Upselling has a bad name because most of it is done badly. Done well it is not selling at all: it is offering something the guest would have wanted if they had known it existed. The difference between the two is entirely in what you know about the menu and when you say it.
Why the standard scripts fail
“Would you like fries with that?” works in fast food and nowhere else. In a restaurant it lands as a transaction, and guests who feel sold to tip worse and return less, which is the opposite of what the upsell was for.
The three scripts that reliably fail: offering an upgrade with no reason, offering the same thing to every table, and asking a yes-or-no question at the wrong moment. All three signal that you are working from a policy rather than from the table in front of you.
The manager-driven version fails hardest. Told to push the premium spirit on every drink order, a server produces a mechanical line that guests recognise instantly, and the restaurant gets a small number of upgrades plus a general sense of being handled.
What good upselling actually is
It is information, delivered at the moment it is useful, about something the guest cannot see on the menu.
A guest ordering a steak does not know that the peppercorn sauce is made in house and takes twenty minutes. Telling them is not selling, it is telling them something relevant.
A table ordering three mains does not know that the sides are large enough to share and that two would be plenty. Telling them costs you a side and buys you a table that trusts you.
That second example is the key to the whole thing. The most effective upselling includes occasionally selling less, because it establishes that your recommendations are about the meal rather than the bill. Guests then take your suggestions seriously for the rest of the service.
| Weak upsell | Strong version |
|---|---|
| ”Would you like a starter?" | "The bread comes out hot at six, if you want something while you decide" |
| "Premium spirit in that?" | "That cocktail is built for the rye, it is what the bar uses in the house version" |
| "Any sides with that?" | "The steak comes on its own, so people usually add the greens or the potatoes" |
| "Dessert menu?" | "The tart takes ten minutes, so I can put it in now if you want it” |
| Offering to every table identically | Offering what fits this table’s order |
The four moments that work
Upselling is a timing skill more than a language skill, and there are four moments where an offer is welcome rather than intrusive.
While they are still deciding. This is where a suggestion is genuinely helpful, because the guest has an open question. Anything offered here reads as assistance.
At the point where the menu is unclear. When a main comes without sides, or a dish is smaller than it sounds, saying so prevents a disappointment and adds an item at the same time. This is the highest-value moment and the most under-used.
When they mention an occasion. A birthday, an anniversary, a first visit. Guests marking something want the marking acknowledged, and a suggestion here is part of that rather than an intrusion.
At the transition to dessert, with a reason attached. “The soufflé needs fifteen minutes, so it has to go in with the mains” is information; “can I tempt you with dessert” is a script.
The moment that does not work: after the guest has finished ordering and closed the menu. That is a decision made, and reopening it reads as pressure.
What you need to know to do it well
The knowledge that powers good upselling is not the menu list, it is the layer underneath it.
What comes with what. Which mains include sides, which do not, and which sides are large. This single piece of knowledge prevents more complaints and adds more items than anything else, and it is the layer a fast menu-learning routine usually skips.
Preparation times. Which dishes are slow, which are fast, which have to be ordered early. Guests appreciate being told, and it converts.
What pairs with what. Not a sommelier’s answer, just two or three honest pairings per section that you can offer without a pause.
House versions and specials. What the kitchen makes rather than buys, what is new this week, what the bar is proud of. Daily specials are where this changes fastest.
Portion sizes relative to each other. So that “is that enough for two” has a real answer.
Those five fields fit on the back of a flashcard, which is roughly the point: short-term memory runs out at a handful of items, so the information has to be stored somewhere you can drill it. MenuFlashcards builds a deck from a photo of your menu, and the fields worth adding are exactly these, since they are what the upselling moments require.
| Knowledge | Upsell it enables |
|---|---|
| What comes with what | Sides, when a main arrives alone |
| Preparation times | Ordering a slow dessert early |
| Pairings | A drink alongside a specific dish |
| House versions | An upgrade with an actual reason |
| Relative portion size | Adding, or honestly removing, a dish |
The language that does not sound like selling
Three patterns cover almost everything.
State a fact and stop. “That one comes on its own.” The guest decides what to do with it, and most of the time they add a side without being asked.
Attach a reason. “The rye is what the bar builds that with” is persuasive because it explains rather than pushes, and it works for the same reason a one-sentence dish description does.
Offer a timing choice rather than a yes or no. “Do you want the dessert to come with the mains or after?” assumes nothing and gives a real choice.
What to avoid: superlatives you cannot support, the word “just” as a softener, and any question that is transparently a sales prompt.
When to sell less
Deliberately, and more often than feels comfortable.
Three mains for two people. Two starters when four were ordered. A cheaper wine that suits the food better. A guest who does not need the upgrade they asked about.
Each of these costs a small amount now and buys credibility that makes every later suggestion land. It also prevents the specific bad outcome of a table that over-ordered, ate half of it and felt embarrassed rather than looked after.
Managers occasionally dislike this and the numbers usually support it, because average spend per guest is a weaker measure than repeat visits and tips combined.
Practising it in ten minutes
Take your menu and mark every main that arrives without a side. That list is your highest-value upsell information and most servers have never written it down.
Then note the three slowest dishes and the three fastest. Those are your timing offers.
Then pick one honest pairing per menu section, something you would say to a friend.
Then rehearse three sentences out loud: a fact stated plainly, a reason attached to an upgrade, and a timing choice. Rehearsing by recall rather than by re-reading is the difference that research on the testing effect keeps finding. Three sentences, said naturally, cover most of a shift.
A deck with sides, timings and pairings on the back makes the first three steps a lookup rather than a research project, which is the practical reason to build one.
The numbers behind selling less
The argument for occasionally talking a table out of a dish is not sentimental, and it is worth being able to make it to a sceptical manager.
A pushed upsell adds one item to one bill, once. A guest who trusts your suggestions takes them for the rest of the meal, which is typically several items, and takes them again on the next visit. The second pattern compounds and the first does not.
There is also the cost side that rarely gets counted: over-ordering produces waste, comps, and the occasional removed item, all of which land on the same profit line the upsell was supposed to help.
None of this means never suggesting an addition. It means the suggestions have to be right often enough that guests keep listening, and the only way to be right is to know the menu well enough to judge fit.
Upselling as a team rather than a script
Most upselling failures are systemic rather than individual, and they get fixed at the pre-shift rather than by coaching servers one at a time.
Name the two things worth offering today, with the reason. This pairs naturally with answering what do you recommend, since both rely on the same underlying knowledge. Not a target, two specific items and why a guest would want them: the bread is out of a new oven, the special takes twenty minutes so it has to go in early.
Spacing those briefings out matters too, because spaced exposure produces better retention than one long session before a menu launch. Give the floor the timing information. Which dishes are slow tonight, which are close to selling out, what the kitchen wants moved before it turns. Servers who know this sell better without being asked to sell.
Say what not to push. A dish the kitchen is struggling with, a wine that is nearly gone, an item that is arriving inconsistently. Protecting the floor from selling something that will disappoint is as valuable as directing it.
And drop per-head targets, or at least stop announcing them. A server working to a number produces the mechanical line that guests recognise, and the number arrives anyway when the floor knows the menu.
What to do when the guest says no
Accept it in one beat and move on. The second attempt is where an offer becomes pressure, and it is the single most common mistake.
“No problem” and a step back is the whole technique. Guests who declined a starter and were not pushed frequently order dessert; guests who were asked twice generally do not.
The exception is genuinely new information. If the guest declined dessert and the soufflé needs fifteen minutes, saying so once is useful rather than pushy, because they were deciding without a fact that changes the decision.
Practising with an unfamiliar menu
New job, new menu, and the upselling knowledge takes longer to acquire than the dish names because it is not written on the menu.
Ask the kitchen three questions on your first shift: which mains come without sides, which dishes are slowest, and what the chef is proudest of this week. Those three answers cover most of the upsell surface.
Ask the bar two: what the house cocktail is built with, and which wines they would pour with the three most-ordered mains.
Write the answers on the back of your menu cards the same day, before they blur, then revisit them across the week rather than in one sitting, since distributed practice beats massed practice for exactly this kind of material. That is five questions and about ten minutes, and it is the fastest available route from new starter to someone whose suggestions guests take.
Frequently asked questions
How do I upsell without sounding pushy?
State facts rather than asking sales questions, attach a reason to any upgrade, and offer timing choices instead of yes-or-no prompts. “That one comes on its own” adds more sides than “would you like a side with that”, because it is information the guest can act on rather than a request they have to refuse.
When is the right moment to upsell?
While the guest is still deciding, at any point where the menu is unclear about what comes with a dish, when they mention an occasion, and at the dessert transition if you have a real reason such as preparation time. The moment that fails is after they have finished ordering and closed the menu.
Does upselling actually hurt tips?
Pushy upselling does, because guests notice being handled and it shows up as lower tips and fewer return visits rather than as complaints. Informative upselling has the opposite effect, since a guest who was told the main comes alone, or that three plates is too much, feels looked after and tips accordingly.
What should I know about the menu to upsell well?
Five things per dish: what comes with it, how long it takes, what pairs with it, whether there is a house version, and how big it is relative to other dishes. That is a different layer from the menu description and it is what every good upsell relies on. It also fits on the back of a flashcard.
Should I ever tell a guest they are ordering too much?
Yes, and more often than feels natural. Saying that three mains is too much for two people costs one dish and establishes that your suggestions are about the meal rather than the bill, which makes every later recommendation land. It also prevents the table that over-ordered and felt embarrassed.
