Why we don't do discovery calls anymore.
The problem with discovery calls.
The discovery call exists because it used to be the cheapest way to qualify a lead. A 30-minute Zoom beats a salesperson spending half a day researching an account. That math was correct in 2015.
It is not correct now. An agent stack can research a prospect's website, competitors, category, and citation footprint in under an hour. Cheaper than a salesperson's salary for the same window. The research advantage that justified the call has evaporated.
What is left is lead-qualification theater. Both sides pretend to listen. The prospect recaps their site and their goals. The salesperson asks questions they could have answered by reading the homepage. Forty-five minutes later, everyone promises to "take this to the team." The bad fits are more obvious than the good ones. And you had to run the call to find either.
What we do instead.
We replaced the discovery call with an audit-first workflow. The prospect submits a form. An agent stack produces a real deliverable. A human editor reviews. The prospect gets a 20-page PDF in 48 hours.
Only the prospects who read the audit and want more book a call after. That call is not a discovery call. It is a scoping conversation anchored to a concrete document both sides have already read.
The workflow
- Prospect submits. Website URL, top competitor, target service, email. Four fields.
- Agent stack runs. Site crawl (Firecrawl), tech audit (Winston tech-audit Skill), GEO audit (winston-geo-tech-audit Skill), citation analysis (scheduled agent queries against ChatGPT, Perplexity, Claude, AI Overviews), and SERP delta (DataForSEO).
- Human editor reviews. The senior editor reads every audit, sharpens recommendations, flags anything the agents missed, and cuts anything generic.
- Audit delivered. 20-page PDF emailed within 48 hours of form submission.
- Prospect opts in. If they want more, they click a "book a call" link inside the PDF. The call is substantive, not discovery.
What the prospect receives
Tech audit summary (crawl errors, Core Web Vitals, indexation gaps). 3 pages.
GEO audit (chunk scores, schema gaps, entity signal). 6 pages.
Citation analysis (where competitors get cited and you do not). 4 pages.
Competitive SERP delta. 3 pages.
Prioritized roadmap. 2 pages.
Named next step. 1 page, with honest recommendation on whether we are the right partner.
Cover + TOC. 1 page.
What this does to close rate.
We do not publish a close-rate number. The honest answer is that we cannot cite one yet without making up a comparison. We have not run a controlled A/B against our own old sales motion. What we can say is this:
- Fits are cleaner. The prospects who book calls after an audit have already seen the work. The calls open with "this recommendation on page 12. How would you execute it?" not with "tell me about your business."
- Calls are shorter. Thirty-minute calls collapse to fifteen because the context is already shared.
- Prospects who do not book have self-selected out for free. They got a real deliverable. They did not have to burn 30 minutes to learn the partnership was not right. We did not have to either.
The trade-off is audit production cost. Each audit costs us model tokens and an editor's hour. We think that is a better trade than salesperson-hours on discovery calls that do not close.
How to implement this yourself.
You do not need our stack to do this. You need three decisions and one commitment.
Three decisions
- Pick the top three audit dimensions that matter to your sales motion. For us, those are tech, GEO, and citation analysis. For a PPC agency it might be account structure, landing-page quality, and conversion tracking. For a cannabis dispensary branding shop it might be compliance readiness, menu coverage, and local SEO.
- Build or prompt-engineer a Skill that produces each dimension from a URL input. You do not need Claude Code specifically. You need any framework that lets you chain research + analysis + formatted output. Skills happen to be the cleanest packaging we have found.
- Ship a simple intake form. Formspree or a Cal.com-adjacent form works. Four fields: URL, competitor, target service, email. Do not ask for phone number on the first contact. You are not asking for a call.
One commitment
Commit to a 48-hour SLA. The audit is worthless if it arrives seven days later. The prospect has already booked a discovery call with someone else. The 48-hour clock is what makes the difference between "AI-enabled" and "AI-native." It is also the number that makes marketing copy honest.