Content research and briefing was eating half the week. A 25-person agency was producing good work — slowly. We gave them a research assistant that does in 20 minutes what used to take four hours, without touching editorial judgment or creative voice.
Same team size. Full editorial control retained. The research just stopped being the bottleneck.
[MARKETING AGENCY, US-BASED] A 25-person digital marketing agency specializing in SEO-driven content for B2B SaaS, fintech, and professional services clients. The agency manages content strategy and production for 18 retainer clients, with a content team of 8 strategists, researchers, and writers. They had a strong editorial reputation — clients renewed — but the team was stretched thin and capacity had become a ceiling on revenue growth.
The founder had tried general-purpose AI writing tools and found them unsatisfying. The output was generic — it didn't reflect the depth of competitive analysis the agency's clients expected, and it didn't translate into the SERP gains that justified the retainer. What they needed wasn't an AI that wrote. They needed one that researched.
When we mapped the content production workflow, the constraint was obvious: research and briefing was the bottleneck, and it was entirely manual. For each piece, a strategist would spend 90 minutes on keyword research — pulling data from multiple tools, cross-referencing search volume, difficulty, and conversion intent. Another 90 minutes reading competitor content to identify gaps, angles, and structural patterns. Then another hour assembling a brief that the writer could actually use.
Four to five hours per piece. The agency's most experienced people were spending a majority of their working week on structured, repeatable research work — the kind that had clear right answers, consistent methodology, and no genuine need for creative judgment.
The writers, meanwhile, were waiting. Briefs came out in batches when strategists finished research runs. Writers had uneven workloads — overloaded for two days, then idle waiting for the next round of briefs. The editorial calendar was driven by research capacity, not publishing opportunity.
The agency had the talent to grow. It didn't have the process infrastructure to scale that talent without hiring more researchers — which would have increased cost without changing the model.
We built a research automation system that operates in three phases, each corresponding to a distinct part of the brief preparation workflow. The system doesn't write content — that's an explicit design decision, and one the agency insisted on. It researches, synthesizes, and structures. The writer still writes.
The strategist inputs a topic and target audience. The system pulls keyword data from their existing SEO tooling via API, identifies clusters by intent (informational, commercial, navigational), scores each cluster by estimated conversion probability based on the client's historical performance data, and returns a ranked list of target terms with supporting rationale. What used to take 90 minutes takes under 10.
For the selected keyword cluster, the system reads the top-ranking competitor pages, identifies structural patterns (what headings appear most frequently, what questions are answered, what angles are commonly taken), surfaces content gaps (questions users ask that no current result answers well), and flags differentiation opportunities. This replaces 90 minutes of manual reading and note-taking with a 12-minute synthesis.
The system assembles a writer brief in the agency's exact template format: target keyword, secondary terms, recommended word count, proposed H2 structure with notes on what each section should cover, suggested examples or data points to include, competitors to reference and differentiate from, and editorial guidance specific to the client's voice. The strategist reviews and edits — they typically modify 15–20% of the draft brief — and publishes it to the writer queue.
The system integrates with their project management tool so that completed briefs drop directly into the writer queue with the relevant client context attached. Writers stopped waiting for emailed briefs. The editorial calendar moved from batch-based to continuous.
From 4–5 hours to under 75 minutes on average. The remaining time is the strategist's editorial review and client-specific customization — judgment work that couldn't and shouldn't be automated.
The same research team now briefs twice as many pieces per month. Writers have a steadier workload and fewer idle gaps. Publishing frequency increased across nine of eighteen client accounts.
Average across the seven clients where the agency has access to revenue attribution data. Higher publishing frequency and better-researched content both contributed. The agency used this figure in four new business pitches.
The agency had budgeted for two additional researcher headcounts to hit their revenue target for the year. They didn't hire them. That's roughly $180K in payroll that didn't happen.
[CLIENT TESTIMONIAL PLACEHOLDER — confidential, will be added upon client approval]
One finding we didn't anticipate: brief quality improved, not just speed. The competitor synthesis surface patterns that human researchers, under time pressure, would often skim over. Writers noted that the briefs were more specific and more actionable than before — even though the strategists were spending less time on them.
We'll look at your current workflow and identify where the research bottleneck lives. One conversation, no commitment — and we'll tell you honestly what's automatable and what's not.
Book a consultation →