I have done somewhere north of two hundred listings audits in my career, and the one I want to walk you through is not the cleanest or the most successful. It is the one that taught me the most, because it forced me to throw out half my standard checklist on day three. If you manage local presence for any business with more than a couple of locations, the pattern will be familiar.
What follows is how I actually ran the audit, what I found, what I would do differently if the same client walked in with a smaller budget, and the principles I now apply to every engagement. Take it as a senior colleague’s notes, not a manual.
The client that triggered this audit
A regional HVAC chain with 47 locations
The client is a heating and cooling business operating across three states in the US Midwest, grown partly organically and partly through two acquisitions in 2023 and 2024. By the time they reached me in October, they had 47 active service locations, three call centres, and a corporate marketing team of four people who were, by their own admission, drowning. They had switched from a managed listings provider to an in-house workflow eighteen months earlier to save money. That decision is relevant.
The symptom: lead volume dropped 22% in Q3
Inbound leads (form fills plus tracked phone calls) had fallen 22% year-over-year for the third quarter. The marketing director’s initial theory was a competitor pricing war, which is the kind of thing marketing directors say when they have not looked closely at the data. Paid search performance was flat. Organic non-branded traffic was down only 6%. The bleed was concentrated in what their attribution tool called “local and direct,” which is usually code for Google Business Profile, Apple Maps, and the long tail of citation sites that feed voice assistants and embedded map widgets.
Initial hypothesis vs what we actually found
My working hypothesis going in was a Google Business Profile (GBP) policy issue. Q3 2025 had been rough for HVAC profiles; Google had tightened service-area business verification, and I had three other clients in the same vertical who got hit. I expected to find suspended or restricted profiles. I was wrong on the cause but right on the symptom location. What I actually found was a slow accumulation of NAP (name, address, phone) drift across the citation graph from the two acquisitions, compounded by phone numbers that had been changed during a call-tracking platform migration in June. Twenty-two percent decline, no single cause, dozens of small ones.
Did you know? According to Birdeye’s analysis, fully populated, verified profiles surface 80% more often in search and generate roughly four times more website visits than incomplete or unverified listings. The penalty for inconsistency is not linear; it stacks.
Pulling the listing inventory
Why I start with a citation crawl, not Google Business
Most agencies open Google Business Profile Manager first because that is where the dashboard lives and the data is cleanest. I do the opposite. I start with a citation crawl across the wider directory ecosystem because GBP shows me what the client thinks is true; the citation graph shows me what the internet thinks is true. The gap between those two is where the revenue leaks.
For this audit I used a combination of Yext’s audit export (the client still had a legacy seat), a separate run through BrightLocal for an independent second opinion, and a manual pull from Apple Business Connect because automated tools still miss Apple inconsistencies more often than vendors admit. The total surface area: 47 locations across an average of 31 surfaceable directories each, so roughly 1,457 listings to reconcile. The actual number after deduplication was 1,612, because the acquisitions had left ghost listings behind.
The 11 directories that actually moved their needle
If you have ever read a “top 50 citation sites” list and felt your soul leave your body, you are not alone. In 2026, the directories that actually drive measurable outcomes for a US HVAC business are a much smaller set than the vendor lists suggest. For this client, after correlating citation accuracy with call-tracking data over a 90-day window, eleven sources accounted for 94% of trackable listing-driven contacts.
| Directory | Share of tracked contacts | Update latency | Duplicate risk | Effort to fix |
|---|---|---|---|---|
| Google Business Profile | 61% | Hours to days | Medium | Low |
| Apple Business Connect | 14% | 1-3 days | High | Medium |
| Bing Places | 7% | 3-7 days | High | High |
| 5% | Immediate | Medium | Low |
The remaining seven, in descending order of contact share, were Yelp, BBB, Angi, HomeAdvisor, Nextdoor Business, Foursquare (still feeding Tripadvisor and a few embedded widgets), and Data Axle. Everything else, including the long tail of niche directories, drove the residual 6% combined. I am not saying ignore them, but I am saying do not let them set your priority list. For most service businesses, a curated, well-maintained presence on a handful of strong directories beats a sprawling mess of 80 listings you cannot keep current; if you are operating at single-location scale, a vetted general directory like Business Directory alongside the eleven above will outperform a scattershot approach every time.
Spotting zombie listings from acquisitions
The acquisitions had left behind 84 zombie listings: profiles for the prior business names, still indexed, sometimes with old phone numbers, sometimes redirecting customers to addresses that were now empty offices. The fastest way to find these is not a tool. It is a spreadsheet of every legal entity the company has ever traded as, every prior phone number, every prior address, run as exact-match queries across Google, Bing, Apple, and Yelp. I found seven zombies that no automated tool flagged because the address had been updated but the business name had not.
Myth: A good citation management tool will find every duplicate and inconsistent listing automatically. Reality: Tools find roughly 70-85% of issues at scale. The remainder, often the most damaging ones, require manual hunting with prior business identities as your search terms.
Triaging inconsistencies by revenue impact
Scoring NAP variants against call tracking data
Once I had the inventory, I had 1,612 listings and 893 distinct NAP variants. You cannot fix 893 things in any sensible timeframe, and worse, you should not try, because most of them do not matter. The triage step is where audits earn or lose their fee.
I scored each variant on three dimensions: the volume of contacts that directory drove in the past 90 days (from call tracking and UTM-tagged links), the visibility of the inconsistency to a customer (a wrong suite number is worse than a missing “Inc.”), and the propagation risk (does this directory feed others). The scoring is rough; I am not pretending it is science. But it gives you a defensible order of operations rather than a panicked sweep.
Why I ignored 60% of the discrepancies
Roughly 538 of the 893 variants were things like “Street” vs “St”, missing commas in the address line, or trailing whitespace in business names. They are technically inconsistent. They do not measurably affect rankings or customer behaviour in 2026, because the major data consumers (Google, Apple, Bing) have been normalising these forms for years. Spending time on them is a way to look busy. I batched them for a quarterly cleanup run and moved on.
The 355 variants that mattered fell into three groups: wrong phone numbers (mostly fallout from the call-tracking migration), wrong or outdated suite numbers at three locations that had relocated within the same building or complex, and category mismatches where acquisition-era profiles still listed the business as “air conditioning contractor” only, missing the heating category that drove most Q4 search demand.
The two locations that needed emergency intervention
Two locations were on fire. One had a duplicate GBP that ranked above the official profile for branded searches; the duplicate had a five-year-old phone number that rang at a defunct answering service. Direct revenue loss, every day, until fixed. The other had been merged with a competitor’s profile during a Google data refresh (this still happens, it is maddening), so the address pin showed the competitor’s location three miles away. I escalated both through Google’s business redressal complaint form and via a paid Google support contact, and resolved them in 11 and 19 days respectively. Everything else waited.
Quick tip: Before you start fixing anything, snapshot every listing as it currently appears. Screenshots, exports, the lot. When something breaks two months later (and something always breaks), you need a baseline to compare against. I keep a dated folder per audit and have never regretted it.
Working through the fix sequence
Why Apple Business Connect went first this time
Conventional wisdom is to fix Google first because it drives the most traffic. I usually agree. This time I started with Apple Business Connect, for two reasons. First, Apple’s update propagation through to Siri, CarPlay, and embedded Maps widgets had been running 24-72 hours, which was faster than Google’s at the time for this account (Google had been slow-walking edits on legacy profiles flagged by their 2025 verification sweep). Second, 14% of tracked contacts came through Apple’s surfaces, and the iOS share of the client’s customer base skewed higher than average because the service area included two affluent suburbs.
I fixed Apple over five working days, watched the call-tracking data, confirmed no breakage, and then moved to Google. The lesson is not “always start with Apple.” It is “start with the directory where you can move fastest and confirm the fix worked, so you build evidence before tackling the messier surfaces.”
sankey-beta All Listings,Google Business,61 All Listings,Apple Maps,14 All Listings,Bing Places,7 All Listings,Facebook,5 All Listings,Yelp & Angi,7 All Listings,Long Tail,6 Google Business,Contacts,61 Apple Maps,Contacts,14 Bing Places,Contacts,7 Facebook,Contacts,5 Yelp & Angi,Contacts,7 Long Tail,Contacts,6
Handling the duplicate suppression on Bing
Bing Places is the directory that practitioners love to mock and clients love to ignore, but for this HVAC client it drove 7% of tracked contacts, mostly from Windows desktop searches by older homeowners replacing furnaces. The duplicates were ugly. Microsoft’s duplicate suppression workflow in 2026 still requires a manual claim, a verification call to the listed phone number, and a wait period of 5-10 business days. For 47 locations with an average of 2.3 Bing listings each, that is a small project on its own.
I batched the work, assigned it to a coordinator with a spreadsheet, and accepted that this would run in the background for a month. Some audit deliverables are sprints. Some are slow drips. Be honest with the client about which is which.
Negotiating with aggregators that still matter in 2026
The big data aggregators (Data Axle, Localeze/Neustar, Foursquare’s data feed) are not the force they were a decade ago, but they still feed enough downstream sites that ignoring them lets bad data regenerate. I submitted corrections to all three. Data Axle’s response was fast and clean. Localeze took three rounds because their automated verification kept flagging two locations as “unverifiable” against their existing record. The fix in those cases was to provide a utility bill, which felt very 2015 but worked.
What if… the client had refused to pay for aggregator submissions and insisted I only fix Google and Apple? Based on the data, I estimate they would have recovered about 75% of the eventual gain, but the bad data from aggregators would have re-poisoned 15-20% of secondary directories within nine months, requiring a re-audit. I have run that experiment unintentionally on two prior clients. The aggregator step is unglamorous and worth it.
Measuring what changed
The 38% recovery in direction requests by week 6
Six weeks after the bulk of fixes landed, direction requests across the 47 locations were up 38% compared to the pre-audit baseline. Phone calls were up 19%. Website visits from GBP were up 27%. Total inbound leads, measured the same way as the original 22% decline, had recovered by 16 percentage points, leaving a residual 6% gap that I attributed to genuine seasonal and market factors rather than listing issues.
I want to be careful here. Recovery of 38% in direction requests is not the same as a 38% revenue uplift. Direction requests are a leading indicator. The revenue picture took longer to clarify because HVAC sales cycles for replacement work run 2-6 weeks from first contact to booking. By week 12, booked job revenue from local search sources was up 31% versus the Q3 trough. Some of that is the audit. Some is normal Q4 seasonality. I would credit the audit with roughly 22 of those 31 points based on the year-over-year comparison, but I would not stake my reputation on the precision of that split.
Did you know? Birdeye’s analysis attributes 48% of Google Business Profile interactions to website visits, 34% to direction requests, and 17% to phone calls. The mix varies by industry; for service businesses like HVAC, plumbing, and legal, the phone call share runs noticeably higher than the aggregate.
Leading indicators I track before revenue catches up
Revenue is a lagging indicator. If you wait for it to confirm your audit worked, you will have spent six weeks staring at noise. The leading indicators I track, in the order they typically respond, are: impression share for branded queries in GBP Insights (responds in 3-10 days), direction requests (1-3 weeks), category-relevant non-branded impressions (2-4 weeks), and review velocity (4-8 weeks, because customers have to actually receive service before they leave reviews). If the first two move and the third does not within a month, something else is wrong, usually content or category configuration rather than NAP.
What didn’t improve and why
Two things did not improve. First, review volume stayed flat for the first eight weeks because the audit did not touch review acquisition workflows, and the client’s existing process was sending requests to the wrong contact field for about 30% of customers. That is a separate project and I flagged it. Second, three locations in one suburban cluster continued to underperform regardless of how clean their listings were. Investigation revealed a competitor had opened two new locations in the catchment area during Q2. No amount of listing hygiene fixes a genuine competitive shift. The honest answer to the client was that those three locations needed a different conversation entirely, probably about service differentiation and paid search.
How this changes for smaller operators
Single-location version of the same playbook
If you run a single-location business, the framework is identical but the scale collapses. You are auditing maybe 30-40 listings rather than 1,600. The eleven priority directories are still the same eleven, weighted slightly differently for your industry. The triage step is faster because you can hold the variants in your head. The whole audit, end to end, is a one-day exercise for someone who knows what they are doing, or a weekend for someone learning. The mistake single-location operators make is treating it as a one-time event rather than a quarterly check.
When a $200 tool beats a $2,000 one
For the HVAC client, an enterprise stack made sense. For a single dental practice, it does not. Tools like Uberall’s free listings audit or BrightLocal’s mid-tier plans will surface 80% of what a full Yext deployment would, at a fraction of the cost. The remaining 20% is mostly enterprise governance features (approval workflows, multi-user permissions, API access) that a single location does not need. I have seen small businesses pay for enterprise listings management out of vague anxiety, and it almost never pays back. Spend the saved budget on review acquisition instead; that is where the compounding returns sit.
| Operator size | Suggested tool tier | Annual tool cost (approx) | Audit frequency | Time per audit |
|---|---|---|---|---|
| 1 location | Free or entry-level | $0-300 | Quarterly | 4-8 hours |
| 2-10 locations | Mid-market SaaS | $1,200-4,000 | Quarterly | 2-4 days |
| 10-50 locations | Mid-market + manual review | $8,000-25,000 | Twice yearly + monitoring | 2-3 weeks |
| 50+ locations or franchise | Enterprise platform | $30,000+ | Continuous | Ongoing |
The timeline shift for franchises and multi-brand portfolios
Franchise audits run longer for a reason that has nothing to do with technology: governance. In a franchise system, you cannot just edit a franchisee’s listing without going through whatever brand standards process exists, and franchisees vary enormously in how much they care. I have run franchise audits where 40% of locations refused to implement fixes because the local owner did not believe the corporate team. Plan for 6-12 weeks for a 50-location franchise audit, mostly because of the human coordination, not the technical work.
For multi-brand portfolios (a holding company with several different brand identities), the additional complication is that each brand needs its own audit. Do not try to do them in parallel unless you have separate teams. The categories, customer language, and competitive context differ enough that switching context between brands produces sloppy work. I learned this the hard way on a portfolio with three restaurant concepts.
Myth: Once you have done a thorough audit and fixed everything, you can leave the listings alone for a year. Reality: Listings drift constantly. Google rolls out attribute changes, employees update phone numbers without telling marketing, aggregators re-introduce stale data, and competitors occasionally try to claim profiles that are not theirs. Continuous monitoring is the only stable state.
Principles I take into every audit now
Treat listings as distribution, not SEO
The mental shift that took me too long to make is that local listings are a distribution channel, not an SEO tactic. When you frame them as SEO, you obsess over rankings and signals and theoretical citation strength. When you frame them as distribution, you ask the questions that matter: where is my customer looking, what do they see, can they contact me, and is the experience good when they do. The HVAC audit succeeded because we measured contacts and revenue, not rankings. I have seen audits that improved rankings without improving revenue, and they are worse than useless because they convince clients the work is done.
Audit cadence depends on churn, not calendar
Quarterly audits are a sensible default. They are not always right. A business with high operational churn (staff changes, location moves, phone system migrations, acquisitions) needs more frequent checks, possibly monthly. A stable single-location professional services firm with no recent changes can probably stretch to twice yearly and be fine. The right question is not “how often should I audit” but “how often does my underlying business data change in ways that affect listings”. Answer that, then set cadence to match.
The AgencyAnalytics audit framework is honest about this: audits are time-consuming, and over-auditing wastes both agency hours and client patience. Calibrate to actual risk.
The one check most agencies still skip
This is the one I want you to take away if you take nothing else. Check the phone numbers on your listings by calling them. Not by looking at them. By calling. From a phone that is not associated with the business, ideally at a moment when the business is open. About one in twelve audits I run surfaces a phone number that looks correct on the listing, was not flagged by any tool, and rings to a wrong destination, an old call-tracking number, a disconnected line, or, on one memorable occasion, a competitor’s call centre. Tools cannot catch this because the string of digits is right. A human dialling the number catches it in thirty seconds.
For 47 locations, that is roughly four hours of someone’s time and at least one or two saved revenue leaks. For a single location, it is six minutes. There is no excuse not to do this, and almost nobody does it. If you start one new habit after reading this, make it that one. Then go pour another coffee and start your inventory pull.
