Restaurant Marketing — Answered Straight

They’re priced and positioned differently. OpenTable’s Core plan runs about $249/month plus $1/cover (Basic is $39/month plus $1.50/cover), but you’re paying for access to OpenTable’s own diner marketplace — real organic discovery from people browsing the app. Resy is flat-rate ($249-$899/month, no per-cover fee) and cheaper at volume, but its diner network skews toward its curated top-tier city markets. Toast Tables is essentially free if you’re already on Toast POS and holds roughly 5% market share — good for cost control, weak for standalone discovery. If new-customer discovery matters more than saving a few thousand a year, OpenTable’s network usually wins; if you already have a strong reputation and just need booking infrastructure, Resy or Toast Tables is cheaper.

One verified Google Business Profile per physical address — never a single profile representing multiple locations. Each location page also needs 600-1,200 words of genuinely unique content (neighborhood details, local team, area-specific FAQs), not the same copy with the city name swapped, because duplicate content across location pages dilutes relevance and can cause locations to cannibalize each other in search. Use a keyword map that assigns each location its own core-city and neighborhood terms so they’re not competing against your own pages.

These days are outliers, not normal Sundays — treat them like a separate campaign. Toast’s 2025 data shows Mother’s Day same-store revenue up 57%, transactions up 18%, average check up 32%, and reservations running 3x normal volume (with the 2pm brunch slot seeing 6x the reservations of a typical Sunday). Valentine’s Day sees steak sales jump 135%. Start email/SMS and reservation-page promotion 4-6 weeks out, lock in your deposit/reminder flow, and staff for a booking spike that starts well before the actual date.

Faster than you think, and faster than most competitors bother to. Only 73% of reviews get any response industry-wide, and 75% of businesses never respond to negative reviews at all — that’s your opening. Restaurants specifically face the highest urgency: diners increasingly expect a same-day or next-day reply. Businesses that respond to all reviews see roughly an 18% sales uplift and are chosen 89% more often over ones that don’t respond, and consistent responding nudges your average star rating up (about +0.12 stars) — Google also treats response activity as a local ranking signal.

A properly built one handles the real call mix, not just simple bookings. Based on typical restaurant call volume, about 48% of calls are reservations/waitlist, 26% are takeout/delivery orders (with live payment routing into the POS), 14% are menu/dietary/allergen questions, and the remaining 9% are hours/location/FAQ-type calls. That means it needs to actually modify an order, apply a discount code, or explain an ingredient substitution in real time — not just log a voicemail. It should escalate to a live human for genuinely heated complaints; anything trying to fully replace de-escalation is overselling what the tech does well.

Almost always because your menu lives only as a PDF or an image. Search engines and AI tools can’t reliably parse PDF menus, so your dishes, prices, and dietary tags never make it into rich results or an AI Overview/ChatGPT answer. The fix is Menu and MenuItem schema markup — structured data that exposes each dish, price, and dietary label (vegan, gluten-free, halal, etc.) as machine-readable content, which is also what lets your hours, reviews, and reservation link show up as a rich result in Google.

They drive real visits. 74% of diners use social media to decide where to eat and 68% check a restaurant’s profile before going, and 40% say they’ve tried a new restaurant specifically after seeing food photos online. On TikTok, 55% of users say they’ve visited a restaurant after seeing its menu on the platform. The catch is you need a steady stream, not a one-time photoshoot — Instagram’s #food hashtag alone has over 250 million posts you’re competing against, so consistency and encouraging guest UGC matters more than production value.

Build a dedicated catering landing page with its own inquiry form and phone routing — don’t rely solely on third-party catering marketplaces, which take a commission on top of your margin. It’s worth the effort: the catering market is headed toward $124B by 2032 (6.2% CAGR), average catering checks run $381 versus far lower standard takeout tickets, and a single catering order exposes your brand to roughly 25 new guests at that event — making it one of the highest-leverage lead sources you have if you capture the inquiry directly instead of through a marketplace.

A meaningful chunk of the phone-related labor gap, specifically. Restaurant labor costs already run 25-40% of revenue depending on segment (fine dining at the high end), and the labor market is still volatile — quits spiked to 785,000 in June 2025, the highest since early 2023. A host who’s also answering phones during a rush is why restaurants miss roughly 1 in 3 calls at peak — often 20-30 hours a week of staff time is tied up on the phone instead of the floor. AI phone coverage doesn’t replace your floor staff, but it removes the phone as a competing task during your busiest windows.

List everywhere your specific customers already search, with Google as the non-negotiable baseline since it’s free and captures people who never open a reservation app at all. Add OpenTable if you want its larger diner marketplace for new-customer discovery, and Resy if your market and price point line up with its more curated, lower-fee model. Running only one exclusive platform means you’re invisible to everyone who defaults to a different app — most multi-unit groups run Google Reserve plus one paid platform rather than trying to maintain three.

Everything scales in complexity. You need a unique, non-duplicate landing page per location (not templated pages with the city name changed), a separate verified Google Business Profile per address, a keyword map that assigns each location distinct neighborhood terms so locations don’t cannibalize each other in search, and centralized reporting to see which location’s marketing is actually working. This is also where review reputation gets harder to manage — one location’s slow response streak can drag down brand-level trust signals even if the other four are doing fine.

A chatbot alone leaves the largest single channel uncovered. Roughly half of all restaurant customer contact still happens by phone call (reservations and takeout orders combined make up about 74% of typical call volume), and restaurants miss close to a third of calls during peak service when staff are slammed. A chatbot helps with website visitors who are already typing, but it does nothing for the person calling from their car deciding where to eat tonight — that’s a phone problem that needs phone coverage, not a text widget.

Split them. A single scrolling menu page can only realistically target one intent, but “brunch near me,” “downtown catering,” and “date night dinner” are different searches with different competitors, so each deserves its own URL, title tag, and internal links. This also lets you attach category-specific schema (MenuSection, MenuItem) instead of one generic block, which is what feeds GEO/AEO citations later. We build this structure as part of SEO Engineering rather than bolting it onto a template that wasn’t designed for it.

Yes, indirectly but meaningfully — dietary-tag pages capture long-tail, high-intent searches like “gluten-free pasta [city]” that a plain menu listing never ranks for, and they’re exactly the structured attributes AI answer engines pull from when someone asks about dietary restrictions. It also reduces bounce from people who land, can’t confirm an option quickly, and leave. The work is tagging every item consistently in the CMS, not just adding a note somewhere on the page.

More than most owners assume — a slow-loading mobile page is where a “let’s just go there” decision turns into “let’s just go somewhere else,” and Google’s own research has shown abandonment rises sharply once mobile load time crosses a few seconds. It also affects whether Google bothers to rank you well for “near me” searches in the first place, since mobile speed is a ranking signal. This is core technical SEO work — compressing images, cutting bloated plugins, fixing render-blocking scripts — not a redesign.

It helps, if it’s structured right — fresh, dated seasonal content (a “Fall Menu” page, a limited-time dish landing page) gives Google new signals to crawl and gives you something timely to link from social and email. The mistake is deleting last season’s page outright instead of redirecting or archiving it, which throws away accumulated authority and creates dead links. We build a seasonal-menu workflow into the content SEO plan so drops are additive, not destructive.

Third-party listings don’t hurt your rankings directly, but they compete for the same “order [restaurant name] online” search real estate, and if your own site isn’t optimized for that query, the delivery app’s listing — with its 15-30% commission — wins the click by default. The fix isn’t abandoning delivery apps; it’s making sure your own ordering page outranks them for branded searches through better title tags, page speed, and a Product/Menu schema the apps don’t bother with.

Restaurant schema alone gets you found; Menu, MenuItem, Review, and AggregateRating schema get you cited with specifics. AI Overviews and tools like Perplexity pull structured facts — price range, cuisine, specific dish names, dietary tags — to answer questions directly, and unstructured prose menus mostly get skipped. GEO Optimization is largely this: making sure every fact an AI might quote (hours, price tier, signature dish, accessibility) exists in a machine-readable form somewhere on your site.

These tools lean heavily on a mix of structured data, third-party review consensus, and authoritative mentions — a local food publication review, a “best patios in [city]” roundup, consistent NAP and attribute data across your site and listings. If “outdoor seating” isn’t explicitly tagged as an attribute anywhere machine-readable, the model has no clean signal to match you against that specific query even if you have a great patio. AEO work here means structuring content as direct, quotable answers to attribute-specific questions, not just describing the space in marketing copy.

Yes, and it’s one of the more winnable categories because most competitors don’t bother tagging dietary attributes at the item level. It takes a dedicated dietary-options page with each dish explicitly labeled, ideally with schema markup, plus a few credible external mentions (a local “best gluten-free spots” list, a review that specifically praises the options). Answer engines favor sources that state the fact plainly rather than implying it.

Yes — inconsistency is one of the fastest ways to get excluded, because these engines cross-reference multiple sources and tend to suppress citations when they can’t reconcile conflicting facts (your site says $$ , Yelp says $$$, Google says closed Sundays but you’re open). Authority signals matter, but consistency across every place your restaurant appears matters just as much for GEO. This is a recurring audit, not a one-time fix, since third-party sites drift out of sync on their own.

Specials need to exist as standing, structured, evergreen content — a dedicated “Happy Hour” or “Lunch Specials” page with specific times, prices, and items — not a rotating Instagram story that disappears in 24 hours. AI engines can’t cite content that isn’t crawlable and current, so if your specials only live on social, you’re invisible to that query entirely. We build these as permanent AEO-structured pages that get updated, not recreated, each time the offer changes.

Yes — this is a rule-based routing setup, not a limitation. The agent (built on Vapi, Bland, or Retell depending on what fits your stack) can take the full request — party size, date, special requirements — confirm it’s provisional, and immediately notify the manager via text or your booking system for approval, rather than either falsely confirming a table you don’t have or dropping the lead to voicemail. You define the threshold — 8 tops, 20 tops, whatever triggers manual sign-off — and it enforces it every time, consistently, unlike a rushed host on a Friday night.

It integrates with your actual reservation and POS data where an API or booking system exists — checking real table availability through your OpenTable/Resy/Toast Tables setup rather than guessing or over-promising. Where a system doesn’t expose live availability, we set clear fallback logic (offer the next two likely time slots, take a callback request) instead of leaving the caller with a vague “we’ll check and call you back.” The integration work is scoped per client because POS and reservation stacks vary a lot restaurant to restaurant.

Yes, and this is the core advantage over a human host — the AI agent handles concurrent calls without a queue, so nobody gets a busy signal or an endless ring while your one phone line is tied up with a to-go order. Each caller gets full attention: reservation taken, question answered, order placed, at the same time as five others. That’s the exact scenario (“phone unattended during rush”) that drives most of the missed-call revenue leak restaurants don’t track.

Yes — it works from your actual menu and pricing, so it can quote accurate totals, suggest add-ons (“want to add a side of fries or a drink?”) the way a trained order-taker would, and repeat the order back for confirmation before it’s sent to the kitchen. This is where AI voice agents pay for themselves fastest, since phone-in ordering is usually the most understaffed, most error-prone part of takeout operations. It’s configured against your real menu structure, not a generic script.

Yes — multilingual handling is a configuration choice on setup, not a separate product, so the agent can detect or be set to a specific language and hold a full conversation in it, including taking orders and reservations. This matters more for restaurants than almost any other business type, since a huge share of missed-call revenue in diverse markets is a language mismatch, not disinterest. We set the default languages based on your actual customer base, not a generic English-only script.

You control this through a simple schedule/override rather than editing a phone tree — set holiday closures, private-event buyouts, or early closing hours in advance, and the agent applies them automatically on the day, including offering the next open date if relevant. This eliminates the classic problem of an outdated voicemail greeting three holidays out of date. It also means front-of-house doesn’t have to remember to flip a setting on the way out the door.

Your own site should own the ordering flow whenever possible — every order through a delivery app costs you 15-30% in commission plus the customer relationship, since you don’t get their email, phone, or order history. We build ordering directly into the site, wired into your POS, so pickup and delivery orders convert on your domain and feed your CRM; delivery apps stay useful for discovery and overflow capacity, not as your default checkout. The math almost always favors owning the transaction.

It’s a real problem, not a minor one — PDFs are slow to load on mobile, unreadable without pinch-zooming, invisible to Google for menu-item search, and completely unreadable by AI answer engines, which means you lose SEO, AEO, and GEO visibility simultaneously from one bad file format. We rebuild menus as structured HTML with schema markup so every dish is its own indexable, citable piece of content. It’s one of the highest-leverage, lowest-drama fixes we make on a restaurant site.

Yes, and it should be — sending someone off-domain to book adds friction and a chance they abandon on an unfamiliar page. We embed the widget natively so booking happens without leaving your site, matching your branding instead of the platform’s generic look, while still writing to the same backend system your host stand uses. It’s a standard integration regardless of which reservation platform you’re on.

Each location needs its own page — separate URL, separate hours, separate menu if it varies, separate local schema — because that’s what lets each location compete individually in local search instead of all of them fighting for relevance on one generic page. A shared locations-overview page is fine as a hub that links out, but it can’t replace individual location pages if you actually want each address to rank in its own neighborhood. This is standard for any multi-location build we do.

It depends on your POS — most modern systems (Toast, Square) already have gift card and loyalty infrastructure, and the right move is usually surfacing that natively on your site rather than bolting on a separate plugin that doesn’t talk to your POS or CRM. We build the site to pull from and write to your existing systems so a gift card purchase or loyalty signup on the website shows up correctly at the register, not in a disconnected silo.

Every extra second past roughly two to three seconds measurably increases abandonment, and it compounds during your busiest windows — exactly when a slow site costs the most money. We build ordering pages on fast frameworks (React/Next.js) with optimized images and minimal third-party script bloat specifically because speed on the ordering flow has a direct, trackable revenue impact, unlike speed on, say, your about page.

Yes — this is a segmentation-and-trigger setup: the CRM tracks what a customer orders or shows interest in, tags them accordingly, and fires an automated text when that item returns to a seasonal or limited menu. It’s the kind of personal touch that used to require a manager remembering a regular’s order by heart, now running consistently across your whole customer base instead of just the handful of guests staff happen to recognize.

Yes — a no-show can trigger an automated flow instead of disappearing into the void: a gentle “we missed you” message, a note on their profile visible next time they call, or a re-engagement offer if it’s been a pattern. Left unmanaged, no-shows are pure lost revenue with zero recovery attempt; automated, they become a data point that improves how you handle that customer next time, including whether to require a card on file.

Yes — that’s the core of the CRM: leads from Meta and Google ads, on-site forms, and the AI chatbot all funnel into one system automatically, tagged by source, instead of living scattered across an inbox, a DM folder, and a notepad by the phone. From there, automated follow-up sequences (email, WhatsApp, SMS) go out immediately rather than whenever someone gets a free minute, which matters enormously for catering leads that go cold within hours.

Yes — timing is configurable and matters a lot; a request sent right as someone’s paying feels transactional and gets ignored, while one sent a couple hours later, once they’ve had time to actually enjoy the meal, converts better and reads as genuine. We tie the trigger to your POS checkout or reservation completion time so it’s automatic and consistently timed, not dependent on staff remembering to ask.

Yes — the CRM captures order history and visit frequency, so you can send a loyalty-tier offer to your top regulars, a “come back, it’s been a while” nudge to lapsed customers, and a first-visit welcome offer to new ones, instead of blasting the same generic promo to everyone. Treating a twice-a-week regular the same as someone who visited once for a birthday is one of the most common ways restaurants waste their own customer list.

Yes — once birthday or anniversary data is captured (through a loyalty signup, reservation form, or email opt-in), the workflow triggers the offer automatically on the right date every year, with no manual list-checking required. This kind of automation is exactly what workflow automation is built for: recurring, rules-based tasks that get skipped when they rely on a busy staff member’s memory.

It starts with a specific, dated offer promoted through paid social targeted at people near you who haven’t visited recently, not organic posts hoping the algorithm favors you that week. We build content calendars around specific dayparts and slow shifts — a Tuesday trivia night, a mid-week prix fixe — and pair the organic post with a small retargeting budget aimed at past customers, since that combination consistently outperforms organic reach alone for driving a specific night’s covers.

Organic keeps your existing audience warm, but it rarely reaches new people at meaningful volume anymore, and it can’t be targeted to “people within 3 miles who haven’t ordered in 60 days.” Even a modest paid budget aimed at geo-targeted locals and retargeting past visitors outperforms organic-only for actually filling seats, especially for time-sensitive pushes like a slow night or a new menu launch. We typically run both — organic for brand and retention, paid for acquisition and specific pushes.

These need a separate funnel from your regular catering-quote form, since the buyer researches differently (longer timeline, higher budget, comparing multiple venues) and the ad targeting is different (event planners, engaged couples, corporate admins vs. someone ordering lunch for the office). We build a dedicated private-events landing page and lead-gen campaign with its own qualifying questions, then route those leads into the CRM tagged separately so follow-up matches the sales cycle instead of getting a generic catering auto-reply.

Yes — the key is the offer and framing, not the targeting itself; a warm “we miss you, here’s 15% off your next visit” ad to someone who’s genuinely dined with you before reads as a normal loyalty touch, not surveillance, especially compared to cold prospecting. We build these win-back segments directly from your CRM’s visit data so the audience is accurate and the messaging is specific to lapsed regulars rather than generic.

Start building anticipation well before opening — behind-the-scenes construction content, staff hiring posts, a countdown, local influencer previews — because a cold social launch on opening week has no built-in audience and misses the pre-opening buzz that drives a strong first month. Paired with lead generation (an email signup for opening-week reservations or a soft-launch invite list), this turns your following into actual first-week covers instead of just impressions.