is on its way: first-timers this month, the share of them who got a personal welcome, replies to a pitch, the number of confirmed commitments for a launch. Leading numbers move sooner and are usually closer to something you can change tomorrow. In a honky tonk, leading numbers are often things the staff do, and lagging numbers are things the guests do. The welcome rate, the share of first-timers introduced to a regular, and the number of lesson students walked to the floor are leading. Second visits, regulars, and margin are lagging. That pairing gives every staff member a number they can affect tonight.

You need both, because each fails alone. A room that watches only lagging numbers learns about a problem when it’s a quarter old. A room that watches only leading numbers can feel busy and successful while the result never arrives. Pair them. If your goal is more regulars, the lagging number is regulars per month. The leading numbers might be first-timers per month and the second-visit rate at fourteen days. When the leading numbers rise and the lagging one doesn’t follow within a few weeks, something between them is broken, and that’s where to look. Plan the review dates around both. Look at leading numbers weekly, in the fifteen-minute check from Chapter 34, and at lagging numbers monthly or quarterly. Looking at a lagging number every week mostly shows you noise, and looking at a leading number only once a quarter means acting too late.

A leading number is a guess about the future, so check its reputation. Does it actually predict the result in your room? Compare for a few months. If the second-visit rate rises and regulars rise a couple of months later, it’s a good leading number for you. If it rises and nothing follows, drop it. The pairing is a hypothesis you test with your own history, not a law of nature. Chapter 25 puts the funnel numbers in order, and Chapter 31 shows how to test a change without fooling yourself. Some leading numbers will fail the test, and that is useful. If more welcomes don’t lead to more returns in your room after three months, either the welcome isn’t the problem or the welcome isn’t being done well. Both are worth knowing, and neither would be visible without the comparison. A simple chart helps: put the leading number and its lagging partner on the same page, month by month, and look at whether they rise and fall together.

LEADING AND LAGGING PAIRS BY ROLE

Owner: leading, first-timers welcomed within ninety seconds. Lagging, thirty-day return rate.

Booker: leading, pitch replies within a week. Lagging, rebook rate over the year.

Musician: leading, sign-ups at the merch table. Lagging, fans who come to a second show.

Bartender: leading, sends per reach on room clips. Lagging, guests per month who say content brought them.

Dance Teacher: leading, students walked to the floor after the lesson. Lagging, second-lesson rate.

Check each leading number against its lagging partner after three months. Keep the pairs that move together.

22.3 Vanity and Actionable

Eric Ries, in The Lean Startup, popularized a useful pair of terms: vanity metrics, which look good and tell you little, and actionable metrics, which show cause and effect and tell you what to do. Followers are the classic vanity number. They only go up, they’re easy to boast about, and they don’t say whether anyone came. Likes are close cousins. Total reach, impressions, and page views have the same property: they grow whether or not the room fills. None of these is worthless. They’re context, like the temperature. They just shouldn’t sit at the top of the page. Vanity numbers can become actionable when they’re narrowed. Total followers is vanity. New local followers this month who later came in, counted with the how’d-you-hear card, is actionable. Total reach is vanity. Sends per reach on clips of the room is closer, because it tells you which ideas travel.

An actionable number has three traits. It’s tied to a specific group and a specific window, so you can compare like with like. It’s a rate or a count of people, not a cumulative total. And it points to something you could change. “Followers who visited within thirty days” is actionable. “Followers” is not. “Sends per reach” on a clip is actionable, because it tells you which kind of post to make more of. “Likes” are not, because they don’t. The conversion is usually to ask “out of whom?” and “within how long?” until the number describes people and behavior, not volume. Cumulative numbers, such as total guests since opening or total followers since the account began, are always vanity in this sense, because they can only rise. Turn them into rates over a window, such as this month’s first-timers or this quarter’s new regulars, and they start to tell you something.

Be careful with Goodhart’s law, which Chapter 21 introduced: when a number becomes the target, it stops being a good measure. If you reward staff for sign-ups, you’ll get sign-ups from people who don’t care. If you judge a night by door count alone, you’ll discount the cover until the room is full of people who spend nothing. Choose numbers that are hard to game without also doing the right thing, and pair each with a check. Door count, with spend per head. Sign-ups, with active rate. A pair like that is difficult to inflate without helping the room. Pairs also make conversations easier. When a bartender’s sign-ups rise but the active rate falls, the pair points to a specific fix, such as clearer promises at sign-up, without anyone feeling accused. When a staff member proposes a new number, ask the three questions together, out loud. Most people can tell on their own whether a number passes, once they’ve heard the questions.

22.4 The Metric Tree by Funnel Stage

A metric tree is a goal at the top and a few numbers beneath it, branching by funnel stage. Below is a starting set of examples for each stage, written for the five roles. Notice what they have in common: every one is a count or rate of people, every one has an obvious place to collect it, and every one could prompt a decision. You won’t use all of them. Pick the stage your goal lives in and take one lagging and one or two leading numbers from that row, then add one guard number to keep the first honest. If a goal spans two stages, such as “more first-timers who become regulars,” pick the stage where the room is weakest according to the funnel snapshot in Chapter 3, and build the tree there first. A second tree can come next quarter.

The tree matters because it shows how the numbers relate. If your goal is “more regulars,” the tree might read: regulars per month at the top, the second-visit rate and visit frequency below it, and first-timers per month at the base. Each lower number feeds the one above. When the top moves, you can look down the branches to see which one carried it. When it doesn’t, you can see which branch stalled. A plain tree, drawn on one page, will explain more about your room than a dozen dashboards. Draw the tree on paper first. Boxes and lines, with the goal at the top, two or three numbers under it, and one guard to the side. Many owners find that drawing it shows them a number they were missing, such as the share of lesson students who reach the floor, which links the teacher’s work to the owner’s goal.

Keep the tree small enough to say in one breath. If you can’t recite your numbers to a bartender without notes, you have too many. Write each number’s owner, who collects it and when, and where it lives, such as the door tally, the point-of-sale report, or the platform’s own analytics page. Then set a review date. A tree with owners and dates is a plan, and a tree without them is a poster. The worksheet at the end of this chapter gives you the form, and Chapter 34 shows how the numbers feed a monthly scorecard. Once a quarter, read the tree aloud at the staff meeting and ask whether every number on it is still useful. A tree that never changes has probably stopped being read.

STARTER NUMBERS BY FUNNEL STAGE

• First-Timer. Lagging number: New faces per week or month. Leading numbers: Reach to profile visits; map views; shares of a clip; pitches answered. Where it lives: Door tally; platform analytics; business profile.
• First Win. Lagging number: Second-visit rate within 30 days. Leading numbers: Share greeted in 90 seconds; first-night survey answer; sign-ups with permission. Where it lives: Door tally; host notes; list tool.
• Regular. Lagging number: Regulars per month (3+ visits in 60 days). Leading numbers: Visit frequency; 30- and 60-day return; active rate of the list. Where it lives: Cohort sheet; point-of-sale; list tool.
• Ambassador. Lagging number: Guests brought per regular; share of first-timers sent by a friend. Leading numbers: Regulars in a role; thank-yous given within a week. Where it lives: How’d-you-hear tally; ladder list.
• Protection / Infrastructure. Lagging number: Failures avoided; margin per night. Leading numbers: Consent-log completeness; review response time; hours worked. Where it lives: Consent log; ledger; review tracker.

Pick one lagging and one or two leading numbers from a single row, plus one guard number. Five numbers total is the ceiling.

COMMON MISTAKES WHEN PICKING NUMBERS

Picking the number the platform puts first. The top of a dashboard reflects what the platform wants you to care about, not what your goal is.

Mixing stages. A goal about loyalty measured by reach, or a goal about discovery measured by regulars, will make good work look like failure.

Leaving out the window. “Returns” means nothing without “within thirty days.”

Changing definitions quietly. If the definition changes, the baseline has to be rebuilt and the date noted.

No guard. Every number that people will try to improve needs a partner that would catch them improving it the wrong way.

Too many numbers. If you can’t say them in one breath, cut until you can.

22.5 Five or Fewer, and When to Retire One

Five is a ceiling, not a target. A room with one clear goal this quarter may need only three numbers, and that’s better. Choose them by running each candidate through the So-What Test and the “out of whom, within how long” question. Keep the ones that pass. Then check the set as a whole: does it have at least one lagging number, one leading number, and one guard that would catch a cheat? If so, it’s done. If not, swap rather than add. The test from Chapter 21 works well here. Could this number go up while the room gets worse? Does it have a known normal swing? Would a high or low reading change what we do? A candidate that fails any of the three is context, not a number for the page.

Numbers need a retirement plan. When a goal changes, the numbers attached to it should change with it, and any number that has gone three months without prompting a decision should be dropped. Dropping isn’t forgetting: keep the history in a sheet so you can bring it back if the goal returns. Review the list at the quarterly keep, fix, or kill review from Chapter 34. The discipline of retiring a number is the other half of choosing one, and it’s what keeps the list from slowly growing back into the dashboard you started with. Retiring a number doesn’t mean deleting the data. Keep collecting it if it’s cheap, and keep it in the background. If something changes later, the history will be there. What changes is that it no longer has a place in the meeting.

Finally, remember what the numbers are for. They’re there to help you make a decision, not to prove that you’re doing well. If a number is bad news, the right response is curiosity. If it’s good news, the right response is suspicion, at least until you’ve checked the baseline and the noise. Chapter 23 builds the baseline, which is the thing that lets you tell the difference. A short list of numbers, defined clearly, with a known normal swing, is the foundation of everything that follows in this part of the book. It also lets you be candid in public. A room that can say “our second-visit rate fell this quarter, and here’s what we’re doing about it” to its staff, its partners, and its own owner is a room that is learning.

A NUMBER’S ID CARD

Name: what you call it. Definition: exactly what is counted, out of whom, in what window. Source: where the count comes from. Collector: who counts. Baseline: average, median, and swing. Decision: what you’ll do if it rises or falls outside the swing. Guard: the number that would catch cheating. Started: the date you began counting it this way.

If you can’t fill in every line, the number isn’t ready for the scoreboard.

TRY THIS WEEK

Write your goal at the top of a card. Under it, write one lagging number, one leading number, and one guard, each with its definition and window. Show the card to one other person in the room and ask whether they’d count each number the same way. Fix any definition they’d read differently.

CASE STUDY

Fourteen Numbers Down to Four

Percy books a Wednesday night at a dancehall. His goal for the quarter is “turn more Wednesday first-timers into regulars.” He starts with 14 numbers he looks at every week: followers, post likes, post reach, profile visits, email opens, email clicks, door count, bar sales, first-timers, second-visit rate, regulars, reviews, average rating, and tips. He runs each one through the So-What Test.

Step 1 — The cut. For ten of them he can’t finish the sentence. Followers, likes, reach, opens, reviews, rating, tips, bar sales, profile visits, and email clicks don’t tell him what to do about turning first-timers into regulars. That leaves door count, first-timers, second-visit rate, and regulars. 4 of 14 survive, which is 4 ÷ 14 = 29%.

Step 2 — The tree. Lagging: regulars per month. Leading: second-visit rate within thirty days, and first-timers per month. Guard: door count, so that a rise in regulars isn’t just the result of fewer first-timers coming at all.

Step 3 — The first month. First-timers: 36. Second-visit: 9. Rate: 9 ÷ 36 = 25%. Regulars: 6. Door count averaged 48. He writes each number’s source on the page and sets the review date.

Step 4 — What he gives up. He keeps watching reach and likes but doesn’t put them in the meeting. Twelve weeks later, the rate is 36%, and regulars are 11, a rise of 5 over the first month. He checks the leading and lagging pair: the second-visit rate rose first, and regulars followed about six weeks later, which supports the pairing for his room. All figures are illustrative.

REAL TALK

From the booking desk:

I used to bring a stack of printouts to every Monday meeting. The owner’s eyes glazed over by page two. Now I bring one index card with four numbers on it. We actually argue about them, which means they matter.

HEADS UP

Choosing numbers about people can create privacy duties. Tracking guests by name, phone number, or visit history may bring consent and data-handling rules, including the federal CAN-SPAM Act for marketing email, federal and Texas rules on marketing texts, and the Texas Data Privacy and Security Act, which exempts most small businesses but not every obligation. This is general information, not legal advice. Prefer counts and tallies that don’t identify anyone, and before you collect personal data, ask a licensed Texas attorney, and see Chapter 35 and Appendix J.

YOUR METRIC TREE

Fill this in for one goal, using the Chapter 4 plan. Five numbers or fewer. Every number needs an owner, a source, and a decision.

Fill in: Goal, Funnel stage, Review date.

• Lagging number The result you care about. Write its definition: who, how many, within how long, from where.
• Leading number 1 An early sign the result is coming. Definition, source, owner.
• Leading number 2 Optional. Definition, source, owner.
• Guard number A number that would catch a cheat or a side effect, such as spend per head or active rate.
• So-what For each number: “If it goes up or down, I will…” Write one line each.
• Retire when What would have to change in the goal before you drop each number?

MEASURE IT

The number: The So-What score: of the numbers you watch, the share that pass the test, with a written goal, a clear definition, and a decision attached.

By hand: List every number you check regularly. For each, write the sentence “If this goes up or down, I will…” Count the numbers where you could finish it. Divide by the total watched. Example: 4 of 14 = 29%. Aim to keep five or fewer, all passing.

Watch out: Don’t delete a number just because it’s inconvenient. If a number embarrasses you, ask what decision it should prompt before you drop it.

Go deeper: Chapter 23 (the baseline) and Chapter 34 (the scorecard and the review rhythm).

ROLE BOX

• Owner: Pick one funnel stage for the quarter and take one lagging number, one or two leading numbers, and one guard.
• Booker: Pair rebook rate (lagging) with replies to pitches (leading), and guard with settlement disputes.
• Musician: Pair rebooks (lagging) with replies and confirmed dates (leading). Guard with the pay you actually received.
• Bartender: Pair guests who visit because of your content (lagging) with sends per reach (leading). Guard with followers who visited.
• Dance Teacher: Pair students who reach the live floor (lagging) with the second-visit rate (leading). Guard with how many the first lesson turned away.

TALK ABOUT IT

• Which number do you check most often? What would you do differently if it went up or down?
• Which of your numbers is a vanity number? Why do you keep watching it?
• If you could only keep three numbers for the next ninety days, which three would you choose?