Nonna's Table Walkthrough: A 90-Day Modeled Switch from Delivery Apps to Direct Ordering

This is an illustrative scenario based on modeled assumptions for a family Italian restaurant doing roughly $80,000 per month before switching. Not a real customer outcome. "Nonna's Table" is a stand-in we use to walk through what a 90-day shift from delivery apps to direct ordering can plausibly look like when the tactics line up. Every number in this walkthrough is modeled, not measured. The assumptions behind those numbers (commission rates, AOV ranges, processing fees, retention math) are pulled from public sources and cited inline. Your real numbers will differ.

We rebuilt this post because the original 800-word version skipped the math. Restaurants kept asking, "what does this actually look like, week by week?" That is what the next 16 minutes covers.

What "Nonna's Table" Looks Like Before the Switch

Picture a family-run Italian restaurant in a U.S. metro market: about 12 years old, 60 seats, classic red-sauce menu plus weekly specials, owned by a husband-and-wife team with a kitchen crew of six. Modeled baseline numbers:

The owners' diagnosis going in: covers are healthy, the kitchen is full on weekend nights, but profit margin is sliding. They run the math one Sunday and discover roughly $5,500 to $6,500 a month is leaving as marketplace commission. That is the trigger for the 90-day project.

The Goal: Flip the Channel Mix Without Losing Volume

The goal is not to cut delivery apps out. It is to flip the dominant channel. Modeled north-star metrics for week 13:

• Direct-order share moves from ~35% to ~55% of off-premise
• Marketplace commission paid drops by roughly $3,800 to $4,400 per month
• Total off-premise covers hold steady (no net volume loss)
• A captured first-party customer list of about 2,100 records
• Phone order handling time falls by roughly half

The reason retention math drives this whole project: it costs five to twenty-five times more to acquire a new customer than to retain an existing one, and a 5% increase in retention can lift profits 25 to 95% in services businesses (Bain via Harvard Business Review, "The Value of Keeping the Right Customers"). Once Nonna's Table has a customer's email and order history, every future order from that customer is a retention play, not an acquisition cost.

Weeks 1 and 2: Setup, Menu Import, Photography

The first two weeks are unglamorous. They are also where most rollouts go off the rails.

What gets done:

• DirectOrders account live on day 1, branded ordering site provisioned on a custom subdomain
• Menu imported from the existing Toast POS, then manually corrected (modifier groups, photo placeholders, allergen flags)
• A Saturday afternoon photo session for the top 30 items, using natural light and a borrowed DSLR
• Stripe processing connected, business verification submitted (Stripe Pricing)
• Google Business Profile updated with the new ordering link as primary action

What goes sideways:

• Toast menu export missed two weekly specials and three combo meals. The owner notices on a Wednesday lunch when the website still shows the old menu. Fix: 90 minutes of manual editing.
• Two photos look worse on the live site than they did on the camera. Reshoot the Friday after.
• The branded subdomain (order.nonnastable.com) takes 26 hours to propagate. Plan for it; it is normal.

Modeled cost: $0 incremental software (DirectOrders Pro at $249/month replaces an existing booking and online ordering tool of similar cost). Roughly 14 hours of owner and manager time, on top of normal operations.

End of week 2: Site is live but doing essentially zero direct volume. The marketing has not started yet.

Weeks 3 and 4: Launch the QR Insert and Reset Google

Week 3 is when the story changes. Two tactics go in at once.

Tactic 1: QR-coded packaging insert

Every takeout bag, every delivery bag (including app deliveries the kitchen still fulfills), and every dine-in check folder gets a 4 by 6 card. Front: "Order direct next time, get 15% off." Back: a QR code, the URL, and a one-line reason ("we keep more of every dollar, you get faster pickup, and the price you see is the price you pay").

This works because of QR adoption: 89 million U.S. smartphone users were forecast to scan a QR code in 2024, more than triple the 2019 number (Statista, U.S. mobile QR code scanners). The behavior is no longer fringe.

Tactic 2: Google Business Profile reset

The "Order Online" link on the GBP gets pointed at the new direct site. Photos refresh. The owner posts a new GBP update every Sunday for the rest of the project. This costs nothing and tends to be the single highest-ROI free move available to an independent restaurant.

Modeled outcome at end of week 4:

• ~180 direct orders captured
• ~$6,800 in direct revenue (roughly 8% of total month, up from near zero)
• ~140 emails captured (some customers skip checkout email; that is normal)

What did not work: A QR sticker on the host stand. Diners do not scan things at the host stand. The card in the bag is the workhorse.

Weeks 5 and 6: Turn on AI Phone Ordering

Phone is where most Italian restaurants leak revenue on Friday and Saturday nights. The line rings, the host is plating salads, the call goes to voicemail, and a customer who would have spent $52 orders pizza from somewhere else.

In week 5, AI phone ordering goes live. The setup is opinionated: it answers in two rings, takes routine orders end to end, transfers to staff for catering and large parties, and texts an order confirmation to the customer.

Modeled assumptions for the phone channel (illustrative ranges from typical AI phone deployments described in Popmenu's Restaurant Consumer Trends Report, which finds 35% of restaurant orders still come by phone):

• Average phone order ticket: $42 (slightly higher than online, because callers are usually ordering for groups)
• Phone handling time per order: drops from ~4 minutes (human) to ~30 seconds of staff time on the AI-handled portion
• Phone calls per Friday night: ~95, of which ~70 are routine

What goes wrong in week 5: The AI mishears two callers' addresses. The owner listens to the call recordings, retunes the prompt, and the issue largely clears up by week 6. This is the first lesson of AI phone: it works out of the box for clean orders, but you have to listen to the first 50 calls and tune the prompts.

End of week 6:

• Direct order share: ~22% of off-premise (up from ~10% at the end of week 4)
• Phone order handling time: down from ~4 hours/night on weekends to ~1.5 hours
• Captured customer records: ~520

Weeks 7 and 8: Reorder Marketing and the First Email Send

By week 7, the customer list is meaningful. Time to use it.

The reorder workflow is intentionally simple:

• Day 14 after first order: "Loved seeing you. Here is 10% off your next order this week" (no code needed, applies via personalized link)
• Day 30 after last order: "We miss you. Free dessert with your next order"
• Day 60 after last order: "It has been a minute. Here is 20% off, our way of saying come back"

These are not novel. They are the standard restaurant lifecycle workflows that have been published in CRM playbooks for a decade. What changes is who owns the list. With marketplaces, this email goes out under the marketplace's brand, to a customer the marketplace owns. On direct, it goes out under Nonna's brand to a customer Nonna owns.

Modeled response rates (modeled from typical email benchmarks for independent restaurants; numbers vary widely by list quality and frequency):

Across roughly 700 customers in the active list at the end of week 7, the first email cohort generates about 35 incremental orders worth roughly $1,400 in revenue at $0 in commission.

What did not work: A segmented "vegetarian only" send the team rushed out. The segment was 18 people and the email got reported as spam by one of them, hurting domain reputation for a week. Lesson: do not send to segments smaller than 100, and do not skip the warm-up.

Weeks 9 and 10: First-Time Diner Acquisition Push

By week 9 the engine is running on existing customers. To grow direct beyond the existing base, the next push is acquisition: paid social ads with a "first order 20% off" hook, plus a partnership with a neighborhood mailer that drops 4,000 doors.

Modeled CAC (cost to acquire a customer):

• Paid social: $14 per first-order customer (modeled from typical local-targeted Meta ad performance for independent restaurants)
• Mailer drop: $0.42 per piece x 4,000 = $1,680, generating an estimated 28 first-order customers, so $60 CAC

Compare that to the retention math: the cost to keep a known customer ordering is essentially the cost of the email send (a few cents) plus the discount given. Bain's classic finding is that acquiring a new customer is five to twenty-five times more expensive than retaining one (Harvard Business Review, "The Value of Keeping the Right Customers"). The math at Nonna's Table is consistent with that range.

End of week 10:

• Direct order share: ~38% of off-premise
• Captured customer records: ~1,300
• First-order CAC blended: ~$22
• Reorder rate (orders 1 to 2 within 30 days): ~28%

Weeks 11 to 13: Compounding the Lift and Stress-Testing

The last three weeks are about compounding. New customer acquisition continues, but the bigger story is repeat behavior on the now-meaningful direct list.

Tactics added in this stretch:

• Loyalty: every $100 spent direct earns a $10 credit. Modeled redemption rate: ~32%.
• A Sunday family-meal bundle pushed only via direct ordering and email. Modeled to lift Sunday revenue ~12%.
• A second wave of QR cards in every bag, with a different message ("Saved $4.20 on this order vs. ordering through the app").

What did not work, again, honestly:

• A "tag us on Instagram for a free tiramisu" promo got 11 redemptions over two weeks. Cool, not material. Pulled it.
• An attempt to move catering inquiries to a self-serve form failed. Catering is high-touch by nature; the AI phone routes those to staff and that stays.

End of week 13 (modeled):

The Tools Used in This Walkthrough

For readers building their own version, here is the modeled stack:

This is a deliberately small stack. We have seen rollouts fail because owners try to layer in five marketing tools at once. One ordering platform, one POS, one ad channel, one local push, and one email engine is plenty for the first 90 days.

Modeled Before / After P&L

The single page of math most owners want to see. All numbers modeled for an $80,000-per-month restaurant with the assumptions above; not measured outcomes.

The acquisition spend tapers off. By month 6 in the modeled scenario, paid acquisition drops to ~$300/month and the net margin impact rises to roughly $2,800 to $3,500/month, before any catering or private-event upside.

You can model your own version using our commission calculator and break-even calculator with your real numbers.

The 90-Day Channel Mix Shift, in One Table

Pickup and dine-in fill the rest of the mix. The point is the marketplace line, not the direct line: it moves from 70% to 30% over 13 weeks, without total volume falling.

What Required Hand-Holding (Honest Tradeoffs)

A few things this walkthrough makes look smoother than they would be in real life:

• Photography is harder than you think. A reshoot is almost certain. Plan two photo sessions in the first month, not one.
• The first AI phone calls need supervision. Owners who do not listen to the first 50 calls and tune the prompts often abandon the channel prematurely.
• Email deliverability needs warm-up. Sending to 2,000 cold addresses on day 1 lands you in spam. Ramp from 100 a day to 500 a day over two weeks.
• Catering and special events should stay human-routed. The model assumes AI handles routine, humans handle high-touch. Trying to automate catering inquiries hurts conversion.
• Marketplaces are not enemies. The walkthrough keeps DoorDash and Uber Eats live for discovery; it just stops treating them as the primary channel. Many restaurants overcorrect and shut off marketplaces entirely on day 30, then panic when discovery slows. Don't.

Reorder Math: Why This Works Long-Term

The 90-day window is the setup. The compounding happens after.

Bain's research, summarized by Harvard Business Review, finds a 5% bump in retention can lift profits anywhere from 25% to 95% in service businesses (HBR, "The Value of Keeping the Right Customers"). Restaurant repeat economics are similar: a customer who orders three times in 90 days is dramatically more valuable than three one-time customers, because the marginal cost of getting them to order again is essentially zero.

Modeled illustration:

• 100 first-time direct customers acquired at $20 CAC = $2,000 spent
• If 28% reorder within 30 days (modeled rate), that is 28 customers worth roughly $42 each = $1,176 in low-CAC revenue
• If half of those reorder a second time, another ~$590 at near-zero acquisition cost
• After 90 days, the LTV-to-CAC ratio is climbing fast even on conservative assumptions

That is the engine the QR insert, the AI phone, and the email workflows are feeding. The 90-day window builds the asset (the list). The next 12 months monetize it.

Bottom Line: This Was Modeled, Here Is How to Model Your Own

To restate, because we know readers skim: none of the numbers in this walkthrough are measured customer outcomes. They are modeled, with the assumptions cited, for a hypothetical Italian restaurant doing roughly $80,000 per month before switching. Your AOV, your commission rate, your marketing effort, your kitchen capacity, and your local market will all push the numbers up or down.

The point of the walkthrough is not the specific dollars. It is the shape of the project: setup, channel reset, AI phone, email engine, acquisition, loyalty, repeated over 13 weeks with honest hand-holding where it is needed.

If you want to model your own version with your real numbers, the fastest path is our commission calculator for the marketplace side and our break-even calculator for the cost side. If you want to see the underlying platform that this scenario sits on, book a 20-minute demo and we will walk through your menu and order volume on a screen share.

The retention math in the Bain study is the part you should not let go of. Customers you can email are worth more than customers you cannot. Channels you own beat channels you rent. The 90 days at Nonna's Table is the work it takes to flip that ratio.

For deeper coverage of the underlying tactics, see our 90-day migration playbook, our AI phone ordering guide, our breakdown of same-day payouts and cash flow, and the customer database guide. All of those are practitioner posts; this walkthrough is the modeled scenario that ties them together.