Updated November 2, 2026 — added direct answers on homeowner AI search paths, machine recommendations, and visibility investment vs paid ads, from our own published work.
The tactics in this playbook are the same ones we publish for every market: be readable to the machines, be named, be cited, measure it with the denominators showing. What changes in Puget Sound is that Puget Sound is not one market. It is at least six, stacked on top of each other along one highway — and the move that wins in Tacoma wastes money in Everett, while the move that matters most in Edmonds barely registers in Seattle. This is the local edition: the same doctrine, market by market, with our own measurements where we have them and honest labels where we don’t.
It has two halves, because the playbook was always two books bound together. The first half is visibility: how the answer engines decide whose name a flooded homeowner hears. The second half is the specialty work — the fire extinguisher tag, the records room, the server rack, the art on the lobby wall — the plays that get a restoration company inside commercial buildings before anything breaks. Same Sound, same six markets, same rule in both halves: a tactic only earns its place here with a local reason, and where there is no local reason, we say so instead of swapping in a city name and calling it local.
Every number below that carries a percentage was measured on our rig on October 18, 2026: real homeowner questions, asked mid-flood style — “who should I call,” “who’s best,” “who has the most reviews” — put to ChatGPT, Perplexity, Gemini, and Google AI Mode, with every answer scored for who got named and whose pages got cited. Google AI Overview returned no answer on any question in these batteries and is excluded everywhere below. A snapshot is not a census. It is one dated sample, with its denominators showing, and it is still the only honest way we know to talk about this.
The doctrine, before the geography
Named and cited are two different jobs. An answer engine can use your page as a source and still recommend somebody else — we have watched it happen, and the section on Everett below shows the receipt. Being named is the engine volunteering your company in the answer text. Being cited is your page appearing in the source list underneath it. The first job gets you the call. The second job is how the first one eventually happens. Any report that blends the two into one “visibility score” is hiding the only distinction that matters.
Your website is a minority of what the engines say about you. The frame we use on our own offer comes from McKinsey’s research: a brand’s own website is a small share of the citations about that brand. The rest comes from the pages the models already trust — directories, association profiles, local references, roundup pages. That is why the work is the service pages, the insurance and certification language, and the outside profiles, in that combination. A beautiful website with thin directory presence is a company the engines can describe but rarely do.
The competitor set is fixed, visible, and smaller than you think. Ask an engine who handles water damage in a city and it names one or two companies, out of a frame of maybe eight that ever appear. The frame repeats. The same names show up answer after answer, which means the game is not “rank higher” — it is “get into the frame, then take share inside it.” You can read your frame any time you want. We run a fixed list of prompts for exactly that reason.
Reviews are machine food now, and the machines cross-check. The engines quote review counts back at homeowners — “over 160 five-star Google reviews” is the kind of clause that decides an answer. But the footprint has to agree with itself across the surfaces the engines read. A deep review base on one surface and a thin, stale profile on another reads to a model the way a half-finished job reads to an adjuster: possible, but not the safe recommendation.
Each engine reads different sources, so “AI visibility” is not one number. In our measurements, Perplexity and Google AI Mode cite Yelp in roughly one answer in six; Gemini has never cited Yelp in our rig at all — zero of 35 answers. A tactic that feeds Perplexity may do nothing for Gemini. Anyone selling you a single score for all of AI is averaging away the part you could have acted on.
Directories are citation infrastructure, not paperwork. Across 143 answers measured the same week, the directory order was BBB first, then Thumbtack, HomeAdvisor, and Yelp fourth — 21 Yelp citations, ahead of Expertise and Angi. When a homeowner asks an engine whether a specific company is any good, the engine goes to that company’s own directory pages: in all three Puget Sound branded checks we ran, the company’s Yelp business page was the page cited. Claimed, complete, current is not hygiene. It is the asset.
Measure first, fix the biggest gap, re-measure. Every engagement we run starts with the same sequence: a fixed prompt list, scored answers, one highest-leverage fix, then the same list again after the fix so the movement is provable. The fix is usually unglamorous — a service page that does for one city what another of your pages already does for a different city, a directory profile completed, a name collision cleaned up. Small repairs, precisely aimed, beat a website rebuild aimed at everything.
Seattle metro: the saturated frame
Seattle is the hardest frame in the region to enter, because it is the one market here where franchise density and review volume stack on top of each other. When we asked ChatGPT which Seattle-area water damage company has the most reviews, the answer crowned listing names — shops whose review counts the engine could read and repeat — and never reached a local company holding 100+ five-star reviews of its own. In a saturated market, the engine’s shortlist is built from whatever is most legible at scale, and legibility at scale is what national brands manufacture for a living.
The tactic that matters most in Seattle is citation capture on the questions franchises answer sloppily: the specific ones. “Who handles sewage cleanup in a condo building” is a different frame from “best water damage company,” and it is thinner. The tactic that matters least is another generic service page saying you do water damage in Seattle — the frame already has that page, written by a brand with two thousand locations. In Seattle you do not out-shout the wall. You find the doors in it.
The Eastside: commercial money, quiet homeowners
Bellevue, Redmond, and Kirkland run on a different economy than the rest of the Sound: higher-value homes, a deep commercial and property-management layer, and franchise-owned restoration territory. We have not run a dedicated Eastside measurement battery yet, so this section is doctrine, labeled as such — it will get its own snapshot before we treat it as anything more.
The doctrine says the Eastside rewards the commercial signals: insurance billing fluency, documented response time, capacity language a property manager can forward to an owner. The buyer asking the question is often not the homeowner at all. It is the manager of a flooded HOA or a warehouse, asking an engine who can handle it — and the citation has to prove commercial capability, not just residential warmth. Review volume matters less here than it does in Seattle; a smaller, current, consistent profile across the directories beats a big stale one, because the questions are fewer and the reader is checking, not browsing.
Tacoma and Pierce County: the winnable frame
Tacoma is where we can prove a local leader can own serious share of the frame. In our October 18 battery — seven unbranded questions, 27 answers that named a company — the leading independent was named in 41% of answers. That is a real seat at the table. But the same battery shows the split that decides everything: the independent was cited in only 11% of answers, while SERVPRO was named in 59% and cited in 52%. The engines know the local company exists. They name it from memory and cite the franchise from evidence.
The Tacoma battery also produced the single most instructive answer in our Puget Sound work. On Perplexity, asked for the best water damage company in Tacoma, the answer opened by calling SERVPRO of Tacoma “the strongest overall choice” — in the same market where the independent’s own Tacoma page ranks first on Google for that question. Winning Google and winning the answer are now separate competitions, and Tacoma is where the gap between them is easiest to close: the naming is already there, so the work is the cited evidence underneath it — the service pages, the directory records, the outside pages Perplexity actually reads.
One more Tacoma lesson, and it is about humility at the engine layer: asked about a burst pipe in Kent — where the leading independent is headquartered — Gemini hedged on whether Kent was in Washington or Ohio, and named SERVPRO. Engines get local geography wrong in ways a human never would. Part of the local play is feeding them geography so plainly stated, in so many consistent places, that the machine cannot confuse your city with one in Ohio.
Everett and Snohomish County: the wall, and the split
Everett is the wall. In our battery — seven unbranded questions, 28 answers — SERVPRO was named in 71%. The leading independent, holding the deepest review base of any independent in the corridor (682 Google reviews, displayed on its own website), was named in 29% and cited in 18%. That is not an unknown company being ignored. That is a known company losing to a frame: SERVPRO, PuroClean, Roto-Rooter, and a rotation of local names divide the answers, and the independent’s share is real but third-tier.
Everett also gave us the cleanest proof in the region that cited and named are different jobs. On one Gemini answer, the source list included the independent’s own Everett page — the engine read it, used it, and then recommended three other companies by name. If your reporting only counts citations, that answer looks like a win. If you sell water damage work, it is a loss with a citation attached. The Everett fix is therefore not “get cited more.” It is converting existing citation into naming: the answer assets, the directory records, and the review surface have to say the same name the page already earns.
Asked about the company by name, all four engines named and cited it — four of four. That pattern repeats in every Puget Sound battery we have run, and it is the most hopeful fact in this playbook: nobody we have measured is unknown to the machines. They are un-volunteered. The distance between branded and unbranded is the entire business.
Edmonds and the small cities: branded invisibility
Edmonds shows what happens in a small-city market when an established local company never built the second job. In our battery — 27 answers on unbranded questions — the Edmonds-headquartered company was named in 26% and its website cited in 7%, while SERVPRO took 70% of namings. Gemini named the company zero times in six answers, including both questions about Edmonds, its home city. And on the four questions that named the company directly, every engine named it back — while not one of the four cited its website. An engine that will confirm you exist but never send a reader to your page is a directory that has your name and not your number.
The small-city lesson cuts the other way too, and it is the best news in this playbook for independents. The expected national rival in Edmonds was not SERVPRO’s shadow brand — 1-800 WATER DAMAGE was named in just 4% of answers and cited in none. In small markets, the second franchise seat is often empty. The wall is one brand deep. A local company with a complete directory presence, consistent reviews, and one strong city page is not fighting a category — it is fighting a single incumbent’s local footprint, and footprints can be out-built.
Small-city tactics therefore invert the Seattle order. In Seattle, directories are table stakes under a mountain of franchise content. In Edmonds, Lynnwood, Mountlake Terrace, and their kin, the directory record and the branded page are the game: they are what the engine has to read, and there is little else competing for its attention. Completeness beats cleverness in a thin market.
Kitsap and the rural edges: thin ice, honest label
West of the ferries and out past the suburbs — Bremerton, Port Orchard, Poulsbo, and the rural edges of the counties — we have not run a measurement battery yet. This section is doctrine, and it stays labeled that way until a Kitsap snapshot exists.
The doctrine is encouraging. Rural and small-market frames are thin: fewer franchises with real local pages, fewer directories competing, fewer answers being generated at all. In a thin frame, one complete, consistent presence — claimed Google profile, complete Yelp page, BBB record, one service page per town that says plainly what you do and where — can outrank a franchise’s tenth location page, because the engine assembles its answer from whatever exists and there is not much. Word of mouth still closes most rural work; the engines fill the silence for everyone who moved there last year and does not have a neighbor to ask. Being the one full record in a thin file is the cheapest seat in this whole playbook, and the first shop in each rural market to take it seriously will be hard to dislodge.
Part Two: the specialty playbook, localized
The first half of this playbook is about being named. This half is about being inside the building before anything breaks. Water mitigation is the call every facility already has a vendor for; the specialty plays are the door past that vendor — the work aimed at the assets an adjuster cannot simply cut a check against. The model, as we publish it: you do not perform the specialty work yourself. You hold the agreement, you stabilize in the first hours, you coordinate a pre-qualified specialist bench, and the chain of custody is the product. What follows is each play as published, then where in the Sound it actually earns — and where it doesn’t.
The fire extinguisher tag: the compliance door
Expired extinguisher tags are a real compliance problem — fire marshals, insurers, and corporate safety programs all care — which is why a manager who ghosts every salesperson takes this meeting. You run the inspection as a real service, walk the building with a restoration operator’s eyes while you do it, and leave a written report with your after-hours number in the manager’s file. The ethics are the strategy: if the tags are fine, you say so and leave. The play earns where buildings stack and one manager covers several addresses — downtown Seattle’s office and hospital stock, the Eastside’s office parks, the multifamily corridors running down to Tacoma. It does not earn in rural Kitsap, and the published play itself says why: it dies aimed at buildings too small to matter. On the far side of the ferries the tag check rides along on other work, or it doesn’t run.
The Cintas lesson: cadence is the product, and the route rep is a channel
Cintas looks like uniforms; what it actually sells is a weekly stop inside a multi-year auto-renewing agreement. The restoration translation we publish is the emergency response agreement with scheduled walkthroughs and a quarterly review — sell the stop, not the emergency, and stack accounts on one corridor so a crew can work a loop instead of chasing singles. The second half of the lesson is who carries your name into buildings: the route reps of Cintas, Aramark, and their regional competitors are inside a hundred or more commercial properties a week, they know every property manager by name, and almost nobody in restoration talks to them. That channel runs on the I-5 corridor exactly the way the doctrine wants — Seattle to Tacoma is one long route. In Kitsap the ferry splits the corridor in two, and both the national reps’ books and yours have to be built as separate loops. Density is the whole mechanic; where the water cuts the route, the mechanic halves.
Documents and records: sell to the records manager
Paper is the first specialty build, because it is the most frequent specialty loss and the most time-sensitive: wet paper molds in forty-eight to seventy-two hours. The response is freeze first, chamber second — vacuum freeze-drying at a specialist like Polygon or Document Reprocessors — and the restoration company’s product is everything around the chamber: the first-twelve-hours stabilization, the packout, the frozen transport, and a chain-of-custody package the customer can hand to their carrier, their regulator, and their lawyer. The buyer is not the facilities director. It is the records custodian — the office manager, the records clerk, the information governance director — the person whose job becomes the records problem the moment water hits the file room.
Seattle holds the densest records map in the region: a major research university and its medical system, county and municipal archives, and a downtown full of law and accounting firms whose files are their business. Tacoma runs a close second with the county seat, a large health system, and three universities. Everett’s version is the county government and its hospital system. The Eastside’s records sit inside corporate headquarters — but the biggest campuses run their own machines, so the sellable layer there is the professional firms around them, not the logos. Edmonds is the honest small version: a hospital’s records and a city hall’s, real but counted in ones and twos. And Kitsap carries a limit worth printing: much of its records mass sits behind federal fence lines at the shipyard and the naval base, which procure through federal channels. The local play is the county, the cities, the school districts, and the clinics — not the fence.
Electronics: the seventy-two-hour corrosion window
Circuit boards have a faster clock than paper: acidic soot starts corroding traces inside a day, and the recoverable window closes around seventy-two hours. The restoration company’s work is the stabilization — a controlled power-down coordinated with the client’s own engineers, humidity driven below forty percent, exposed boards neutralized, every serialized unit triaged and logged — and the specialist’s ultrasonic tanks do the rest. The Eastside is this region’s electronics capital: the Redmond and Kirkland tech campuses concentrate more servers, labs, and custom-configured equipment than anywhere else in the Sound, and none of it can be re-ordered on Tuesday. Seattle’s version is data centers, broadcast facilities, research labs, and the port’s logistics systems. Tacoma’s is industrial controls around the port. Everett’s is the supplier ring around the great plant — the plant itself buys at national scale, but the shops that feed it don’t. In the small cities, electronics and medical are the same room: the imaging suite. And Kitsap’s heaviest electronics sit behind the same fence as its records, which is why the realistic Kitsap bench is the contractor and supplier shops outside the gates in Bremerton and Silverdale.
Technical drying: the engineered end of drying
No piece we publish carries this exact label; the doctrine underneath it is the structural-drying science — psychrometrics, desiccant versus refrigerant dehumidification, equipment sized to the chamber instead of the truck — and the discipline of getting paid on the data that proves the dry. As a play it is simple: on a big footprint, documented engineered drying is the version of the work that survives an adjuster’s scrutiny. It follows footprint, not population. Port warehouses in Seattle and Tacoma, plant-scale buildings in Everett, hospital campuses anywhere — those are its ground. The Sound adds one local argument: marine air. Structures on the water, and the vessel-adjacent work around the marinas, live at humidities that make desiccant the default rather than the upgrade. In the small cities and the rural edges there is no standing market for it — the capability shows up there as proof on the occasional school or gymnasium loss, not as a line you staff. That difference is written here instead of a paragraph pretending otherwise.
Fine art and collections: stabilize, document, hand off
The art play has the humblest job description in this book: you do not treat the art. The conservator — often designated by the carrier — determines treatment. You stabilize in place, document photographically, isolate the piece from the ongoing damage, and make the handoff clean. What you are selling is a competent first twenty-four hours. Seattle’s anchors are its museums and corporate collections. Tacoma owns a signature class of its own — glass, with a museum built for it and an art museum beside it. The Eastside’s version is the region’s one residential exception to the rule that specialty is commercial: the lakefront estate belt, where the collections hang in houses, and where the published doctrine already says luxury residential is the only residential segment where this economics works. Edmonds proves the play scales down — a small art museum with a real collection is still a collection. Everett’s art center is real too, and it is a line in this paragraph rather than a market, because that is what it is. Rural Kitsap, past one island museum, has no anchor institution to build on, and this play does not pretend to one.
Medical and laboratory equipment: the regulated room
The fourth published category is the one most restoration companies avoid, which is precisely its value: equipment that cannot return to service until the manufacturer recertifies it, worked under infection-control discipline, approved through risk management and biomedical engineering on cycles that run sixty to ninety days. Every market in this playbook has the anchor — the regional trauma center and university medical system in Seattle, hospital systems in Tacoma, Everett, Bellevue, and Silverdale, and a full hospital in Edmonds, which is why a town of twenty thousand supports a regulated-room play at all. What changes by market is not the need. It is the length of the approval path, and the operator who knows that going in prices the patience correctly.
Documentation-only: sell the record
The newest play in this book inverts the usual sale. Some losses will be remedied by the property manager’s own crews or the owner’s contractor — that decision is already made, and fighting it wins nothing. What that loss still needs, inside the first forty-eight hours, is a record that survives the adjuster: the two-audience photo protocol, the moisture map and readings log, the scope narrative, the pre-existing condition log — the packet our published doctrine has always treated as the real product, now sold on its own, flat per loss. The liability line is the whole design: the restorer sells a record, never a certification of somebody else’s remedy. The record states what was observed. It does not state what should be done.
Inside a pre-signed emergency services plan, the play runs at three levels, and a property manager mixes them across a portfolio property by property. Document from the beginning is the record alone — its commercial function is buying the decision-maker time while the evidence is secured. Hold conditions adds preservation and stabilization during the decision window, short of full mitigation — the clocks get slowed, nothing gets solved, and the file says exactly that. Emergency respond is the full response with documentation built in from minute one instead of reconstructed afterward. Which level sells where follows the same density law as everything else in this book: the record-only level sells where portfolios concentrate — Seattle and Eastside property management, owners and general contractors who self-perform — because a buyer with forty doors buys documentation forty times. Hold conditions sells where decision windows are structurally long: institutional owners, carrier-approval chains, the management companies working the military corridor outside Tacoma. And where the density isn’t there, the honest test replaces the pitch: a buyer needs enough doors, losing often enough, to buy records on purpose — in the rural edges that buyer may not exist yet, and the play waits until one does. The cross-reference to the first half of this playbook is direct: a documentation client is a relationship that recurs on every loss, including the losses you were never going to win. The company that holds the record is the company the building already knows when the big one comes.
How the sequence runs, in any of these markets
The order of operations does not change with the zip code. Measure the frame for your city with a fixed list of the questions your customers actually ask, scored for named and cited separately, engine by engine. Read who is in the frame and which outside pages the answers are built from. Fix the single biggest gap first — it is usually a page, a profile, or a name consistency problem, not a budget problem. Then run the same list again and keep the rate as a trailing number, because frames move when a competitor publishes or a model updates.
The specialty half runs its own sequence, and it is just as fixed: build the bench before you sell the category, sign the agreement before the loss, drill the response before the call comes. Sell only the specialties on your bench. Add a category when the teaming agreement is signed, not when the brochure is printed.
That is the whole practice. The Sound just makes it interesting: six markets, six different first moves, one discipline. If you want your frame measured, the free mini-check asks one real question about your company and reports back what the answer said. The full scan runs your city’s whole list. Either way, you will know your frame — and in Puget Sound, knowing which of the six games you are actually playing is most of the game.
How do homeowners find restoration contractors through AI assistants?
Homeowners finding a restoration contractor through an AI assistant follow a three-stage conversational sequence rather than scrolling through a page of blue search links:
- The Ask: The homeowner enters an unbranded, situational question at the moment of discovery (e.g., “Who do I call for water damage in my basement?” or “Best water restoration company near me”).
- The Shortlist: The assistant synthesizes a small, situational winners’ circle—typically naming three to five local operators based on verified entity data, physical proximity, and operational availability.
- The Deciding Signals: The assistant evaluates technical certifications (such as IICRC credentials), review recency and problem-specific sentiment, and documented emergency dispatch capability to construct its recommendation.
In our published doctrine, being “named” on that shortlist and being “cited” as an authoritative source are two distinct jobs: an answer engine can quote your technical drying guide as evidence while recommending a competitor for the physical dispatch, unless your local entity presence and service boundaries are fully verified.
How can a local restoration company get recommended by ChatGPT?
To get recommended by ChatGPT and other conversational answer engines, a local restoration contractor must execute a fixed order of operations:
- Measure the Local Frame: Benchmark your market using a fixed battery of unbranded questions property owners actually ask in your city, scoring being named and being cited separately across each primary engine.
- Standardize Your Entity Footprint: Eliminate cross-platform naming, address, or phone discrepancies across primary local directories, mapping platforms, and search profiles to establish unambiguous entity clarity.
- Publish Verifiable Operational Truth: Explicitly document your geographic service boundaries, verified 24/7 emergency dispatch availability, and technical credentials on your digital properties so machines parse them as operational facts.
- Track Citation Rates Dynamically: Monitor your recommendation rate over time, recognizing that local frames move as competitors publish field content or engine models update.
As demonstrated in our published field experiment in Puyallup, winning recommendation share in competitive local markets is not a budget contest against national franchises; it is won by local operators whose verified operational rubric matches what the machine requires to name a contractor.
Should a restoration company invest in AI search visibility or stick with Google Ads?
Deciding whether to invest in AI search visibility or rely on Google Ads comes down to the structural difference between owning an authority asset and renting temporary placement. Paid ads purchase immediate, auction-dependent clicks where visibility vanishes the moment ad spend ceases. In contrast, AI search visibility builds durable machine authority—positioning your company directly inside zero-click answers and conversational shortlists.
However, honest operators recognize the current limitations of AI referrals: unlike paid search campaigns that offer predictable click-through volume, AI assistant recommendations exhibit shortlist behavior, where secondary contractor names rotate based on situational variables, leading to less predictable short-term lead volume. The strategic approach is not an either-or replacement: paid placement captures immediate demand, while investing in structured entity authority, technical documentation, and citation tracking builds an enduring asset that captures the growing majority of zero-click inquiries. For methodology on auditing where your domain stands, see our published framework on how we track AI citations.
Measurements: Tygart Media citation snapshots, October 18, 2026 (Everett/Snohomish, Tacoma/Pierce, Edmonds/Seattle batteries; 143-answer directory measurement, same week). Rates are one dated sample with denominators shown, not rankings or promises. Companies are named only as they appear in those published measurements. Specialty doctrine: Tygart Media’s published specialty cluster (the fire extinguisher tag strategy; document and records recovery; electronics restoration; the specialty accounts pillar) and Restoration Intel’s recurring-revenue and specialty-recovery pieces, localized as marked — doctrine is labeled doctrine wherever no local measurement exists yet.
