Companies now shop for agencies the way they shop for software: describe the situation to an assistant, get a shortlist. A founder types the industry, the budget, the channel that is not working, and named agencies come back. Here is how the names get chosen.
Agency moments follow marketing pain: the channel that stopped converting, the launch that needs more hands, the rebrand after a pivot, the specialist need, this vertical, this platform, this compliance regime, the agency being replaced. Specific asks dominate: buyers describe their industry and problem, so vertical and channel specificity is what gets matched.
For agency recommendations, assistants read agency directories and review platforms, published case-shaped work, category listicles, community threads where clients speak candidly, and agency sites that state focus, process and pricing posture plainly. Portfolios of logos verify little; a page explaining how you solve a named problem for a named vertical is quotable evidence.
The agencies that keep appearing chose lanes and published proof: named verticals, named channels, honest case write-ups, active review profiles. Full-service generalists lose the specific asks that dominate this category. And agencies, of all buyers, check the vendor test: ask the assistants who the best agencies are in your lane, and the citations show exactly whose evidence is working.
The test costs nothing: ask ChatGPT, Gemini, Claude and Perplexity the questions above, phrased the way a real person would say them, and write down who gets named and what sources appear. Run each question twice on different days, since answers vary and patterns matter more than single runs.
For agency operators, the fixes: declare lanes on pages built to answer them, keep directories and reviews active, and publish case-shaped content clients can be matched against. For marketing agencies that want the systematic version, the free Aethon audit runs your real buyer questions across all four assistants with screenshots, and the agency playbook turns the gaps into a work plan.
By matching the described problem and vertical to evidence: directories, reviews, case-shaped content, community sentiment and clear positioning pages. Specific lanes with proof beat full-service claims in nearly every phrasing.
No. Directory ads buy directory placement, not answer placement. The recommendation text is synthesized from trusted sources on every major assistant.
Because buyers ask with constraints, industry, channel, budget, and the specialist's published evidence matches the constraints. Assistants score match quality, which specialists control and generalists dilute.
Pick the lanes you want, build one honest page per lane, activate reviews and directories, and measure the four assistants monthly on those lanes. The free Aethon audit shows your current shortlist standing, and yes, agencies also run Aethon for their clients.