AI Assistant Acquisition for B2B SaaS. Three layers: understood, recommended, chosen. We measure where you appear across ChatGPT, Claude, and Perplexity, translate the measurement into a fix list, and run the paid pilot when it earns its way in.
Not a monitoring dashboard. A written diagnosis with a fix list. If you already have Profound or Peec AI running, we sit next to them, not in place of them.
Be understood. Entity clarity, technical access, factual accuracy. The engines have to know who you are before they can recommend you. Most of the fixes here are boring and mechanical: schema, canonical pages, crawler eligibility, entity consistency across the web.
Be recommended. Evidence, content, corroboration, citation visibility. Being named is not the same as being the pick. In the audits we publish, the appearance leader and the recommendation leader are usually different companies. That gap is the work.
Be chosen. Paid distribution, landing pages, conversion systems. Earned visibility is a floor. When the fix list says the organic build takes 6 to 12 months, paid ChatGPT Ads become the bridge to the buyer already having the conversation.
The audit measures all three. The read fixes the first two. The pilot delivers the third.
The audit is the diagnosis. The read is the prescription. The pilot is the treatment. Enter at any step. Stop at any step.
A full AI visibility audit for your category, published as a case file in the research library.
A working session and a written document that translates the audit into a three-track fix list with sequencing.
A 30-day paid ChatGPT Ads test. Ad spend billed directly to your account. Written verdict on day 30.
Most agencies port keyword thinking into a semantic system. They bid category terms like “best crm,” “hubspot alternative,” “product analytics.” And they lose to the incumbents who already own that ground.
The audits measure where the AI answers have holes. Symptoms buyers describe when they don’t yet know what to buy. Those responses either name no vendor at all, or name an adjacent category. That’s empty inventory. Semantic, not lexical. That’s what we bid.
The research is not credibility decoration. It is the diagnosis. The read is where we turn it into a fix list. The pilot is where paid becomes the bridge.
Same protocol every time: 30 buyer questions × 3 samples × 3 engines = 270 responses. New case files publish as categories are covered.
HubSpot at 50.7% appearance and 19.3% pick share of total responses. Salesforce appears in 28.5% but is the pick in only 0.4% — 9 of 77 appearances are UNFAVORABLE. Marketo has 15 UNFAVORABLE. Clay is the outlier: 21 appearances, 90.5% recommendation-when-mentioned.
Amplitude leads appearance (71.0%). PostHog leads pick share (28.6% vs 16.7%). Recommendation classification separates mentioned from chosen; the category story flips. PostHog’s recommendation-when-mentioned is 93.9%. Mixpanel is 60% ALTERNATIVE, third by pick share at 6.7%.
HubSpot appears in 78.1% of category-named responses. In problem-first, it drops to 52.4% and Salesforce becomes the top-mentioned vendor.
HubSpot at 57.8% category, 22.2% problem-first. ActiveCampaign and Customer.io close most of the distance when buyers describe symptoms instead of naming the category.
Apollo leads category at 52.2%. In problem-first, video tools like Loom and Vidyard surface more than dedicated sales-engagement vendors when buyers describe outreach quality problems.
71.5% appearance, 30.7% pick share. 41-point gap, the smallest among leaders we’ve measured. AI coding assistants is the most-consolidated category on the recommendation axis. Copilot leads problem-first at 28.6%. Two different buyers, two different winners.
Zendesk is the reference brand for the category. The recommendation is going elsewhere. Intercom and Help Scout are named as alternatives but nobody has broken away as the pick.
Mailchimp appears in 44.4% of AI answers. It is the pick in 1.1%. 43-point gap. Nearly every mention is an example, a comparison, or a historical footnote. The category is wide open for whoever gets picked instead.
68-point gap between appearance and pick share. The biggest in the audited corpus. Snowflake is the incumbent citation, not the recommended answer. BigQuery and Databricks are not filling the gap either. The category recommendation is going nowhere consistent.
Pinecone at 75.2% category-named, drops to 15.9% problem-first. Open-source alternatives Milvus, Weaviate, and Chroma surface more often when buyers describe RAG or embedding-storage symptoms.
Linear at 58.1% category-named. In problem-first, incumbents Jira and ClickUp reclaim the lane. The trendy tool wins the category question and loses the pain question.
Profound leads Anthropic Claude’s AEO-agency mindshare at 50.0% appearance. Otterly.AI at 40.0%. Peec AI at 35.6%. The Hint Co. appears zero times. Day 0 baseline. Right of reply extended to every named vendor. Re-audit in 30 days.
Fidelity leads at 48.4% category, followed by Schwab, Robinhood, and Investopedia. In problem-first, Fidelity takes 36.4% of a 22-response bucket. Small sample, partial three-engine coverage on this run.
One operator, four phases, ad spend billed directly to your account. No retainer, no follow-on lock-in.
Account setup, policy review, crawler eligibility, tracking install (pixel and Conversions API), baselines from your existing channels.
Context-hint research and intent mapping, campaign architecture, creative variants, landing-page recommendations.
Campaign goes live. Daily monitoring, bid adjustments, creative rotation, hint refinement based on early signal.
Cut losing hints, scale winners, final creative pass, performance report, and a written recommendation.
Evidence supports a permanent seat. Cost-per-qualified-conversion is competitive with your other channels, trending flat or improving.
Signal is inconclusive. Delivery, category fit, or a single execution variable clouded the read. One adjustment, one more window.
The channel isn’t for you yet. We say so on the last day. No retainer pitch. Capital returned to your other channels.
Twenty minutes. If your category, budget, or timing doesn’t fit, we’ll say so on the call. No follow-up sequence.