ApiaryActiveLive
Try: pause · settings · learn · wipe
← Community / Reading Room
TE
craft · 24 min read

The Empty Bin Is a Signal, Not a Stockout

A co-op keeps plain binder clips in a shared drawer. On Monday the front tray is empty. One person assumes the stationery lead reordered. Another finds a box…

AI disclosure: Drafted with AI assistance for Apiary using published pull and visual replenishment references. The shop, drawer contents, usage, delivery windows and outcomes described below are fictional. This isn't an inventory rule for critical materials.

A co-op keeps plain binder clips in a shared drawer. On Monday the front tray is empty. One person assumes the stationery lead reordered. Another finds a box in the back and quietly pours it into the front. By Wednesday all the clips are gone, and the first lead is blamed even though nobody actually gave them a request. A tidy divider would not have fixed that. To make the drawer useful, the team needs a precise transition: when the active stock reaches its agreed trigger, somebody changes which stock can be used and creates a single visible replenishment request. Its recipient confirms that it was received. A later observation establishes what actually arrived.

The back bin is time, not a secret second stash. Its job is to cover ordinary use through the response window while the next supply is obtained. If people pull from both bins whenever convenient, or someone says “we're covered” because an order was drafted, the system loses its ability to distinguish a usable reserve from an uncertain future delivery. A short pilot in a noncritical drawer can reveal that gap without buying software or seeking a perfect prediction.

Start with two states and one item people recognize

Choose a single benign consumable that moves often enough to observe and isn't critical for safe operation. The invented stationery drawer uses one specific size of binder clip for practice materials. The clips are not sold by weight, have no safety function in the example, and are easy to count in pairs or units. Do not group visually similar pieces under one bin label if the users cannot substitute them for each other. A medium and large clip might look close in a photograph and still be the wrong thing for the local use. Decide what one piece, reusable partial package, and sealed pack mean before attaching a label.

Check the actual quantity and identity in both bins. Put a short spec or reference description on the labels so next shift can distinguish the intended item. Treat mixed, bent, or unsuitable pieces as distinct from usable stock. A box in the drawer with the right color does not prove its label matches its contents, and an unmarked sleeve does not suddenly become the defined reserve. If you discover unidentified items, hold them out of the pilot quantity pending an ordinary check instead of filling the spreadsheet with guesses. The initial counted result is the starting evidence, not the count the team hoped they would see.

Mark the uses separately. The first bin is for ordinary draw. The reserve is protected until the chosen trigger. The simplest event for the fixture is that the last usable piece in the active bin is removed. At that time the user exposes a signal card and switches to the reserve under the local rule. Some systems use a low line instead of completely empty active stock. It can be a valid alternative if everyone can recognize the line and the quantity model accounts for which usable pieces are available after signaling. Avoid using an invisible definition of low that relies on the most experienced person always being there.

The Lean Enterprise Institute's kanban lexicon describes kanban as a signal device for authorizing or instructing work or withdrawal in a pull system. A paper card in a drawer can serve as a small local signal if it communicates a known action to people who can act on it. Toyota's description of its production system offers industrial context for how parts can be pulled forward in response to use. Neither system says that a desk with two containers has automatically become lean or that full scale industrial rules belong in a school supply room. Borrow the visible signal and responsibility, not a claim to have transplanted an entire production system.

Spell out the one cycle that replaces an empty bin

A drawer has several moments that can all look like replenished if nobody agreed on the words. Imagine the user takes the last from Active A, pulls out Marker M, and changes the current draw to B. The card next to the board now shows that the old active bin must be restored. The stationery lead accepts the signal, checks whether a packet is already inbound, and follows the agreed internal purchasing or stock transfer process. The requested amount and responsible role are clear. None of those actions puts a single physical clip back in the drawer until somebody receives something useful and verifies it.

For the pilot, record at least lower level signal, accepted by owner, supply arrival, checked units, and return to protected reserve. Add an explicit partial or mismatch status when the accepted supply isn't sufficient. This is a chain of evidence, not five ways to write order open. A button pressed in an inventory app might have the same logical meaning as moving a card, but it must still reach the right responsible person. Conversely a card physically lying on an unrelated workbench is not necessarily an accepted replenishment. Test exactly where a reminder becomes acknowledged work. Until that point, a missed handoff is distinct from the delivery actually being late.

Set a clear reset rule. After new usable stock arrives, one operational approach continues using whichever bin is active and restores the other as reserve, switching the labels only after verifying it has enough. Another uses a consistent front position and replaces the reserve behind it. Choose whichever is unambiguous. Do not let a newcomer move loose pieces between both bins because the equally spaced dividers look like one storage slot. Write the replenishment decision and who checked it on the record. A weekly review can then reconstruct how a box moved from received to usable reserve instead of assuming all material labeled delivered is available.

Suppose the accepted request corresponds to a pack of 20 but only 14 usable items are supplied. The pilot doesn't reset the order state to complete just because a carton arrived. Count the verified pieces, record the shortage and whether the remaining supply is expected, and leave the reserve short visibly until the agreed condition is met. If the local policy allows accepting and using partial stock, say which bin received it and what can be used. A partial refill is productive, but it does not buy the full amount of time. Keep written expected amounts separate from landed counts and note any unresolved identification rather than docking the inventory to fit a desired number.

Estimate reserve from an observed window, not an average alone

Use a modest synthetic history from the fictional drawer. Over eight ordinary working days, users consumed 4, 6, 3, 8, 5, 4, 7, and 2 clips. Say replenishment from a recognized internal supply room commonly takes two working days after somebody accepts the request, and acceptance can take another working day when the handoff arrives late. For a simple pilot, consider which working-day spans could happen before useful replenishment is actually checked into the drawer, including signal detection and intake. Do not conflate time quoted by a supplier with total time before usable local stock arrives.

The six observed three-day sums are 13, 17, 16, 17, 16, and 13. The largest is 17, reached twice: by the 6 + 3 + 8 and 8 + 5 + 4 windows. A single day's maximum of eight does not determine what the drawer needs across that three-day response window. A shop might try a protected 20 for this low-consequence pilot: the documented observed peak plus a small practical allowance with a whole existing pack. The 20 is a reversible test setting, not a confidence bound or promise. The eight days don't show unusual exhibits, closed supply rooms, different holiday shifts, or long-delay failures. Don't turn them into a business guarantee merely because the figure looks convenient.

Clarify whether usage was measured while the item was continually available. If Tuesday's count is low because the drawer was empty half the day, actual use understates attempted demand. The short history can still tell the truth if it labels periods with missing stock, observed substitutions, or unavailable counts. The user may only need a broad screen, but an appropriate exception note is better than calibrating reserve from an artificial quiet period. A hypothetical event day consuming 18 or 25 units is not evidence that the team actually had such a day; it can be added as a deliberate stress case and labeled as such. It may suggest a known extra reservation process without controlling every normal drawer.

Check where the working-day convention matters. Someone may expose the signal just after the internal supplier's cutoff, or the receiving clerk may be present only on certain afternoons. A reserve based on “two days” can be too small if it ignores those limits. A paper sequence can make them concrete. The hypothetical request appears late Thursday; the lead confirms it Friday; the supply room works Monday and Tuesday; intake is noticed Wednesday morning. The staff who use the drawer care about the period until Wednesday, not just the pair of days in a fulfillment quote. Include the actual period in your pilot assumptions or revise the handoff instead of adding stock indefinitely to hide recurring delays.

The arithmetic also depends on when the old active bin and reserve become available. When the trigger is front empty, the post-trigger stock initially equals the usable reserve if nobody has dipped into it. A low-line trigger can leave ordinary pieces in the active bin plus whatever verified reserve exists, but only if both are really eligible and the design expresses how to use them. A card uncovered once inactive stock is already depleted has a different effective trigger. Test the actual event and stock available immediately afterward. Otherwise a mathematical reserve setting can be blamed for an implementation whose physical supply never matched it.

Practice the shortage timeline before tuning the quantity

Use a documented fictional cycle with time anchors, not an accusation. On Monday morning the active bin empties when the remaining four pieces are used. Usable reserve at that point should be 20. The drawer user exposes the marker and takes another five clips later that day, leaving 15. If the responsible lead first sees the card Tuesday morning, note acceptance then. Tuesday use is eight, leaving seven. On Wednesday another six are used, leaving one. If the replenishment expected Wednesday turns into a Thursday arrival and two pieces are wanted first thing Thursday, the reserve is short by one for that request. The figures are small enough to inspect by reading the record.

Compare three alternative explanations. In the first, the marker was created late because an ordinary browser notice failed or a paper card went to the wrong place. A better signal path can shorten the interval and might solve the problem without buying more. In the second, the signal and acceptance were on time, but the actual supply room response took longer than usual. A calendar-aware buffer, a revised acknowledged handoff, or an approved alternate supply policy may be relevant. In the third, the actual reserve was less than 20 because someone borrowed five clips earlier without recording that the secondary supply was open. The remedy there may be enforcing a visible reserve change or revising access rules, not changing the arithmetic.

Write each version as a set of observed or hypothetical events. Do not retroactively call an event logged at 2 p.m. a signal sent at exactly 8 a.m. just because someone remembers opening the drawer in the morning. An interval or coarse description is sufficient when the exact time is unknown. Write usable received Wednesday afternoon only if a check established that material actually reached the drawer. A dispatch status isn't verification of quantity, local placement, or suitability. These distinctions may feel overcareful until a timeline proves the lead did everything they could and the package was sitting unchecked somewhere else.

An example review report can say “Only one verified item remained when two were needed. Card acknowledged Tuesday, valid stock not counted into the drawer until Thursday afternoon. We don't yet know why the delivery was delayed.” That tells a supervisor whether to investigate supply timing, regular event forecasts, or the handoff. A number called stockouts last week cannot by itself identify which boundary to change. In a low-risk pilot record what the user did to continue, such as use another approved binder or postpone a practice batch. That qualitative evidence helps judge whether the pilot is important enough to expand.

Make repeated signals safe without hiding requests

A recurring failure starts with good intentions. The first drawer user exposes the card but sees no acknowledgement, so another shift exposes a second one. Now the supply coordinator has two ordinary requests that appear to mean the same thing. In the fictional process, give a signal a known open identifier, even if it is just a dated slot on the board. A second user can point to the unresolved one, add a note, or escalate through the same path without silently making a duplicate purchase. Identify how genuinely distinct requests look too. Two bins emptied independently at different stations may need two replenishments, even when the labels and amounts match.

Try a paper rehearsal with a day and night shift. The morning lead already accepted a signal, but nobody marked it acknowledged. The night shift sees an empty active bin and a nearly full reserve. They should be able to find the open signal, tell whether somebody actually accepted it, and present any newly urgent context without pretending the first request never existed. A second exercise hides the old card entirely. Now the steward should record that signal state is unknown and reconcile manually according to the ordinary approved process, not imply that no order could possibly be pending. The important distinction is not avoiding all repeated communication. It is avoiding duplicate physical changes where their result matters.

Make corrections visible instead of overwriting history. Imagine a card identifies medium clips when large was documented on both the physical label and initial note. Correct the request before anyone reassigns a box. Keep the old identifier or the note linking it to the correction, and verify which request the supplier will actually fulfill under its process. An unattended corrected field in a secondary spreadsheet does not necessarily reach the person filling the cartons. For a benign internal pilot a discussion followed by a receipt is enough. The larger practice is tying a change to the people and objects it affects, not adding a brand-new purchasing system.

An alert can be small. Decide how long a visible unacknowledged signal should wait in ordinary hours before somebody checks it, and explicitly note any closed periods when no acknowledgement is expected. A one-line routine may use the staffed board rather than a text message, depending on how local people actually work. Don't solve a weak process by flooding everybody with duplicate warnings. Test a handoff at a shift change and a weekend closure with wholly invented cards. If an alert makes the recipient mark the item ordered just to dismiss it, the problem may be that the UI has no honest receipt for “I saw the signal and will review it at the required time.”

There is no need to track people more closely than the work requires. The record can use a responsible role, receipt number, or known location with one authorized escalation. The point is to know when the next action became somebody's responsibility, not to reconstruct an employee's whole day. Public training examples can omit names entirely. Shortage data that plainly identifies why stock was unavailable lets the team improve the workflow without turning a low-risk practice drawer into a disciplinary ledger.

Watch counting and relabeling from the receiver's side

Suppose a storage lead replaces two worn paper labels during a shift. The green tape traditionally means currently active, but new tape is blue and someone interprets it as reserve. A refill occurs while labels are being updated. Treat this as a setup change with its own small receipt. First identify the current draw bin, the verified reserve balance, and any accepted pending signal. Then produce labels that preserve those meanings and test them from two approach angles. A stock movement that happened during relabeling shouldn't disappear simply because both bins look neater afterward. Check whether the change displaced the signal or accidentally marked a partial refill ready.

Use a label test with an ordinary person who didn't design the drawer. Ask them where to take one piece this afternoon, when to stop and expose a signal, and what the marked pack count actually means. Then ask a receiving clerk what evidence is needed to mark the empty side restored. If the first reader thinks reserve means available overflow or the second thinks delivery means invoice sent, revise the labels. Familiar stationery colors can help recognition, but action words and a clear position should also work when color itself isn't distinguishable. The sign doesn't need formal jargon. It needs to explain what should happen before someone makes a silent substitution.

Count the same units the users take. A box of twenty and twenty individual clips are equivalent only when the labeled box is known to contain the right usable pieces. A sealed mislabeled carton, an open half-used pack, and a case containing several smaller packs need separate handling. In a synthetic trial a receipt says one unit was supplied; the sender meant one carton, while the clerk meant one clip. The quantity field should carry its unit and packaging relationship or remain unresolved. Avoid treating the quantity as a scalar that loses the package meaning between sending and receiving. This matters before any reorder formula is applied.

The process can be deliberately tactile. Place a printed card where it becomes visible naturally as the last few active pieces are drawn; define whether visibility alone is the signal or moving the card is required. Give an opener a known way to signal without dumping clips from both bins into one pile. Give a receiver a place to mark held or partial stock without moving good supply out of reach. A cardboard mockup of the paper flags and event stages is enough to test comprehension; avoid physical changes to an operational workshop beyond the authorized pilot. The best label is the one the novice interprets correctly under ordinary hurry.

Trace a change in demand without attributing every shortage to growth

One fictional week with an exhibit can use more clips than four ordinary weeks. If the exhibit is known in advance, a reserve designed for normal draw need not itself absorb every large scheduled pull. The station can offer a separate planned-supply request or assign extra approved stock for the event. Record that decision independently from the test of the routine signal. Otherwise scheduled exceptional use might make every normal setting look insufficient. Conversely if exceptional use has become ordinary and predictable, don't keep excluding those periods merely to present a reassuring profile. Your stated operating scope should match the work readers recognize.

A stocking policy can help when work is interrupted by seasonal schedule change. The provider might ordinarily refill after two business days; during events, the queue behaves differently or the intake clerk receives boxes a day later. A modest test might distinguish normal calendar, scheduled exception, and disruptions whose reason is still unknown. Each has a different fitting response. If staff cannot identify the usual wait from any observations, mark it as an assumption instead of silently treating a verbal estimate as measured performance. A useful historical receipt labels who or what confirmed each boundary without collecting irrelevant private details.

Try comparing two shortages with identical counts but different histories. In the first, protected stock began at 20 and was consumed as intended; an unusual last-minute draw reached 25 before the scheduled refill. In the second, the usual use was only 12 but reserve began with eight verified clips because an unrecognized early draw had already entered it. Raising the target by five may appear to solve both cases in a spreadsheet. It does not fix reserve erosion in the second. An investigator can ask how much shortage came from nominal policy, counted starting stock, timing, and a mismatch in demand assumptions without forcing one arbitrary explanation to cover the whole week.

If the pilot records missed use, decide which data are truly observable. An empty bin with two people waiting might indicate two tasks potentially needing stock, not a measured requirement of three clips per task. A signed inventory note could count twelve drawn before the shortage and state that further demand was not measured. An approved substitute or rescheduled task can be recorded qualitatively. Keeping a coarse unfulfilled-use note is more honest than marking a zero and calling demand surprisingly low. Distinguish forecast, estimate, observed withdrawal and usable receipt in a chart. They are not interchangeable series merely because they use the same unit.

Another useful exercise keeps the bins constant and moves only one process boundary. Compare notice clearly acknowledged by midday with notice not acknowledged until the next shift under the old process. Hold expected receiving time conventions explicit and run the same hypothetical demand sequence. The reserve may last under one policy and not the other. That doesn't establish a live improvement until observed, but it helps decide whether to pilot a better response rather than buying far more stock. NIST MEP's lean and process improvement overview discusses visual inventory signals and disciplined continuous improvement. The lesson for this drawer is that a signal becomes valuable when people can observe its failure and change one cause, not that every local issue needs an industrial transformation.

Keep an ordinary recovery rule ready before something goes missing

A pilot works better when it says what to do if the reserve isn't enough. For the benign binder-clip case, staff can switch to a previously accepted alternative, delay an optional activity, or report which item is unavailable. They should not count unauthorized material from another department, make a purchase beyond their authority, or mislabel incompatible pieces to keep the station looking full. Keep material limits and the written request flow in scope. This article does not say to reduce protective or critical inventory or apply the toy demand arithmetic to emergency materials, regulated stock, or anything whose failure would cause harm.

Create a short practice table. If no one has accepted the signal, contact the responsible role through the known process. If it was accepted and the arrival date changed, update expected arrival without claiming physical stock came. If only part is actually usable, record receipt and remaining concern separately. If the contents of the drawer are unknown, count the eligible material before relying on an apparent reserve. If there is no useful observation, record unverified rather than promising a precise date or count. Simple conditional responses prevent an alert from automatically leading to an expensive solution. A failure model can be specific about the unknowns without guessing a culprit.

Consider substitutions carefully even for low-risk work. The smaller clips may be entirely fine for short handouts, but not a sensible alternative for larger packets. That makes approved substitution a scoped decision, not arithmetic that one clip counts as one clip across all uses. When listing usable stock keep the categories separate unless they are genuinely interchangeable under an authorized rule. A special replacement shipment might only replenish one class. If the team needs many different classes, use a separate item record or choose a simpler alternative methodology; squeezing them into one queue can make every warning ambiguous.

One scenario deserves explicit attention: material marked accepted arrives and its paper record is lost before it reaches the drawer. Someone can physically count what is present, but that may not tell them which delivery it came from or whether the requisition remains open. Restore only the facts that evidence supports. The valid response may be “drawer contains twelve checked clips; closure of request unknown.” Avoid making the reconciliation easier by marking all accepted requests complete or going on blind additional orders. A low-cost inventory system can tolerate ambiguity if its records say it exists. It starts failing when every process demands a tidy, invented result before the day can close.

Review a couple of cycles, then decide whether to keep the system

Plan a bounded pilot on one SKU under an agreed area owner. Before it starts, capture measured quantities, intended refill window, the event that reveals the card, and any known high-use closures or other exceptions. Declare the review bar. A reasonable small experiment might wait for a few complete shipping or internal transfer cycles or a fixed short time, whichever provides useful evidence for the specific outlet. If no cycle occurred, say that. You can still evaluate whether people understood the labels; you cannot infer that replenishment works flawlessly just because the reserve was never touched.

Count useful intervals separately. Time from signal to acknowledgement reflects one boundary. Time from acknowledgement to usable receipt reflects another and may include supplier and intake substeps documented in larger systems. Percentage of received usable against requested measures delivery agreement, not how quickly people moved a card. Number of shortage requests reflects some use impact, but only within the recorded scope. For a pilot use the raw dates and actions where a summary might mislead. A table with three verified cycles teaches more than a chart with unstated assumptions about all days when no one wrote a note.

Invite the receiver, drawer user and responsible lead to explain one ordinary and one failed cycle from the same artifact. If the lead sees a purchase closure while the drawer user sees only empty slots, the record didn't reach the intended use. If the user sees a signal open two days later and the receiver can show a valid replenishment entered earlier, find out which state never got updated. Fix the route of information before deciding that the amount of reserve is always the problem. The Lean Enterprise Institute's account of pull production emphasizes downstream information about needs. The simple relationship is useful to a small shop because feedback from actual use is required for the next appropriate action.

Avoid treating visual improvements as service outcomes. Clear labels are promising, but they are not evidence that delayed supply reached the drawer. A neat photograph taken at the start verifies that two bins and a card exist under the selected arrangement. It says nothing about whether the emergency scenario, known closure, or partial delivery was handled. The training packet can truthfully say the usability check worked while live cycles remain unapparent or too few to claim change. No independent outcome exists merely because the board looks like a familiar example in a guide. Earn stronger claims with records that actually support them.

At the end compare the pilot against the easiest alternative. If the drawer item hardly moves, periodic counted restocking may be less burdensome than a visual trigger no one uses. If purchasing is centralized and the manual signal reliably goes unseen, a clearly recognized internal intake route might help more than new containers. If the actual problem was an incompatible item sold in variable quantities, fix the identity and receiving check first. Designs can change or retire after a short trial without blaming the people who tested them. The job is to keep ordinary work supplied, not to protect the picture of the solution you started with.

Leave a small receipt the next shift can trust

In the fictional successful cycle, a colleague counts twenty verified reserve pieces, uses the active bin last, exposes the right card, sees its acceptance, and eventually verifies the actual replacement. In the failed cycle, only one usable item remains for a two-piece need, and the clerk knows when the gap was observed without claiming an unsupported reason. Those experiences are different enough to require different lines, but both are evidence the co-op can learn from. Preserve the first simple method that records them clearly. If a second label or form does not help a real decision, drop it rather than making people do a fine imitation of tracking.

For a handoff, display current draw location, verified reserve when relevant, signal status, known arrival state, open exceptions and who owns the next step. Avoid an estimated date presented as received or a box label presented as physically verified quantity. A day later the same reader should be able to understand how late evidence changed the status. The careful details worth retaining are ones somebody would need to locate a weak boundary. Once the system makes that boundary observable, it has achieved more than an empty compartment with an attractive label.

A final communication test asks two people to read the word empty. One should interpret the empty active bin as the planned moment to change what stock is available and signal the next refill. They should recognize an empty available reserve while work requires the item as a different event, carrying more direct consequence. If the labels and receipts reliably expose that difference, this tiny inventory pilot is doing its important work. It is respecting what was actually received, used, and requested instead of converting uncertainty into a neat false count.

Check whether two partly empty bins can still serve the promise

Observe a late pilot cycle in which the active bin contains four pieces and the restored reserve contains only six. The supplier's system has closed the transaction because a parcel was delivered, but the drawer's ordinary reserve target is twenty. A clerk can acknowledge that six usable pieces arrived while recording reserve not yet restored to policy. Use gate labels that distinguish the supplier transaction, visible quantity, and local stock decision. The tool's own replacement milestone need not control every other receipt just because it is easier to export.

A supervisor might allow a smaller reserve temporarily for a pre-announced quiet shift. That is an explicit exception, with a time window and a known fallback if use is higher. It does not prove the original reserve was supplied and shouldn't be rolled into an ordinary success rate. An accepted temporary condition can be appropriate where consequences are small and options are known. For the pencil-and-paper exercise, ask somebody resuming work the following morning whether the exception is still in force and what stock they can actually see. That is the moment a vague late receipt would otherwise turn into a surprise shortage.

Primary sources

AI disclosure: Drafted with AI assistance for Apiary. Usage history and timelines are invented examples, not records of a measured pilot.

Frequently asked
What is The Empty Bin Is a Signal, Not a Stockout about?
A co-op keeps plain binder clips in a shared drawer. On Monday the front tray is empty. One person assumes the stationery lead reordered. Another finds a box…
What should you know about start with two states and one item people recognize?
Choose a single benign consumable that moves often enough to observe and isn't critical for safe operation. The invented stationery drawer uses one specific size of binder clip for practice materials. The clips are not sold by weight, have no safety function in the example, and are easy to count in pairs or units. Do…
What should you know about spell out the one cycle that replaces an empty bin?
A drawer has several moments that can all look like replenished if nobody agreed on the words. Imagine the user takes the last from Active A, pulls out Marker M, and changes the current draw to B. The card next to the board now shows that the old active bin must be restored. The stationery lead accepts the signal,…
What should you know about estimate reserve from an observed window, not an average alone?
Use a modest synthetic history from the fictional drawer. Over eight ordinary working days, users consumed 4, 6, 3, 8, 5, 4, 7, and 2 clips. Say replenishment from a recognized internal supply room commonly takes two working days after somebody accepts the request, and acceptance can take another working day when the…
What should you know about practice the shortage timeline before tuning the quantity?
Use a documented fictional cycle with time anchors, not an accusation. On Monday morning the active bin empties when the remaining four pieces are used. Usable reserve at that point should be 20. The drawer user exposes the marker and takes another five clips later that day, leaving 15. If the responsible lead first…
References & sources
  1. Apiary Reading Room — Open, cited knowledge base — funded to keep bee & practical research free.
From the Apiary Reading Room. Opinion & editorial — not financial advice. We don't overclaim.
More from the Reading Room