10xSearch results and case studies

What our own baseline says

On August 13 the fixed 50-prompt panel produced 100 raw observations across Gemini and Claude. Thirty-nine responses carried cited URLs, but 10xSearch received zero mentions and zero recommendations. The run completed with zero errors. We publish the miss because it is the honest starting point for measuring whether this work improves real visibility.

How to read these results

A case result is useful when its unit, date, denominator, source, and limitation are visible. AI question wins do not equal leads. Lighthouse performance does not equal revenue. Qualitative public-surface inclusion does not equal a proprietary traffic lift. We publish each result at the level the evidence supports.

Why the scoreboard changed

The earlier scoreboard treated 43 won ledger questions as one current headline and used the phrase 0 to 43. The public evidence now shows 40 current strict questions and 3 historical carry-forward questions. It also declines to claim a comparable zero baseline because the April scan used an older recognition contract.

What customer evidence we still need

Public AI recognition and technical performance are useful, but a complete business case should also include qualified lead volume, conversion rate, closed revenue, cost, attribution rules, and a comparable pre-period. We will add those only when the source data and publication permission support them.

What outside sources currently verify

Inman's editorial site lists 10xSearch as a 2026 Top Luxury Tech/Tool finalist. A National Association of REALTORS article names Rick Janson as the creator of 10X Search, links to the company, and describes its page-evaluation system. NAR NXT and McKissock independently identify him as 10xSearch's founder and a real-estate AI educator. A separate Inman announcement distributed through PRWeb verifies event participation, while KE Team Hawaii discloses a customer-provider relationship. These sources corroborate the company, founder connection, public method, recognition, subject expertise, and disclosed relationship. They are not independent performance audits and do not prove traffic, leads, or revenue.

The evidence ladder behind a case-study claim

Case evidence becomes stronger as the observation moves from a promotional statement to an independently inspectable record. The first level is a named customer statement or public provider disclosure. It verifies the relationship and the customer's reported experience, but not an independently measured outcome. The next level is a public technical or search observation with a named tool, URL, date, and reproducible method. Prompt-level AI evidence adds the exact question, engine, model, answer, cited URLs, verdict, and capture hash. Commercial evidence adds comparable lead, opportunity, and revenue records with an attribution rule. Independent editorial or third-party analysis can corroborate the company, method, or result, depending on what the source actually reviewed. No single level should borrow authority from another. A customer-site credit cannot prove revenue, and an award cannot prove a product outcome. The case page should state the highest supported level, link the source, and identify the evidence still missing.

• Relationship evidence: named customer statement, public disclosure, or live-site credit.
• Technical evidence: public URL, named tool, device or scope, date, and reproducible capture.
• AI evidence: prompt, engine, model, raw answer, citations, verdict, and hash.
• Commercial evidence: comparable lead, opportunity, revenue, cost, and attribution records.
• Independent corroboration: an outside source limited to the facts it actually reviewed.

How a before-and-after result is designed

A useful before-and-after comparison freezes the baseline before the intervention and preserves both states. For a technical case, that means the same public URL, tool, device profile, test settings, and a clear capture window. For search visibility, it means the same query set, market, interface or API scope, and measurement method. For AI visibility, it means the same panel version, verdict logic, engine coverage, and successful-denominator rule. For pipeline, it means comparable dates, qualification definitions, attribution logic, and seasonality notes. The intervention ledger records material site, content, profile, source, and campaign changes between measurements. That ledger improves interpretation but does not automatically prove which change caused the outcome. If a baseline is unavailable or incompatible, the case must say so and publish a narrower current-state observation. A favorable after screenshot without the comparable before state is evidence of current visibility, not evidence of lift.

• Capture and retain the baseline before implementation begins.
• Use the same unit, scope, settings, and denominator in the comparison.
• Record interventions and external changes between the two dates.
• Disclose missing, partial, or incompatible baseline evidence.
• Avoid causal language when the design supports only an observed association.

Why current, historical, and cumulative wins stay separate

AI answers vary over time, so a result ledger needs more than one time dimension. Current strict recognition describes questions that qualify in the newest comparable batch. Historical wins describe questions that qualified in an older batch but not the newest one. A cumulative ledger can show every question that has ever qualified, but it must not use that total as the current recommendation rate. The Doug Leibinger evidence illustrates the distinction: 40 questions qualify in the newest August 10 batch, while three additional questions are older wins. The 43-question ledger remains useful history when those rows retain dates and captures. The current headline is 40, not 43. A compatible baseline also matters. If an older scan used different prompts, engines, or verdict logic, the page should not imply a clean zero-to-current lift. This reconciliation prevents ordinary model volatility from becoming permanent marketing inflation and keeps future regressions visible.

• Current: qualifying questions in the single newest compatible batch.
• Historical: older qualifying questions that do not qualify in the newest batch.
• Cumulative: every distinct question that has ever qualified, always labeled as a ledger total.
• Baseline: an immutable earlier batch comparable under the published contract.
• Regression: a prior current win absent from the newest compatible batch.

How customer permissions shape the public result

A public case study should never outrun the client's permission or the source data. Permission can cover the customer name, logo, relationship, site URL, screenshots, prompt answers, traffic, lead counts, revenue, testimonial language, and publication duration separately. If the client authorizes only a qualitative account, the case can describe public inclusion and the work performed but cannot fill the gaps with estimated numbers. If the client permits aggregate reporting but not identification, the record should stay anonymous and avoid combinations of facts that reveal the subject. A current permission record should be connected to each public claim because authorization can change. 10xSearch uses narrow claims on the Kink Team case because proprietary rankings, traffic, and conversion counts are not authorized for publication. That makes the case less dramatic, but more reliable. Evidence discipline protects the client and keeps marketing copy from becoming a conflicting source that answer engines repeat.

• Record permission separately for identity, logo, URLs, screenshots, metrics, quotes, and duration.
• Publish only the narrowest claim supported by both evidence and current authorization.
• Keep unapproved clients anonymous and remove identifying combinations of facts.
• Recheck permission before republishing a result in a new report or channel.
• Remove or qualify a claim when the source or permission no longer supports it.

How visibility evidence connects to pipeline evidence

Prompt recognition and citations are leading visibility indicators, not customer outcomes. The commercial chain begins when a visitor reaches a tracked page or identifies the source during a form submission. The lead record should retain the landing page, referrer, campaign fields, and approved click identifiers. A scheduling event becomes stronger evidence when it links to the persisted lead rather than existing only as an analytics click. Sales qualification, opportunity creation, and closed revenue each add a durable downstream state. The report must name its attribution rule because first-touch, last-touch, self-reported, and influenced-pipeline views can assign the same client differently. Direct traffic and dark social can also obscure the first exposure. When the chain is incomplete, the case should stop at the last verified stage. A prompt win with no visit remains an AI visibility result. A booked call with an AEO-page first touch is a pipeline result. A closed client with reconciled CRM and billing records is the business outcome.

• Keep prompt observations, web sessions, leads, bookings, opportunities, and revenue as linked but distinct records.
• Persist source context when the visitor submits, not only in a transient browser event.
• Use documented qualification and attribution rules for every pipeline headline.
• Reconcile CRM and billing evidence before labeling revenue closed.
• Stop the claim at the last durable stage when later links are missing.

What a buyer should compare across case studies

Compare the decision relevance and evidence quality before comparing the size of the number. A national brokerage, a luxury agent, and a local team may have different markets, entity structures, publishing permissions, and conversion cycles. Check whether the case matches your problem, whether the baseline is compatible, which work the vendor controlled, and which external factors changed. Follow the public sources and inspect whether the page names the tool, prompt set, engine, model, location, dates, denominator, and limitation. Then compare scope, responsible team, content ownership, technical ownership, measurement access, reporting cadence, pricing, contract term, and exit rights. A small reproducible result may be more useful than a large percentage without a denominator. Ask for references or private evidence where public disclosure is limited, but do not accept confidentiality as a reason to provide no inspectable method at all. The provider's own honest baseline is also revealing because it shows whether the same standards apply to its marketing.

• Match the case's market, entity type, starting condition, and objective to your situation.
• Inspect dates, denominators, methods, sources, limitations, and publication permissions.
• Separate controlled work from recrawling, model changes, seasonality, and other external factors.
• Compare asset ownership, data access, responsible team, cadence, price, term, and exit rights.
• Prefer narrow reproducible evidence over an unsupported universal lift claim.