In most of this county the person typing into Google is not a shopper. She is a coordinator with a start date, a budget line and four names to call — and a report built for shoppers will tell you almost nothing about her.
Search Console has been recording her visits since the day the property was verified, free, refreshed every morning. What it returns is a ledger, not a verdict. The distance between those two things is where the wasted quarters happen.
A county of suppliers, not storefronts
Count the invoices moving through this basin on any Tuesday. A grip truck out of Sun Valley for eleven days. Forty thousand blanks cut and sewn in Vernon for a brand headquartered elsewhere. A drayage run booked out of Wilmington because a container came in short. A machine shop in Torrance quoting a bracket against a drawing. Clinics and restaurants exist too and behave like consumer businesses anywhere, but they are not what this county runs on.
Everything on that list has the same shape. The customer is assembling something temporary: a shoot, a production run, a build, a case, a delivery window. She is filling a slot in a plan that already has a date on it, and when the slot is filled she stops looking.
The clock is inside the query
Somebody with a call sheet for the eighteenth is not browsing. She types a specification and takes the first credible answer that confirms availability.
- Second place converts far worse than the position gap suggests
- Hiding capacity behind a contact form loses the job outright
Price is the last filter, not the first
She screens on capability, capacity and whether you can be there. Rate comes up once you survive that screen.
- Cheap-and-fast copy screens you out of serious work
- Sizes, tonnages, turnaround and certifications do the selling
Hold that reader in mind through every screen below. Read on retail assumptions — volume good, position good, more impressions good — and the same account will rank your best month below your worst, with numbers that are all accurate.
Your own log versus the public page
The panel puts Search Console analytics and rank tracking a click apart, and they look like two views of one thing. They watch different objects. When they disagree, most teams decide one of them is broken; both are usually right about the question they were built for.
Search Console is Google's record of your property: every time something of yours went in front of somebody, and every time that somebody picked it. Definitive about you, silent about everyone else. When a post house in Burbank takes a term you held for a year, no flag appears — the line bends and you invent a reason for the bend.
Rank tracking reads the same results page from outside, signed out, under controlled conditions. So it can name the domains above you, attach Domain Authority to each and count the keywords you hold in common — and it sees nothing past the click. Eight analytics views cover one side of that split and six rank-tracking views the other, the practical argument for keeping both records in one workspace instead of two browser tabs and a spreadsheet.
| What you actually want to know | Which record answers it | What the other one gives you instead |
|---|---|---|
| Did anyone book us off search this month? | Clicks, in the analytics views | A ranking, which books nothing |
| Who else is bidding for this term? | The competitors view, with authority scores | Silence — rivals never appear in your own log |
| Is this term worth a page at all? | Volume-ranked keywords in rank tracking | Impressions, which count exposure, not demand |
| Are we visible in the South Bay specifically? | Neither, until you configure for it | A county-wide average that dissolves the difference |
The headline figures fail individually
Impressions, clicks, click-through rate and average position sit at the top of every screen in the account. Each is measured properly. Each becomes misleading the instant it is quoted without the other three, and the four fail for reasons that have nothing to do with one another.
- Impressions only count appearances. Listing at position thirty-four registers exactly as listing at position two. A climbing line often means you now surface deep in the results for phrasing no buyer in your trade would type.
- Clicks are honest but jumpy at low volume. Someone chose you, which no other number proves. Yet a fabrication shop that went from nine clicks a week to five has recorded a holiday and a rainy Monday, not a decline.
- Click-through rate is meaningless without its position. Four percent at position two means the title and description are failing a fight you already won. Four percent at position sixteen is roughly what sixteen pays out.
- Average position is an impression-weighted mean. Whichever query produced the most appearances writes most of the number — often a broad term you never set out to win, in a part of the county you do not serve.
Together the four describe a funnel: shown, chosen, at what rate, from what slot. Alone, none of them describes anything. The trap in a supplier market is that broad terms generate the impressions, so they dominate three of the four figures while producing almost none of the revenue.
Specification queries and the long tail that pays
The query view lists what people typed; the pages view lists what Google served. Different questions, and the mistake is to skim one and act on it.
Sort the query list by clicks and you get the obvious head terms. Read the bottom of the export instead and you find the demand this economy actually runs on: a stage dimension, a fabric weight, a certification, a lane, a turnaround time, equipment named by model. Single-digit impressions, and the jobs are in there. A coordinator typing a specification has already decided what she needs and is only looking for someone who has it.
By intent, not by size
Sort the export into three buckets: terms naming a capability, a place, or you. Each needs a different page.
- Capability terms deserve their own detailed page
- Place terms belong to service-area pages
- Branded terms tell you whether reputation is doing any work
Query against page
Filter one page and read the queries it collects. A capability page pulling only general terms is not ranking for what it was written to answer.
- Wrong queries on a page means wrong page for the query
- One page absorbing everything is a structure problem
Keyword dynamics turns that from archaeology into routine: it tracks which terms entered or dropped out of the top three, the top ten and the top thirty. For a supplier the top three is the tier that matters, because a buyer working down a call list rarely reaches the second screen. A term sliding from four to eleven costs little; the same term sliding from three to four can halve the calls.
Who is searching, and on what
Two breakdowns get dismissed as housekeeping. Both carry money here.
The country split. A production company in Atlanta, a brand office in London, a buyer in Toronto sourcing capacity for a run: none of them are in your service area, all of them are real customers who will fly nobody and simply place the order. Rows from outside the state written off as bot noise are demand you never quoted. Worth knowing, too, whether those visitors land on a capability page that states specifications or on a location page written for somebody driving over.
The device split. Phones and desktops are not two skins on one audience. Somebody standing on a dock, a stage or a shop floor searches on a phone, needs a number to call, and gives you eight seconds. Somebody comparing three vendors for a purchase order is seated, with tabs open, and will read a specification table. A page that works on one and not the other is a decision about which half of the buyers you keep.
Drive time, not distance
The most expensive misreading in this market is treating Los Angeles as one place. It is not one market, nor a set of neighborhoods in the way a compact city is. It is a set of catchments defined by the freeway system and the time of day.
A rental house in Burbank and one in Torrance sit thirty miles apart and are not competitors: nobody crosses the basin at four in the afternoon to save a hundred dollars on a day rate. A shop in Vernon and a shop in Commerce, six miles apart, are fighting over identical work. Distance predicts almost nothing here; drive time predicts almost everything.
Search Console has no dimension for this, so build it from three imperfect proxies: query rows that name a city or a corridor, page rows grouped under the service-area page carrying them, and tracked terms checked from a set location rather than a national default. None is exact alone. Together they answer the only geographic question worth asking — is this district producing quotes or only impressions?
| What the county-wide average shows | What is often underneath it | What to do about it |
|---|---|---|
| Average position 8.4, flat all quarter | Position 3 in the San Fernando Valley, 19 on the Westside | Split the report by service-area page before reading it |
| Impressions up 40 percent | A broad term now surfacing across the whole basin at page three | Check clicks on the same rows; if flat, ignore the rise |
| Clicks flat year over year | Half the query set has been replaced | Compare the two exports as lists, not as totals |
| CTR falling on a strong page | The page now ranks for queries it was never written for | Write the missing page instead of rewriting the title |
Reading two windows against each other
A single number has no direction. Date comparison is where a report stops being a status screen and becomes evidence, and in a project economy it has rules.
- Leave the last two days alone. Search Console data settles on a delay the presets already account for. Reading yesterday and declaring a collapse is the standard self-inflicted panic.
- Compare like periods, not adjacent ones. Twenty-eight days against the previous twenty-eight reports on your last change. Twenty-eight against the same weeks a year ago reports on the market. Apparel and production run on cycles that make month-on-month reading nearly worthless.
- Compare sets, not sums. Two quarters can produce identical click totals from entirely different query lists. Where every customer is a temporary project, composition changes faster than the total, and only composition carries information.
- Give a change four to eight weeks. That is the honest window before structural work shows measurable movement. Judging a new capability page after nine days produces noise and a reversed decision.
Rank tracking holds the other half of the time picture: a portfolio-wide ranking score with a twenty-eight-day trend curve, the fastest way to separate a site-specific fault from a shift across your whole competitive set. If four domains dipped in the same week, stop auditing the one you were worried about.
The point at which numbers hand over
Everything above is description. None of it decides anything, and three questions here sit permanently outside the export.
Whether a term is worth chasing at all: the record shows what already happens, never what would happen if you published a page for a service you have never described. Generative market research addresses exactly that blind spot — a competitiveness score with tiered rivals, model-classified intent, competitor strengths and content gaps, read off the live landscape instead of your archive. Whether a job is worth having: a hundred clicks from thin two-week work is worse than eight from quarter-long contracts, and no screen knows your margins. And capacity: winning work you cannot staff in August loses the client permanently.
AutoSEO — the reading and the acting in one loop
For a supplier business with a real service list and nobody whose job is search.
- Candidates come from three directions. Search Console history, the live results page and your own seed terms feed one pool, and each candidate is approved, rejected or deferred one at a time — specification terms your estimator recognizes stay in, tourist terms stay out.
- Link building runs on a partner network. Placements draw on over 230,000 sites, which is the part of the work a two-person marketing team cannot do by hand.
- The full analytics stack is included. All eight Search Console views and all six rank-tracking views, not a summary tile.
FullSEO — when the vocabulary has to be exact
For firms where the wrong term on a page costs a quote, and someone must sign off before anything publishes.
- Manual keyword selection with automatic fallback. You pick the terms your trade actually uses; the automation fills the gaps rather than overruling you.
- Placement with an authority target. Links are placed manually against a Domain Authority goal instead of taken as they come.
- Nothing publishes unreviewed. Switch on the human-review mode and a team of specialists, developers and writers checks the automation’s output before it reaches the site.
Stream sits between the two: a single project feed, in order, carrying answers from the assistant, automatic reports, each new backlink with its donor authority and traffic, to-dos in three states and campaign news. Filter it by links, files or to-do, or search the whole history. Lists of terms and addresses go in by the batch — useful when one page per city has to be maintained across eighty-eight of them. Case write-ups live on our blog; formats of engagement are listed under services.
Five questions that come up in the first month
Our best term has forty searches a month. Is that worth any effort?
Almost certainly, if it names something you actually do. Forty specification searches are a different economy from four thousand category searches. Work out what one job is worth and how many of the forty could become a quote; the arithmetic settles it in a line.
Impressions doubled and the phone did not ring. What happened?
You started appearing for something new and broad, deep in the results. Filter to the queries that gained impressions and read them. If they sit past position twenty, nothing happened that concerns your business, and no title rewrite will change it.
Should we build a page for every city in the county?
Only where you genuinely work and can say something specific: a yard you deliver to, a corridor you run, response times that differ. Near-identical pages compete with each other and produce impressions rather than calls. Check the export first — if nobody names the city, the city needs no page.
Rank tracking says position four, Search Console says nine. Which is lying?
Neither. Tracking checks one query under fixed conditions from one location; Search Console averages every appearance across locations, devices and phrasing variants, weighted by impressions. Both are correct answers to different questions, and the disagreement is usually a geography signal worth following.
How often should a small firm look at any of this?
Clicks and the query export weekly, positions and rivals monthly, strategy quarterly. Exports run to ten thousand rows in CSV or JSON; a rendered PDF stops at 250. Treat the PDF as something for other people and do your own reading in the export.
None of this is an advantage in tooling. Two shops on the same street in Santa Fe Springs are handed the same data, free, every morning. The difference is procedural: take the average apart before you believe it, read the query breakdown and the page breakdown against each other, look at the device split before anybody edits a sentence, and attach a district to every row before saying a word about geography.
What the platform removes is labor, not judgment. Workers running in the background keep every dashboard current with no manual refresh, site tags reduce a portfolio of domains to one filter, and one property at a time can be released to an outside address — the consultant sees that client and no other. Reports can be built to your own template, in your colors, which sounds cosmetic and is not: an unopened report changes nothing. Next to it, query research with intent classification covers the question your own history cannot reach.
One quarter of history and one hour without interruptions is enough to learn what the account has been recording all along. Open the dashboard and read the two breakdowns together before anybody quotes an average in a meeting. The line that changes your year will not be near the top of any list. It is a specification nobody on your team thought to publish, eleven impressions a month, three places behind a competitor four exits up the freeway.
