# Diagnosis, $1,200 One category. Your prompts, frozen before we run them. We measure where the engines currently place you, what they cite when they answer, and which of your competitors they name instead. Source: https://martenfield.com/services/diagnosis/ Price: $1,200 (one-time) Length: 2 weeks Method version: v1.1 --- ## Who this is for Proptech and CRE software companies who suspect they are absent from AI answers and want the size of the problem, dated, before committing to anything else. ## What this will not do Tells you where you stand and how much the engines disagree with themselves across runs. Changes nothing on its own. ## What you get - 50 buyer-intent queries across 5 engines at 3 runs each, frozen and dated before the first run - The raw runs: one row per run, per query, per engine, with every cited URL - The method version that produced them - Source map: which domains the models cite for your category - Entity check: how the models describe you, and what they get wrong - Run-to-run disagreement: how much each engine contradicts itself across runs - Technical review: crawler access, schema, rendering, freshness - The sequence of what to fix first, ordered by effort against expected movement ## What you keep **The runs, the method and the order of work.** You get the raw runs, the method, and the sequence of what to fix first. It is the same instrument we use on retainer clients, run once on yours, so what you are reading is a preview of the work rather than a pitch for it. **The baseline.** If you stop here, you keep the baseline. It is yours, and it is dated, which is what makes any later measurement mean anything, whether we take it or somebody else does. ## What the diagnosis distinguishes **Does the model know what you are?** Resolution. Whether the engine places you in the right category at all, and describes you as the thing you sell rather than as something adjacent. This one gates the other three: if a model cannot place you, nothing downstream can work, and no amount of source presence fixes a company the engine thinks is in a different business. **Does it cite your pages as evidence?** Citation. Whether a URL on a domain you own appears in the sources behind the answer. This is produced by publishing something that owns a specific claim, which means it is the outcome most responsive to work on your own property, and also the easier of the two to move. **Does it name you as a vendor to consider?** Recommendation. Whether you appear in the shortlist when a buyer asks who to look at. This is produced by presence in the sources that discuss options in your category, which are pages you do not control, and it is the metric the retainer with earned mentions contracts on. **Do you survive as the buyer gets specific?** Whether a company named in a general answer is still named once integrations, portfolio size, team size or system of record enter the question. A name that appears for the category question and disappears at the qualifying one is a different problem from a name that never appears, and the query set is built to separate them. **Why four and not one number.** These fail for different reasons and are fixed by different work. A company cited constantly and never recommended has a positioning problem: the engine reaches its pages and does not treat it as an option. A company recommended generally and dropped on specifics has a fit-signal problem: the facts a buyer qualifies on are not published anywhere the model can read them. A single score cannot tell those apart, and the remedy for one does nothing for the other. We are stating what the diagnosis separates, not a pattern we have seen: no client diagnosis has been published, and we are not going to describe a typical finding we do not have. ## How it runs ### 01. Query set agreed and frozen We draft 50 buyer-intent queries from your category, your competitors, and the questions your sales team already answers. You approve the set, and it is frozen and dated before anything runs, because the queries are the contract. ### 02. Measurement Every query runs against 5 engines, 3 times each, logged out. Raw responses are stored, not just the parsed result, so a parser correction later re-reads history instead of re-running it. ### 03. Source and entity analysis We extract every cited domain into a ranked source map, then check how the models describe you against how they describe the competitors they name instead. ### 04. Debrief A working call on the findings and the order of work. This is also where we tell you if the honest answer is that you do not need us yet. ## Not included - Implementation. The diagnosis tells you what to fix, and the retainer is where it gets fixed. - Content production. - Ongoing tracking after the 2 weeks. ## What we do not promise - Pipeline numbers from AI answers - Demo counts attributed to a model - Attributed revenue from generated answers Attribution from generated answers is unreliable in 2026. We contract on a leading indicator, not the invoice: citation share on the first retainer tier and recommendation share on the second. Recommendation share is the proportion of answers, across a defined query set, in which a company is named as a vendor to consider. Citation share is the proportion of answers in which a URL on a domain the company owns appears as a cited source. They are reported separately and never merged. ## Who we turn down - Individual agents and local brokerages - No proprietary data to publish - Budget under $2,500 a month after the diagnosis - First question is the monthly rate ## Questions ### What do I actually get at the end? The raw runs, the method version, the frozen query set, a source map of the domains the models cite for your category, an entity check, a technical review covering crawler access and schema, and the sequence of what to fix first. All of it stays yours. ### Can you guarantee we will get cited? No. A retainer contracts on measuring one metric identically before and after, citation share or recommendation share depending on the tier, and on doing the work the diagnosis specifies. Anyone guaranteeing a recommendation or a citation outcome is guessing, because the engines change ranking behaviour without notice. ### Why five engines and three runs? Generated answers vary between runs. One run tells you what a model said once. Three runs across five engines tells you what it tends to say, which is the thing you can act on and the thing that can be re-measured honestly in 90 days. ### What happens if we stop after the diagnosis? You keep the baseline, the raw runs and the frozen query set. Anyone, including your own team, can run the same set again later and compare against a dated starting point. ### Do you work outside proptech and CRE? Not currently. The evidence, the query packs, and the source maps compound inside one category and do not transfer cleanly out of it. We would rather say no than sell you a generic version. --- Each retainer contracts on the metric its work moves, across a named query set over 90 days, measured identically both times, under method v1.1: citation share on the first tier, recommendation share on the second. Both are reported on both.