How Do I Get My QSR Crew to Upsize and Attach Items?

Signals you actually need this

Most operators don't discover an attach problem — they discover a margin problem and trace it backward. The tell is a store where transaction count is healthy or even growing while average check sits flat month over month. If you rang 900 tickets in April and 940 in May but the check moved from $9.10 to $9.14, your crew is processing orders, not selling them. Traffic is doing the work; nobody at the window is.

A second signal shows up in the spread between crew members. Pull per-cashier attach rate off your POS for a two-week window and sort it. In a store with no scorecard, the gap between your best and worst attach performer is routinely three to five times — one person attaches a drink or a side on 35% of tickets, another on 8%. That gap isn't talent. It's that nobody ever told the second person that attach was part of the job, or measured it, or paid for it. When the spread is that wide, a scorecard closes more dollars than any new menu item you could launch.

Third: your limited-time offer launches land soft. Corporate or your franchisor drops a new dessert or a value bundle with POS signage and a pre-shift huddle, and two weeks later the mix report shows it at 3% of tickets. That's the classic symptom of a crew with no re-aimable target. Without a weighted matrix, "push the new thing" is just words in a huddle that evaporate by second rush. With one, you raise the weight on that attach line overnight and the whole crew re-aims on the next shift because their composite moves with it.

Fourth signal, and the one operators miss: drive-thru speed metrics are excellent and getting better while check is deteriorating. That's the matrix working against you. If the only thing you time, post, and reward is service speed, you've built a machine that optimizes for shaving four seconds — and the fastest way to shave four seconds is to stop asking about the large fry. You didn't get a lazy crew. You got a crew that read your incentive correctly.

Finally, look at your app and loyalty signup numbers per shift. Signups are structurally the same behavior as attach — a five-second ask at the point of transaction that nobody does unless it's counted. If signups are near zero, attach is almost certainly weak too, and both fix with the same instrument.

What good looks like versus what bad looks like

Bad looks like a single number taped to the wall. A drive-thru timer board, a daily transaction count, maybe a "check average" figure for the store with no per-person attribution. There's no line item for upsize, no line item for attach, and no way for a crew member to know whether they're doing well beyond a manager's mood. The bonus, if there is one, is a store-level number split evenly — which means the person who attaches on 35% of tickets and the person who attaches on 8% earn identical incentive. That is a direct tax on your best crew member, and they know it.

Good looks like a published matrix with eight or nine explicit lines: size upsize rate, combo conversion, add-on attach, dessert attach, drink attach, app or loyalty signup, average ticket, and service speed. Each line carries a weight your shift managers set together. Each crew member carries a 1-to-5 level on each line. The composite is the sum of weight times level across every KPI, and it's the only number that determines the shift bonus. A crew member who is a level 5 on speed and a level 1 on attach lands a low composite and can see exactly which line is dragging them.

The behavioral difference is that in the bad version, coaching is vague — "try to upsell more." In the good version, coaching is a specific, small, repeatable move: "your drink attach is a 2, your food attach is a 4; for the next three shifts, name a specific drink instead of asking 'anything to drink?'" That's coachable in thirty seconds at the window and the effect shows up in the next scorecard pull.

The other difference is durability. A pep talk decays in about two days. A visible number that determines pay does not decay, because every shift regenerates the signal. This is the same mechanic RevOps teams use on sales floors — weighted scorecards, published attainment, comp tied to a composite rather than one metric — and it transfers to a QSR counter almost unchanged, because the underlying problem is identical: a frontline human deciding, dozens of times an hour, whether to make an ask.

Building the matrix: KPIs, weights, and levels

Start by listing every behavior a complete crew member should produce, not just the ones that are easy to time. The working list for most QSR counters is: size upsize rate, combo conversion rate, add-on attach (fries, sides), drink attach, dessert attach, app or loyalty signup rate, average ticket, and service speed. Some operators add order accuracy and cash variance. If a behavior isn't on the list with its own weight, your crew will correctly infer it doesn't count.

Weights are where you encode strategy. A common starting split for a store trying to grow check: upsize 20%, add-on attach 20%, drink attach 15%, combo conversion 10%, dessert attach 10%, loyalty signup 10%, average ticket 10%, service speed 5%. Those aren't magic numbers — they're a statement that this store is currently buying check growth with a small amount of speed. A store fighting a drive-thru time problem would invert several of those. The point is that the weights are yours, they're explicit, and they're changeable.

Levels need definitions, not vibes. A 1-to-5 level on drink attach should map to actual rate bands pulled from your own POS baseline — for example, level 1 under 10%, level 2 at 10-17%, level 3 at 18-25%, level 4 at 26-33%, level 5 above 33%. Set the bands off your own two-week baseline so a level 3 is roughly your current median performer. If you set the bands off some industry ideal, everyone lands at level 1 on day one and the scorecard reads as punishment instead of a ladder.

Score on a fixed cadence — weekly is the sweet spot. Daily is noisy and creates whipsaw coaching off a single bad rush. Monthly is too slow to feel connected to behavior. Weekly gives you roughly 300-800 tickets per crew member depending on hours, which is enough volume that the rate isn't random.

Publish it. This is not optional and it's the step operators skip because it feels confrontational. If a crew member can't see their own levels and the gap to the next one, the matrix is a management report, not a motivator. Post it in the crew room, in the scheduling app, wherever the team already looks. Show levels, not raw dollars, if you want to keep it non-punitive — the level is the ladder rung, and the only move that matters is climbing one.

One caution: normalize for shift context before you compare people. A closer working a slow Tuesday night and an opener on a Saturday lunch rush face different attach opportunity. The cheap fix is to compare within daypart. The better fix, if your scheduling data supports it, is attach per labor hour, which strips out both traffic volume and shift length.

Real cost and ROI ranges

The instrument itself is cheap or free. A well-built spreadsheet costs you a few hours to construct and a standing fifteen minutes a week to update off a POS export — free in dollars, real in time, and it carries the risk that it goes stale the first busy week. Purpose-built scorecard and gamification platforms are the paid tier; recognition-and-leaderboard tools commonly run in the low-to-mid tens of dollars per user per month, and coaching-plus-scorecard platforms typically price by quote. Commission and attainment tools that wire a composite to actual pay usually have a free tier and paid plans in a similar per-user range. Your POS — Toast, Square for Restaurants, Loyverse, whatever you run — already captures the per-cashier attach and ticket inputs the matrix consumes; you're not buying data, you're buying the weighting and visibility layer on top of it.

The return math is where this gets interesting, and it's driven by one fact: attach and upsize grow revenue on traffic you already paid for. There's no incremental marketing spend, no additional labor hour, no new equipment. A twenty-cent lift in average check is close to pure incremental margin after food cost.

Run the arithmetic on a single store. At 800 tickets a day and a $9.00 average check, you're doing roughly $7,200 in daily sales. Move the check by $0.25 — one additional small side attached on roughly one ticket in eight — and you add $200 a day, about $6,000 a month, roughly $72,000 a year at one location. Even if you only capture half that, the scorecard has paid for any tool on the market several hundred times over. For a five-unit franchise group, the same lift compounds to a number that shows up in the P&L as a line item, not a rounding error.

Cost the incentive honestly. If you fund the crew bonus at, say, 15-20% of the incremental check lift, the math still works overwhelmingly in your favor, and the crew sees a real dollar tied to a real behavior. A bonus that's too small to notice produces exactly the behavior change you'd expect from a bonus too small to notice. This is the same failure mode RevOps sees in underfunded sales comp plans — the quota moves, the pay doesn't, and behavior stays put.

The hidden cost is management attention, and it's the one that actually kills these programs. Budget fifteen to thirty minutes a week for the scorecard pull and posting, plus roughly two minutes of targeted coaching per crew member per week. If your shift managers can't protect that time, buy the tool that automates the pull off the POS rather than running the spreadsheet — the automation isn't a luxury, it's the difference between a program that survives month three and one that doesn't.

Watch two downside risks. First, an over-aggressive attach push can degrade guest experience if crew start stacking multiple asks per order; cap it at one well-chosen ask per ticket. Second, if you weight speed to zero, drive-thru times will drift. Keep speed on the matrix at a small weight specifically so it stays a constraint rather than disappearing.

How it plugs into the shift and the wider operation

The scorecard lives inside a loop you already run: POS captures transactions, you export per-employee data, you score, you post, you coach at pre-shift, the crew executes, and the POS captures the next round. Nothing about that is exotic — the only new step is the scoring and the posting, and those take minutes once the template exists.

Pre-shift is where the matrix converts to behavior. A generic huddle produces generic results. A matrix-driven huddle sounds like: "team drink attach is a 3, the store target this week is a 4, so the ask today is name a specific drink instead of asking whether they want one." One line, one behavior, one shift. Then it shows up in Friday's pull, and either the level moved or the coaching needs a different angle.

The LTO loop is the highest-leverage use of a weighted matrix and the reason to build it before you need it. A new combo or limited-time dessert drops, you raise the weight on that attach line the night before launch, the composite re-ranks overnight, and the crew re-aims without a single confusing conversation about priorities. When the LTO ends, the weights go back. That responsiveness — pivot the target overnight, whole crew follows — is impossible with a bonus wired to order count and trivially easy with a bonus wired to a composite you control.

The scorecard also feeds decisions upstream of the register. Scheduling gets better when you know which crew members carry high attach levels, because you can weight your strongest sellers into your highest-traffic dayparts — the same logic a retail chain uses to staff its best closers into Saturday afternoons. Hiring gets a real signal, since a new crew member's attach curve over their first month tells you more about fit than any interview did. Training gets targeted, because you stop running blanket upsell modules and start running a two-minute drill for the four people whose drink attach is a 2. And your franchisor's program compliance stops being a separate reporting exercise — if their push is a weight on your matrix, compliance is a byproduct of the number your crew already chases.

There's a neighboring pattern worth stealing. Convenience stores, car washes with membership attach, and quick-service coffee run essentially this same play — a frontline human, a few seconds of decision, a small ask that compounds across enormous volume. The mechanics port cleanly: define the ask, count it per person, weight it, publish it, pay for it. Anywhere the transaction is fast and the attach is optional, the weighted composite beats the single number.

Common failure modes and how to avoid them

The most common failure is scoring what's easy instead of what matters. Service speed is trivially measurable, so it becomes the whole scorecard by default. Resist that. If a KPI is important and hard to measure, approximate it rather than dropping it — a manager-observed 1-to-5 on "makes a specific ask" is imperfect but far better than an absent line.

Second failure: weights that never change. A matrix set once in January and untouched through December is just a slower version of the single-number board. The weights are a steering wheel; if you never turn it, you didn't need one. Review them monthly and hard-reset them for every LTO.

Third: scoring in private. Managers who keep the matrix in a spreadsheet on their own laptop get a reporting tool and zero behavior change. The publishing step is the mechanism, not the paperwork.

Fourth: bonus disconnected from the composite. If you build a beautiful weighted scorecard and then pay a flat store-level bonus, crew will optimize for whatever their manager actually praises out loud, which is usually speed. Wire the pay to the number or don't build the number.

Fifth: no normalization. Comparing a Saturday-lunch opener to a Tuesday-night closer on raw attach rate produces a scorecard the crew correctly perceives as unfair, and an unfair scorecard gets ignored within two weeks. Compare within daypart, or normalize per labor hour.

Sixth, and subtlest: over-indexing on one line until it distorts the guest experience. If dessert attach is weighted so heavily that every guest gets asked about a dessert regardless of what they ordered, you'll see it in the attach rate climbing while repeat visits soften. One good ask per ticket, matched to what's already on the order, outperforms a scripted stack every time.

Related questions

How long before a new scorecard shows up in average check?

Expect a visible move in three to four weeks. The first week is awareness, the second is behavior, and by the third and fourth you have enough ticket volume per crew member for the rate change to separate from noise. If nothing has moved by week six, the bonus isn't connected to the composite.

Should I score part-time crew the same way?

Yes on the same KPIs and weights, but normalize the comparison — per labor hour or within daypart. A twelve-hour-a-week crew member scored on raw volume will always look weak. Levels should reflect rate, not total transactions, which keeps the ladder fair across schedules.

Does this work in drive-thru as well as at the counter?

Yes, with one adjustment: keep service speed on the matrix at a small but non-zero weight so times don't drift while attach climbs. Drive-thru attach asks need to be shorter and more specific — name one item, not a menu tour.

How many KPIs is too many?

Above nine or ten lines the matrix stops being legible to crew and starts being a report. If a line's weight rounds to under 5%, either raise it or cut it. Every KPI on the board should be something a crew member could change on their next shift.

Can I use this to compare stores, not just people?

Yes — roll each store's crew composites into a store-level average and compare across units. It surfaces whether a weak store has a training problem (broad low levels) or a staffing problem (one or two crew dragging an otherwise strong bench).

FAQ

What exactly is a weighted multi-KPI scorecard?

It scores each crew member on several behaviors — upsize, attach, loyalty signup, speed — each carrying its own weight and a 1-to-5 level. The composite is the sum of weight times level across all lines. A fast crew member who never makes an ask lands a low composite, which is the entire point: one strong metric can't hide a weak one.

How do I set the weights without guessing?

Set them with your shift managers against what the store is actually trying to buy this quarter. Growing check means upsize and attach carry the most weight; fighting drive-thru times means speed rises. Start from your own POS baseline, write the weights down, and revisit monthly rather than treating the first draft as permanent.

Won't pushing attach slow my crew down during a rush?

A single well-chosen ask costs a few seconds, not a few minutes. Keep service speed on the matrix at a small weight so it stays a live constraint, and cap the ask at one per ticket. The failure mode isn't a slow crew — it's a crew stacking three asks on one order.

How do I change the matrix for a limited-time offer?

Raise the weight on that attach line the night before launch and drop the others proportionally so the weights still sum to 100%. The composite re-ranks overnight, the pre-shift huddle names the one behavior, and the crew re-aims on the next shift. Restore the original weights when the LTO ends.

Do I need to buy software, or is a spreadsheet enough?

A spreadsheet is genuinely enough for a single store — list the KPIs, set the weights, score 1-to-5, roll the composite off a weekly POS export. Multi-unit groups usually outgrow it, because the weekly pull across several locations is where the manual version dies. Buy automation when the upkeep, not the math, becomes the bottleneck.

Does the same approach cover things beyond upsize and attach?

Yes. Any countable frontline behavior can be a weighted line — order accuracy, cash variance, app signups, prep-station adherence. The instrument is indifferent to which behaviors you pick; what makes it work is that each one is explicit, weighted, visible to the person being scored, and connected to what they're paid.

Sources

• https://www.toasttab.com/restaurant-guide/restaurant-metrics
• https://squareup.com/us/en/restaurants
• https://www.7shifts.com/blog/restaurant-labor-cost/
• https://www.nrn.com/
• https://restaurantbusinessonline.com/
• https://www.qsrmagazine.com/
• https://hbr.org/2016/04/the-performance-management-revolution
• https://www.gallup.com/workplace/236927/employee-engagement-drives-growth.aspx
• https://www.bls.gov/oes/current/oes353023.htm
• https://loyverse.com/

Related on PULSE

• [How Do I Know Where, When, and How Many People to Schedule at Each of My Multi-Unit Retail Locations?](/knowledge/tl0001)
• [How Do I Figure Out How Many People to Schedule Each Day and at What Times for My Single Store?](/knowledge/tl0002)
• [How Do I Know How Many Cooks and Servers to Schedule Each Shift at My Pizza Restaurant?](/knowledge/tl0003)
• [How Many Salespeople Should I Schedule Each Day on My Furniture Store Floor?](/knowledge/tl0004)
• [How Do I Decide How Many Reps to Schedule at Each Store in My Mattress Retail Chain?](/knowledge/tl0005)