Checklist

Direct Ordering Conversion Checklist

The steps where guests give up on an online order, and how to fix each one.

By Appetier team2026-10-015 min

Place a test order on your own phone

Order from your own site on a phone, over cellular data, with one hand. That's how many guests order: in a hurry, standing somewhere, on a slow connection.

Count the taps from opening the menu to a paid order. Every tap you remove is one less place for a guest to quit.

Run three carts, not one. Order a single item, then an item with two or three modifiers, then a four-person order with a drink each. The group order is where long modifier lists and a cramped cart start to hurt.

Test pickup and delivery separately, and test once near closing time. A guest who orders ten minutes before close should see an honest answer, not a pickup time after you've locked up.

Write down every tap and every wait longer than a couple of seconds. That list is your to-do list for the rest of this checklist.

The table below is the whole checklist in one place. Each row has its own section further down.

Direct ordering checklist at a glance
CheckPass when
Test orderYou placed one on a phone, over cellular data, and wrote down every tap and wait
Biggest dropYou know which of the four steps loses the most guests
Menu and cartThe menu loads fast, best sellers have photos, and prices match your POS
Time and totalThe ready time and every fee show before the payment step
Payment methodsA test payment went through with each method you offer
Add-onsThe cart shows one or two suggestions that fit it
Second orderA returning guest can repeat an order in one tap
ChangesYou change one thing at a time and measure it before the next
Broken ordersA test order prints in the kitchen

Find the step where guests leave

Every ordering site loses guests in four places: the menu, the cart, checkout, and payment. Pull a week of numbers for each step and see which one bleeds the most.

Here's a worked example. Say 500 guests open the menu in a week, 150 add something to the cart, 90 start checkout, and 60 pay. That's 30 percent adding to the cart, 60 percent of carts reaching checkout, and two in three checkouts paying.

Overall, 60 of 500 is 12 percent. The biggest loss is at the menu, where 350 of the 500 leave without adding anything.

Now improve only the menu step, from 30 to 35 percent. That's 175 carts instead of 150. At the same rates further down, 25 more carts become 15 more checkouts and 10 more paid orders.

At a $25 average ticket, those 10 orders are $250 a week from one fix. Start where the drop is largest, then move down the list.

No step data yet? Use your own test order instead. The screen where you hesitated or scrolled the longest is usually the screen guests leave from.

The table below lays out the worked example step by step.

The worked example, this week and with the menu step at 35 percent
StepGuests this weekShare of the step beforeGuests with the menu step at 35 percent
Opened the menu500Where the count starts500
Added to cart15030 percent175
Started checkout9060 percent105
Paid602 in 370

Check the menu and cart

The menu loads before you'd give up on it, even on a weak signal.

Your best sellers have real photos, near the top.

Item names say what the dish is. "The Hangover" means nothing to a new guest until the description says it's a breakfast burrito.

Descriptions fit in two lines on a phone and name the main ingredients.

Allergens are listed on the item, so a guest doesn't have to call to ask.

Prices match your POS, and any fees show before the last step.

Modifier lists are short, with the usual choice already selected.

Sold-out items are hidden or marked, so nobody builds a cart they can't buy.

Show the time and the total before the card goes in

Guests abandon at payment when something new shows up there. A delivery fee, a minimum, or a pickup time 50 minutes out all belong earlier.

Put the ready time on the cart screen. If your kitchen takes 15 minutes on a quiet night and 35 on a Friday, the time should say so.

Appetier's checkout shows a pickup or delivery time measured from your kitchen's real prep times. Delivery shows a time once the location sets its travel time.

If a guest filled a cart before picking a location, keep the cart. Appetier flags anything that location can't sell rather than dropping it, so the guest swaps one item instead of starting over.

Take the payment methods your guests already use

A guest who reaches payment has already decided to buy. Losing them because their usual way to pay is missing hurts more than losing them at the menu.

Card is the baseline. Take cards on Appetier Pay at 2.9% + 30¢ online, or keep any PCI compliant processor you already use.

Venmo and PayPal are there for operators who already take payment that way. Venmo becomes a checkout button once you connect a PayPal Business account.

Place a test payment with each method you offer. Then check that the order shows how it was paid, so the counter knows which tickets are settled.

Suggest add-ons that fit the cart

An add-on that matches the cart gets taken: a drink with a burger, a side with a bowl. A fixed list of promoted items gets ignored, and guests learn to scroll past it.

The math is small per order and adds up over a week. If one in five of 60 weekly orders adds a $3 drink, that's 12 drinks and $36. Over a year, it's about $1,870.

Show one or two suggestions, not six. A long row of extras pushes the checkout button below the fold on a phone.

Guests on an Appetier storefront can also ask what to order and get items with prices. An item goes in the cart only when the guest taps Add, and it never checks out for them.

Make the second order one tap

Saved card, saved address, favorites, and a reorder button. The guest who gets the same thing every Thursday should be able to repeat it in one tap.

Keep guest checkout for first timers. Offer to save their details after the order, once they have a reason to say yes.

On Appetier, signed-in guests reorder from their last 50 orders and keep favorites. Sign-in is a link by email, good for 30 days at one brand, so there's no password to forget.

Change one thing at a time and measure it

Change the photo order, the checkout button, and the add-on row in the same week, and you won't know which one worked. Pick one change and give it enough traffic.

Appetier runs two versions of a storefront detail side by side and waits for 1,000 visitors and 100 paid orders on each before showing a result. You pick the winner.

Plan for the wait. At 500 menu visits a week split two ways, each version sees 250 visitors, so reaching 1,000 takes four weeks. At a 12 percent rate, each version collects about 30 orders a week and passes 100 in the fourth week too.

Fix a broken order before you fix the design

If the test ticket never reaches the kitchen, stop there. Pause online orders at that location until a test order prints, because a paid order nobody cooks costs you the guest.

If the price on the site differs from the POS, correct it at the source. With POS sync, prices, hours, and sold-out items update on their own.

If the pickup time was wrong, check the hours for that location, including any holiday hours. On Appetier, hours and sold-out items update without publishing, and other published changes reach open storefronts within about a minute.

FAQs

What's the fastest way to cut checkout drop-off on a restaurant ordering site?
Find the step where the most guests leave, and fix that one first. If it's the menu, make it load fast on a phone and cut the taps to the cart.
Should every guest see upsell suggestions?
Show upsell suggestions only when they fit what's already in the cart, like a drink with a meal. Suggestions that don't fit take up screen space and get skipped.
How do I work out my restaurant's online ordering conversion rate?
Divide paid orders by the number of guests who opened your menu in the same week. If 500 guests opened the menu and 60 paid, your conversion rate is 12 percent.
How long should a restaurant run a storefront test before picking a winner?
Long enough for each version to reach 1,000 visitors and 100 paid orders. At 500 menu visits a week, that usually means about four weeks.

Recommended product

A branded ordering site with one-tap reorders and add-on suggestions built in.

Online ordering