Attract restaurant customers: the before and after of a funnel you can actually measure

Step 1 · Audit your Google listing before touching a single post

Start with the Google Business Profile, because 79% of restaurant searches are NON-BRANDED (Malou, Local SEO for Restaurants 2025): people are not looking for your name, they are looking for food nearby, and 42% of those local searchers click inside the map pack (Semrush 2025, via Malou). The deliverable of this step is a one-page document with six numbers copied from the Insights panel: monthly impressions, clicks to the website, direction requests, calls, star rating and review count. The 92-seat grill I discuss here showed up 41,000 times a month and produced 380 direction requests: 0.9% conversion, at 4.1 stars with 37 unanswered reviews. Verification is blunt: if you cannot write those six numbers from memory in thirty seconds, the listing has no owner, and everything you build on top of it rests on sand. Repairing the listing happens on three concrete fronts and inside two weeks, not over a quarter.

Step 2 · Repair the reputation: reply, photograph, publish the live menu

Answer EVERY pending review —negatives first, under 60 words, no arguing, with one concrete action—, replace supplier stock photos with your own shots of the eight highest-margin dishes, and upload the full priced menu into the menu field rather than as an attached PDF. Keep the underlying volume in mind: searches for «food near me» grew 99% year over year (Restroworks, Google Restaurant Search Statistics 2024) and more than 60% of restaurant searches now start on mobile (Restroworks 2025). The deliverable is measurable: zero unanswered reviews, at least twenty original photos, a current priced menu. And verification is not internal, it is field work: search your category plus the neighborhood from someone else's phone, in incognito, and see where you land. Reviews do not arrive on their own: you ask for them at the exact moment of payment, and that moment belongs to the server, not to the community manager.

Step 3 · Build the new-review machine inside the service itself

Put a QR code on the check that goes straight to the Google review form, and treat the scan as part of closing the table. Guest behavior already favors you: QR scan volume grew 433% in two years (QR Code, QR Code Statistics for Restaurant Usage 2025), and QR ordering lifts check size by 9% against traditional dine-in (Sunday 2025). Set the goal as a weekly number, not an aspiration: twelve new reviews a week take a venue from 37 to 200 in four months. The deliverable is the per-shift tally sheet; verification is the average climbing from 4.1 to 4.5 without buying a single review. This is the most expensive leak of all and almost nobody looks at it: a delivery listing with 60 dishes and generic photos converts around 2%, while the same kitchen with 24 dishes sorted by margin and its own photography moves to 5-6%, without touching the recipe or the selling price.

Step 4 · Sort the digital menu by margin, where delivery leaks

The lever is documented: applied menu psychology lifts average check by 15% or more with no price increase (NeatMenu, Menu Psychology 2026), and a complete digital offer —menu, ordering and payment— moves it between 20% and 30% (Sunday, QR Code Ordering 2025). The deliverable here is a short list: the ten dishes that survive, ranked by contribution margin in currency, not by percentage. Verify it by crossing the platform report against your recipe costing: if the digital menu's hero dish leaves less margin than the third one, the order is backwards. Content sits in fifth place for a cash reason: it takes the longest to pay back. A venue posting daily with an unrepaired listing is handing traffic to whoever ranks first in the map pack, and that was exactly the diagnosis at the 92-seat grill: 148,000 USD in monthly sales, 3,200 in paid media, eleven months of posting, and not one table attributed.

Step 5 · Short-form video: the engine that lags, then compounds

Once the base is ready, content does compound, because it feeds the branded searches that come afterward. Lock a repeatable format —real kitchen, fixed frame, twelve to twenty seconds, no trending audio— and a cadence of three sustainable weekly pieces instead of ten erratic ones. The deliverable is a four-week calendar naming the dish and the angle of each piece. Verification is the branded search curve: if it grows, the content is working. A guest who already came through the door costs a fraction of one you have to attract, and that is precisely where most operators quit. Bain & Company documented that lifting retention by five points moves profits between 25% and 95% depending on the sector, and in restaurants, with food cost under 32%, each recovered visit falls almost entirely into contribution margin. Loyalty data backs the bet: 55% of restaurants report that their members' check grew faster than their menu prices (Paytronix, Loyalty Trends Report 2024) and member spend under one-to-one targeting rises 16.5% year over year (Paytronix, Effectiveness of Loyalty Programs 2025).

Step 6 · Close the loop with retention, the cheap money

Start simple, with two fields: phone number and date of last visit. The deliverable is a permission-based contact base; verification is the share of identified tickets per shift. The mistake I watch repeat most often is inverting the order: paid media and posts first, listing later, which means paying for cold traffic and sending it to a 4.1-star profile that scares it off. The second is measuring vanity —reach, followers, likes— instead of direction requests, calls and identified tickets; if your weekly report has no column in currency, it is not a report, it is a comfort blanket. Third comes buying reviews: Google purges them, the average drops, and the listing carries the mark. The fourth is subtler, leaving the listing in an agency's hands with no owner-level access of your own, and that administrative detail costs weeks when a holiday schedule needs fixing. Here is the real tension: marketing asks for creativity, the cash register asks for sequence, and sequence wins every time, because creativity without a funnel only burns media budget faster.

Closing checklist: how to know everything landed

You will know the guide is executed when you can answer six questions with a number and without opening a file: how many direction requests the listing brought last month, what your star average is and across how many reviews, what percentage of reviews is answered, how many dishes the digital menu carries and in what margin order, how many content pieces shipped, and how many tickets from the shift ended up identified. Under the Masterestaurant method we apply with Diego F. Parra, those six numbers live on a single sheet reviewed on Mondays, next to the week's food cost, because attracting guests and costing the plate are the same conversation seen from two ends. The final test is field work, not desk work: search your category from a borrowed phone at nine on a Friday night, and request directions the way a hungry stranger would. The bottleneck is rarely at the top.

Where the funnel actually leaks?

The grill house already surfaced 41,000 times a month in local search and lost people between the listing and the route, which is where a stale photo and an unanswered review do their quiet work.

The second leak lives in delivery. A listing with 60 dishes and supplier stock photos converts near 2%; the same kitchen with 24 dishes ranked by margin and its own photography moves to 5-6%, without touching recipe or price. The third one gets the least attention and costs the most: the guest who came once and never returned. Bain & Company documented that a 5-point lift in retention moves profits between 25% and 95% depending on sector, and in restaurants — food cost under 32%, low marginal cost to serve — that second visit is almost pure contribution margin. There is a real tension here. The content that wins reach — the kitchen Reel, the steaming plate — is rarely the content that books tables, because it reaches people living 40 km away.

Where the funnel actually leaks — in practice

You resolve it by measuring each function separately and accepting that the viral piece is brand advertising, not direct sales.