The Part of Your Profile You Did Not Write
A business sets up a profile once and then treats it as done. In the background, three things keep moving. Google corrects fields on its own. Customers ask questions that stay public until someone answers. And a verification prompt appears, occasionally, and locks editing until it is resolved.
None of that shows up in a monthly report, because none of it is a metric. It shows up as a customer who found a wrong answer, or a week where nobody could change the profile and nobody knew why. The work of running the profile is smaller than the work of local SEO and it is different: it is attributes, media, the question queue, the Google-made diff, and the verification gate.
The narrow part is the write. Every attribute has to be matched against a documented list before it is submitted, because an unknown attribute is rejected silently. Every field update needs the right mask, because some fields cannot be patched alone. Get one of those wrong and the profile simply does not change, and the team assumes it did.
The reframe
The automatable part of a profile is not the copy. It is the check before the write and the read of what Google changed without you.
How Many Checks a Profile Actually Needs
Move the sliders to match your group. The model counts the checks that a person is doing by hand today, which is the operational load a profile quietly carries.
Checks per month across the group
69
3 locations, 42 attribute checks, and 27 open questions
Hours per year
152
spent checking the profile by hand
Annual cost
$5,776
at your loaded hourly rate
Profile health
Stale media and open questions pull the score down. Neither is a ranking factor you can buy, and both are visible to the customer reading the profile right now.
The surface, by size
The documented writable attribute list and the value type each one accepts
The largest surface and the most constrained: every attribute must be checked against the documented list before a write, because an invented one is a silent rejection.
The media resource, its format rules, and the recency of the last upload
Coverage and recency, not volume. A profile with forty photos from one job is not better than one with ten from ten jobs.
The questions resource and any answer already attached to it
The surface where a silence does the most damage, because an unanswered question is public and a wrong answer is permanent.
The getGoogleUpdated read, which returns what Google corrected independently
Changes the business never made. The only way to catch them is to ask for the diff against what was submitted.
The verification state and whether the profile is eligible to be edited at all
Not a task but a gate. A verification request is a stop signal, and every write behind it waits.
The shares sum to one. Note the bottom two bars: the Google-made diff is small and the verification gate is smaller, and together they are the two things a monthly report never shows.
Illustrative estimate from the inputs above. The dollar, hour and profile health figures will differ for your business; this is not a measured result or client data.
Running the Profile, Stage by Stage
Each stage names the system it touches, the payload it leaves behind, and the concrete failure it prevents.
Target System
The business information locations resource
The Profile Is Read as Google Sees It
The agent starts from the read that matters most and is used least: the Google-updated read. It returns what Google corrected on its own, which is the part of the profile a business never chose. Anything on that list is the first thing to reconcile, because it is the change nobody authorized.
The team edits the profile against its own intention and never sees the field Google already changed, so the two versions drift apart.
Stage payload
Reading Before Writing, Because the Rules Say So
The profile API is not hard to write to. It is easy to write to badly. Two rules decide most of the design. A partial update against just the phone numbers is prohibited, so a patch that does not read the other phone fields first can drop them. And the primary and additional categories cannot be updated individually through the mask at all, so a category change is a full read and replace, not a one-field patch.
Both rules push the same way: read first, then write. The agent reads the whole resource, builds the mask from what it read, and only then sends the update. That is slower than a blind patch and it is the only version that does not lose data.
The second rule is not technical. A name, address, or category change can trigger a verification, and a verification request means editing stops until a person resolves it. The agent treats that as a stop signal and stages the write. It never retries through it, because retrying a verification request is how a business gets a profile flagged instead of updated.
What the Agent Does, What Needs You, and What It Never Touches
The profile faces customers and holds the business's public claims, so the boundary is the design.
Autonomous
- Read the profile as Google sees it and list the changes it made
- Check every attribute against the documented list and value type
- Read the Q&A queue and mark which questions carry no answer
- Build the write plan with the update mask for each field
- Track photo recency and the spread of jobs covered
Supervised
- Draft a Q&A answer for a person to publish
- Stage any write that risks triggering a fresh verification
- Propose a corrected attribute from the documented list
- Recommend a photo upload from real completed work
Never
- Invent an attribute that is not on the documented list
- Retry a write blocked behind a verification request
- Update phone numbers or categories through a partial mask
- Answer a Q&A in a voice that argues or names a price
- Generate a photo or publish work that did not happen
Three Things We Will Tell You Not to Buy
Not a ranking lever
There is no way to request or pay for a better local ranking. A well-run profile removes the ways the business was losing on its own faults. It does not move distance, and distance is one of the three factors. If a vendor ties this work to a ranking promise, that promise is not theirs to make.
Not a photo generator
An image of a job that did not happen is a claim the business cannot stand behind. The agent tracks recency and the spread of real work covered, and it proposes an upload when coverage goes stale. Everything it proposes comes from work the business actually did.
Not an attribute guesser
Every attribute is matched against the documented list before it is written, because an unknown attribute is rejected silently and the business never learns the write failed. A tool that invents plausible-sounding attributes looks busy and changes nothing.
Questions About Profile Automation
Where This Fits in Marketing and Reputation
This page runs the profile. These are the surfaces next to it.
Find What Google Changed on Your Profile
Bring your profile. In thirty minutes we will read what Google corrected on its own, list the attributes that are not actually set, and show you the questions customers asked that nobody answered. You keep the list whether or not you hire us.
