01
Your name, your face
Google Images ranks a photo by the authority of the page hosting it. One canonical headshot, free-licensed and placed on the pages Google already trusts, takes the prime slots a voter sees first.
Be the candidate AI describes correctly.
Voters research you through AI before they research you anywhere else. Today the answer leans on news and your opponent's material. We give the engines your record, framed accurately, and prove the movement.
5AI assistants checked, up to: ChatGPT, Gemini, Perplexity, Claude, Grok
24hto your GEO Score and first three fixes
$5000a month to start.
0changes published without your approval
← Visigent for politics
Be the candidate AI describes correctly.
Voters research you through AI before they research you anywhere else. Today the answer leans on news and your opponent's material. We give the engines your record, framed accurately, and prove the movement.
Three results. One entity.
The same work drives all three: your face in images, your panel beside your name, your positions in the answer.
02
Your knowledge panel
The facts box beside your name. A Wikidata entity plus Person schema and corroborating coverage builds the confidence Google needs to generate one. Then the campaign claims it. Timing is Google's; the result is permanent.
03
Your positions, in the answer
A structured bio, one page per priority issue with a plain stance and FAQ, a record page, and fact-checks that pre-empt the misframes AI tends to repeat. Held in Atlas, so every fact says the same thing everywhere.
The starting point.
What the audit usually finds.
Today · the AI gap
- Answers draw on news coverage and opponent material
- Little first-party content for engines to cite
- Positions underrepresented or misframed
- No baseline, so no proof of movement
Target · owned and measured
- Structured positions the engines cite directly
- Share of voice rising against opponents
- Visibility tracked across eight engines
- Cited-source tracking shows what's working
Defense and offense. Same engine, two jobs.
A challenger plays more offense; an incumbent plays more defense. Both get the same baseline and the same scoreboard.
Defense · hold what's yours
- Correct the misstated position, the wrong vote, the confused identity, before a voter sees it
- Fact-checks that answer the three or four misframes engines repeat about you
- Hold the panel, the prime image slots and page one for your name through Election Day
Offense · take the open ground
- Own the issue questions where today's answer is unowned; that's where a challenger moves first
- Make the head-to-head come from your record: '[you] vs [opponent] on housing' answered by your plan, not their press release
- Be the recommendation on 'best candidate for affordability,' not just a name in the list
The prompts that decide the race.
Hundreds, mapped and tracked across three families. Each cluster gets a content, citation and reinforcement plan.
Recommendation
The race
- “Who is the best candidate for affordability?”
- “Who is running against the incumbent?”
- “Which candidate has business experience?”
Position
You
- “What does [you] believe about taxes?”
- “[Your] housing plan explained”
- “[Your] stance on immigration”
Head-to-head
Versus
- “[You] vs [opponent] on the economy”
- “Most pro-taxpayer candidate for governor”
- “Alternative to [incumbent] on housing”
What we build first.
Campaign sites have always been fundraising tools first and information second: one or two pages and a donate button. Engines can't cite that. We build the information layer, and the fundraising follows discovery.
Build
Structured bio
Record, schema-marked and machine-legible.
Build
Issue pages
Clear stance plus FAQ per priority issue, in the voter's words, not the consultant's.
Build
Track record
What you did before the race: business, service, votes, outcomes, with the documents.
Build
Fact-checks
The three or four misframes engines repeat about you, answered before they're asked.
What we measure.
| Outcome | Measured by |
|---|---|
| AI share of voice and visibility | How often AI names you, and your share against opponents, across ChatGPT, Claude, Gemini, Perplexity, Grok, Copilot, Google AI Mode and AI Overviews. |
| AI sentiment | Favorable, neutral or negative, on name searches and on race questions like "who should I vote for." Trended. |
| Prime image positions | Campaign images in the top-left, top-right and column-top slots. |
| Overall image presence | Share of page-one image results that are yours. |
| Branded search | Share of page one for your name that stays the campaign's. |
| Knowledge panel | Live or not, accuracy, campaign-controlled image. |
All baselined in week one and reported as trends, not single-shot scores.
The honest timeline.
Milestones tied to work shipped and measured trend, not to a binary.
Start
Earlier than feels necessary
For a November race, March is comfortable and summer is workable. Engines build confidence in an entity over months; the last six weeks can only play defense.
45 days
Foundation and early lift
Entity, canonical image and off-site placements landed. Your face appears more, and higher.
3–6 months
Majority image slot share
Campaign photos hold most of the prime positions, tracked as a rising trend.
3–9 months
Your own knowledge panel
Planted in week one; generates as coverage builds Google's confidence. Once it lands it's permanent.
Through Election Day
The answer on the issues
When a voter asks where you stand, the answer comes from your positions. The advantage grows as more voters research this way.
Pricing for political programs
Scoped by race, term and ground to cover. Priced on a call.
A primary, a ballot measure and a twelve-month advocacy program are different jobs. Thirty minutes with us and you leave with a scope, a number and the first three things we'd fix.
Questions.
My name is shared with someone famous. Can you fix that?
Partly, and we'll be precise about which part. A Wikidata item plus Person schema
gives you a separate, confident entity, which splits the two of you in Google's
understanding. That alone doesn't re-rank the image grid; placing your canonical photo on
high-authority pages does. We've run exactly this case.
Can you guarantee a knowledge panel?
No. We build what prompts Google to generate one and report progress monthly.
Most land in three to nine months. A Wikipedia article is the strongest lever but is
community-gated and deletable, so we pursue it as a capstone, never a promise.
Will AI quote our campaign site?
Sometimes, but most citations remain earned and third-party. First-party content
is what makes the answer correct; coverage is what makes it cited. We do both.
Does this replace our digital team?
No. It sits beside paid, email and social. Those raise money and turn out voters;
this decides what a voter is told when they ask a machine about you.
Is this only for statewide races?
No. Our current programs are governor, Senate and congressional, but the biggest
per-dollar effect is down-ballot, where nobody is competing for the answers and a modest
program surfaces a candidate fast.
How does the incumbent factor in?
Incumbents start with the coverage advantage and usually the panel. The open
ground is the issue questions, where most answers today are unowned. That's where a
challenger moves first.
Start with the scan.
Enter the campaign site. In 24 hours: how each engine describes the candidate, which questions you own, and what moves first.