PKG_BEAT_01 // BE FINDABLE
Findable
Findable is the engagement that decides what an AI assistant has to work with when your buyer asks about your category, and then fixes it. It produces a dated baseline, a repair specification your developer can execute, and pages written to be understood, quoted, and checked.
- TIME
- four to six weeks
- BAND
- $5,000 to $15,000, plus a retainer in $1,500 to $5,000
It can be bought on its own, at any point in the sequence, and it does not require the copy work first.
What changed is that your buyer arrives already briefed
Something read your site, your profiles, your last three announcements, and whatever independent sources bothered to describe you. It summarized all of it into a paragraph, and that paragraph is the first thing your buyer believes about you.
You did not write it. You may not have seen it.
So the question is not how to rank. It is narrower, and it can actually be answered: when that brief gets assembled, what is available to assemble it from?
Four things. Pages a machine can lift cleanly. Evidence that survives being checked. One description of the company, used everywhere. And sources you do not own that say the same thing. That is the whole list, and most companies are missing at least two of the four.
Four causes, and they are not the same problem
Most vendors only fix the first.
The text is not there. Content rendered by script after load, or locked inside a PDF, an image, a carousel, or a video. The crawler receives a shell, so nothing on the page can be quoted, including the page that best explains what you do.
The answer is there and cannot be extracted. A section that never names its own subject, so quoted alone it means nothing. A heading that says "why it matters." A claim split across three paragraphs so no single sentence survives being lifted.
The evidence is thin. The page claims and does not prove, in the same words every competitor uses. A machine assembling an answer has no reason to prefer you, because you gave it nothing to prefer.
The company is described four different ways. One sentence on the site, another on LinkedIn, another in a directory, another in a press release. An engine picks one at random, or discards all four.
Eight parts
The first three diagnose. The middle three are the copy. The last two are the parts no one else does.
- 01
The question universe
Thirty to sixty questions your buyer actually asks, sorted into nine classes. Each existing page is mapped against them. The output is a decision: which page owns which question, which pages are near-duplicates competing with each other, and which questions have no owner at all. This is the piece clients cannot do alone, because it requires knowing both the buyer and the sales conversation.
- 02
The baseline receipt
The same sample run across three engines, twice, about a week apart, documented with method, date, and the exact wording used. The finding is recorded as frequency over time, never as a screenshot, because one observation is not a ranking.
- 03
The repair specification
What has to change under the copy for any of it to work, written for your developer rather than for you. Text present before scripts run. Headings that describe their sections. One canonical URL per question. Internal links with real words in them. Claims and prices out of images and PDFs. Transcripts. An author and a company identity visible on the page.
- 04
The answer architecture
One canonical page per question cluster. A section pattern that survives extraction: a descriptive heading, the answer with its scope, the mechanism, the evidence, the limitation, the next step. Passages that name their own subject. Facts, interpretation, and recommendation labelled separately. A key facts table. A how to cite this line on anything original. A named author with a real page behind them.
- 05
The proof density pass
Primary sources linked instead of the article about them. Original data published where the business already sits on it. Cases with dates. Comparisons precise enough to check. Every adjective that could become a number, turned into one.
- 06
Entity consistency
One approved description of the company, used word for word on the site, in bios, in profiles, in directories, and on partner pages. Drift happens because three people paraphrase from memory. Written once, it stops.
- 07
Corroboration
A named list of the third parties whose description would matter: associations, directories, review platforms, publications, and partner or client pages. Bylines drafted. Expert comments drafted. A short, data-led pitch per question cluster. Then a boundary: we write the pitch, you send it. Credibility is lent, never bought.
- 08
The retainer
Monthly. New canonical pages at a cadence the team can actually hold. The ledger continued. A quarterly re-run of the engine sample. A drift audit on entity language. A refresh audit where the update date only moves when the page really changed.
THE NINE QUESTION CLASSES
Every page on this site is written to one of these on purpose. Most company sites are all definition, which is why the commercial and trust questions get answered by whoever else showed up.
| Class | The question behind it | Example |
|---|---|---|
| Definition | What is this and how does it work | What is a messaging system? |
| Diagnosis | Why is this happening to us | Why doesn't our site appear when someone asks? |
| Comparison | A versus B, which fits | Agency, freelancer, or someone in house? |
| Selection | What should we choose, who should we hire | Who should we hire for this? |
| Implementation | How do we do it | What has to change on the page? |
| Commercial | What does it cost, what is included | What is in the band, and what is not? |
| Trust | Is this safe, credible, defensible | What happens if a number is wrong? |
| Local | Who does this near us | Who does this in Minneapolis? |
| Brand | Who is this company | Who is Kaperider? |
Four measures, all leading indicators
Coverage. The share of the question universe that has a canonical page passing the extraction check. This is the number that moves fastest and most reliably, and it is entirely within our control.
Observability. Dated engine samples, counted as frequency across weeks and engines, reported with the method attached.
Consequence. AI referrers in your analytics, branded search volume, and briefs that arrive already naming the question. That is your data, not ours.
Integrity. Entity drift count, and unverified claims left in the proof library. A number that should fall and stay down.
None of these proves that an engine will ever cite you, because nobody can prove that. A vendor who implies otherwise is selling the thing this engagement exists to replace.
KEY FACTS
| Engagement | Findable |
|---|---|
| Produces | dated baseline, question map, developer repair spec, answer-ready pages, corroboration plan |
| Time | four to six weeks for the build |
| Inputs | a working site, a developer or the authority to hand one a specification, and one person who knows the buyer |
| Band | $5,000 to $15,000, plus a monthly retainer in $1,500 to $5,000 |
| Not included | a new website, media, link buying, any guarantee of visibility |
| Stands alone | yes, with or without the other packages |
| Reply | within two business days |
And who it is not for
It is for a company whose buyers now do their first research in an assistant rather than a search box, and who cannot tell what that assistant says about them. Usually that means a considered purchase, a real sales conversation, and a category where a wrong description costs you a meeting.
It is not for a company with no content to work with. There is nothing to repair yet.
It is not for a company that wants a score, a dashboard, or a monthly report it will not read.
And it is not for anyone who needs the answer to be a guarantee. If a vendor promised you visibility, you have already been sold the thing this engagement refuses.
WHAT IT REFUSES
- No promise of a ranking, a citation, a mention, or a recommendation, in any engine, at any time.
- No thirty-to-sixty word answer rule, because it is not a real rule.
- No FAQ markup sold as a rich results strategy. That ended in May 2026.
- No llms.txt sold as a fix, because Google ignores it.
- No "add schema and you become visible." No special markup is required.
- No near-duplicate pages chasing one question.
- No bought links, bought reviews, seeded mentions, or promotional text placed across low-quality sites. That is not a shortcut, it is a liability.
Start with the free Answer Audit
Twenty questions, three engines, two passes, one dated ledger, and the five fixes we would make first. It takes about three hours and we do six a month. If the audit says you are already in good shape, we will tell you that and you will have lost nothing.
If you already know the answer is no, start a brief and pick the band.
FILED BY / Kaperider, LLC / Daniel Stouffer, founder
Human led. AI assisted. Editorially accountable.
Published: 2026-11-16 · Last reviewed: 2026-11-16