Running reviews as an operating metric means holding each location to a monthly standard for how many reviews it earns relative to the work it completes, how completely and how quickly it responds to them, and how its recent rating is moving, and then reporting those numbers in the operating review next to revenue, labor, and margin. It is not the same thing as reputation management, which is a software category and a marketing activity. The operating version starts from a different assumption: the reviews a location earns are an output of the service that location delivered, so the number belongs to the person who runs it.
This guide is written for sponsors and operators of multi-location platforms, and it covers the measurement and accountability side only. The case for local search as an owned channel, the profiles, the location pages, and the citation footprint are covered in our playbook on local search as an owned asset, which treats the review base as one of four assets. This is the build-out of the one line in that playbook that says reviews belong in the monthly operating review.
Every review in a services business is produced by a visit. Someone showed up, did work, and left an impression strong enough to write down. The ask happens in the field, in the last few minutes of the job. The response, when it is any good, is service recovery performed in public. None of those three things happens at headquarters, which is why a central team can build the system and still not move the number.
The practical consequence shows up in what each reporting line produces. Put reviews on the marketing scorecard and you get a software purchase, a dashboard, and a quarterly screenshot of a portfolio average holding steady. Put the same metric on the operating scorecard, with a per-location target and a name against it, and you get a change in what happens at the end of the job. Only the second version produces reviews.
The commercial stakes are larger than the ranking argument suggests. In one five-brand consumer services platform we diagnosed, 38 percent of revenue arrived from untracked word of mouth. Word of mouth in a local services category almost always routes through a search: somebody hears a name, looks it up, and decides based on what is showing. The review base is the last thing standing between demand the company already earned and a competitor sitting one line below it.
Most review dashboards report a rating and a count. Neither is an operating metric: neither is comparable across locations, and neither tells a manager what to do differently tomorrow. Four numbers do, and all four are reported per location.
Reviews earned per hundred completed jobs or transactions. The denominator is what makes the metric work. A raw count rewards the largest location for being large and punishes a small one for being small, so managers learn quickly that the number is unfair and stop taking it seriously. Capture rate puts a twelve-van branch and a two-van branch on the same axis, and it answers the only question actually in the manager's control: of the customers we served, how many did we ask well enough that they wrote something.
The share of reviews that received a reply, measured over the month, with median hours to first response as the secondary number. Coverage comes first and speed second, because a location that answers eighty percent of reviews within an hour is doing worse than one that answers all of them within a day. Reversing the order produces a location that replies instantly to easy reviews and leaves the hard ones alone, which is the opposite of the behavior the metric exists to create.
Not the lifetime rating. A location with four hundred reviews accumulated over six years has a lifetime average that barely moves regardless of what happens this quarter, which makes it useless as a management signal and slightly dangerous as a comfort. The rating computed on the last ninety days of reviews moves when operations move, which is the entire point.
The share of reviews at three stars or below, expressed per hundred jobs rather than as a share of reviews. Reported as a share of reviews it moves whenever capture rate moves, which makes a location look worse precisely when it starts asking more customers. Per hundred jobs it is a clean read on something no other report in the business produces: the rate at which service failed badly enough that the customer was willing to publish it.
A single portfolio-wide target is the most common way this program fails in its first quarter. Locations differ in age, transaction volume, service mix, and competitive density, and a number set for the average produces artificial pressure in half the estate and false security in the other half. Both halves stop believing the metric at the same time.
The workable method is to derive the target from the portfolio's own distribution. Measure the capture rate every location is running today. Take the upper quartile as the standard, because it is being achieved by real locations with the same brand, the same systems, and roughly the same customers, which makes it unarguable in a way a benchmark from a vendor report never is. Then phase each location toward it over two or three quarters rather than announcing it as a cliff.
Two guardrails hold the rest of the design together. Never target a star rating, because a rating is a result and a manager can only be held to behavior; targeting the rating teaches people to manage the denominator by asking fewer customers. And never target a raw count without a denominator, for the same reason capture rate exists. Both mistakes are reliable, and both produce a metric that looks healthier the less honest it gets.
The portfolio average is the least informative number in the pack. The distribution is where the operating content lives, and the useful move is to read capture rate and negative rate as a pair, location by location.
In a platform assembling locations by acquisition, first-year variance is usually inherited process rather than service quality. One Claymore engagement saw a platform grow its footprint by 82 percent in 15 months; at that pace, locations arrive with their own habits about when and whether anyone asks, and the spread on capture rate says more about the seller's prior practice than about today's crews. That distinction matters, because the two problems have different fixes and only one of them is urgent.
The ask has to be attached to a moment in the work rather than to a campaign. The moment that performs is the point of visible completion, when the customer can see the result and the person who produced it is standing there. A verbal ask at that moment, followed by an automated send within a couple of hours, outperforms any batch send by a wide margin, because the automated message is arriving as a reminder of something the customer already agreed to do.
Three design rules keep it honest. Whoever asks has to be able to see the result, or the ask degrades into a script nobody believes in. One channel, one template, and one timing rule across the platform, so that variance between locations reads as behavior rather than as mechanics. And never gate the ask by screening for happy customers first, which violates the terms of every major platform and, more to the point here, destroys the diagnostic value of the negative rate. A review base that only contains satisfied customers tells an owner nothing and tells a buyer even less.
Where the locations are franchised rather than owned, the ask is one of the behaviors a franchisor has to persuade rather than mandate, which changes the design substantially; that split is covered in our guide to marketing for PE-backed franchise brands.
Responses are written locally, against a central standard. Local writing is specific, and specificity is the only thing that makes a public reply credible to the next reader, who is the actual audience. Centralized response teams produce fast, uniform text that reads as automated, and they sever the loop between the complaint and the people who could prevent the next one.
The standard fits on one page: full coverage on anything at three stars or below, a reply within two business days, no arguing the facts in public, an offer to continue off the platform, and no use of the customer's private details. The escalation rule has to be absolute. Anything alleging a safety incident, discrimination, billing fraud, or property damage leaves the location and reaches a named person at the platform within a day. Those are not reputation events; they are risk events that happen to be public, and they are the ones that surface in diligence.
One page. Per location: capture rate, response coverage, median response time, trailing ninety-day rating, negative rate per hundred jobs. Then the distribution rather than the average, and the two or three locations sitting off target with a named owner and a stated action against each. That is the whole report. Review text does not belong in a board pack, and a wall of screenshots is a reliable sign that nobody is managing the underlying numbers.
The reason a sponsor takes the page seriously is that it is one of the few commercial metrics that cannot be bought. In S&P Global Market Intelligence's 2026 Private Equity Survey, 71 percent of general partners and 53 percent of limited partners said they prioritize operational value creation over financial engineering. Review performance is operational value creation with a public scoreboard: it moves when service and follow-through move, and there is no media budget that fakes it. How this page fits the rest of the pack is covered in our guide to board-ready marketing reporting.
Asking behavior changes within a month of a standard being set and a name being attached. Capture rate moves within two. The trailing ninety-day rating takes a full quarter to reflect any of it, because the window itself has to turn over. Local ranking movement comes later than all of that and is the least dependable link in the chain, since the map pack also weighs proximity, category, and whatever the competitors are doing, none of which a general manager influences.
So judge the first two quarters on capture rate and response coverage, the inputs under management control, and treat ranking movement as a lagging consequence rather than a target anyone carries. Committing a sponsor to a ranking outcome is how these programs get canceled in month five, one quarter before the compounding starts. Where to begin depends on what the numbers say today, which is what an eight-week commercial audit establishes, and how we sequence the work from there follows the fact base rather than a template. The pattern across our engagements is consistent: the durable gains come from installing the system and holding someone to it.
It means holding each location to a monthly standard for how many reviews it earns relative to the work it completes, how completely and how fast it responds to them, and how its recent rating is moving, and reporting those numbers in the operating review beside revenue, labor, and margin. The distinction from reputation management is ownership rather than tooling. Reputation management is a software category bought by a marketing team. Reviews as an operating metric assumes the reviews a location earns are an output of the service it delivered, so the number belongs to the person who runs the location.
Four, reported per location and never only as a portfolio average. Capture rate, meaning reviews earned per hundred completed jobs or transactions, which makes a small location comparable to a large one. Response coverage, the share of reviews that received a reply, with median hours to first response as the secondary measure. Trailing ninety-day rating, which reflects current operations rather than the lifetime average a hundred old reviews are holding in place. And negative review rate, the share of reviews at three stars or below per hundred jobs, which is the service-failure rate customers were willing to publish.
From each location's own current capture rate rather than a single number applied across the portfolio, because location age, transaction volume, and competitive density all differ. Measure the capture rate every location is running today, set the portfolio's upper quartile as the standard, and phase each location toward it over two or three quarters. Two guardrails matter. Never target a star rating, because a manager can only control behavior and a rating is a result. And never target a raw review count without a denominator, because that punishes smaller locations and rewards volume rather than capture.
The location, with a central standard and a named escalation path. Responses written locally are specific, which is the only kind customers find credible, and the person who writes them learns what is going wrong. Central ownership produces fast, uniform replies that read as automated and that disconnect the complaint from the people who could fix it. The platform should own the standard, the timing rule, and the template library, and should pull any review alleging a safety issue, discrimination, billing fraud, or property damage out of the location's hands and to a named person within a day.
Longer than the operating metrics themselves, and less reliably. Asking behavior changes within a month of the standard being set, capture rate moves within two, and the trailing ninety-day rating takes a full quarter to reflect it because the window has to turn over. Map-pack movement follows review velocity rather than leading it, and it is also affected by proximity, category, and competitors, none of which a general manager controls. Judge the first two quarters on capture rate and response coverage, and treat ranking movement as a lagging consequence rather than a target anyone is held to.
Have a revenue problem the board is asking about? Start a conversation.