Building Your Test-and-Learn Cadence
Every content update is an opportunity to run a structured experiment and generate compounding insights. Josh Grant shares the exact test-and-learn cadence his team used at Webflow to grow AI-attributed sign-ups 5x in under a year, giving you a repeatable system to turn weekly experiments into measurable visibility and revenue gains.
What You'll Learn
Josh Grant built a repeatable system at Webflow that took AI-attributed sign-ups from 2% to nearly 10% in under a year. That traffic converted at 6x higher than standard non-branded visitors. The defining factor was how fast his team learned from each round of experiments.
- A strong test-and-learn practice depends on how quickly your team absorbs lessons and applies them to the next round.
- Structured testing compounds your advantage over time. Each round of experiments feeds better hypotheses into the next.
- Content needs monthly refreshing to stay cited by AI models. Stale pages actively lose ground in AI search visibility.
What separates experiments from routine work
Josh defines a content experiment by two requirements: a clear hypothesis and defined success criteria.
- An experiment looks like: "Optimize 20 FAQ pages and measure citation rate change over two weeks." The hypothesis and success criteria are what make the insight transferable across your content operation.
- Without that framing, an update improves one page but generates no learning you can apply elsewhere.
How to build a portfolio of bets across time horizons
Josh recommends running dozens of tests per month, stacked across three horizons so your team balances near-term wins with longer-term positioning.
- Horizon 1 (this quarter): Low-risk, fast-feedback experiments like page refreshes, FAQ editions, and schema markup improvements. These generate quick data.
- Horizon 2 (next two quarters): Seed bets on emerging opportunities, such as building content around emerging topics or deploying trigger-based agents to respond to market shifts.
- Horizon 3 (high risk, high reward): Investments like first-party research designed to become future LLM training data. These take longer to pay off but can create durable competitive positioning.
You can also tier your review process by risk level. Fast-moving experimental pieces can go through minimal review, while brand-sensitive content gets extra validation before publishing.
Tying every test back to revenue
Every experiment needs a clear line back to business outcomes. At Webflow, Josh's team aligned their testing program to revenue-connected metrics. They tracked AI search visibility alongside sign-up volume and conversion rate.
- Horizon 1 example: FAQ generation at scale via AirOps produced 330+ new citations.
- Horizon 2 example: Automated content refreshes drove a 24% increase in impressions over three months.
- Horizon 3 example: Programmatic long-tail pages targeting "AI website builder" queries resulted in a 51% increase in topical visibility and a 20% lift in sign-ups.
All of these compounded toward revenue goals. A small team drove 5x growth in AI-attributed sign-ups by connecting each experiment to measurable business impact.
Running a weekly operating rhythm with AirOps
Content Engineers can run a tight weekly cadence using AirOps playbooks to detect citation movement and flag optimization opportunities. Your team approves changes at speed while agents handle execution.
- Automated audits and optimization workflows keep your content performing, and built-in internal linking keeps pages connected. All of this runs without scaling your team.
- At Webflow, this approach compressed time-to-impact by 70% and scaled the team's output without new hires.
- Quill matters here: it handles detection and execution while your team focuses on strategy and quality decisions.
Why velocity of learning compounds
Josh closes with a direct takeaway: velocity of learning is the competitive advantage that compounds over time. Your team's speed of learning from structured experimentation is the advantage that compounds over time. Building that cadence is what sustains growth long after any individual experiment finishes.
Key takeaways
- 330 citations from one betJosh's team used AirOps to generate FAQ pages at scale as a Horizon 1 experiment. That single bet produced over 330 new citations across key topics, proving that structured, lower-risk tests can deliver outsized results when tied to clear metrics.
- The 6x conversion edgeAI-attributed visitors at Webflow converted at 6x the rate of standard non-branded traffic. That gap explains why AI search visibility drives pipeline directly for teams that invest in it.
- Tier your content by riskScale your review depth to the stakes of each piece. Lower-risk test pages and FAQ variations move through faster review cycles, freeing your team to spend more time validating brand-sensitive content. This risk-tiering approach lets you maintain high experiment velocity without compromising standards.
- Agents scaled output without new hiresAirOps agents handled automated audits and optimization workflows while keeping internal linking current, driving 40%+ traffic growth from refreshes alone. The team scaled output while maintaining brand standards across every piece.
- The experiment that opened a new marketProgrammatic long-tail pages targeting emerging queries like "AI website builder" were a Horizon 3 bet for Webflow. That investment drove a 51% increase in topical visibility and a 20% lift in sign-ups, showing how high-risk bets tied to emerging market direction compound into durable positioning.
FAQs
A test-and-learn cadence is a structured weekly rhythm where your team runs content experiments with defined hypotheses and measures outcomes against success criteria. Those insights feed directly into the next round of work. It shifts content operations from ad hoc updates to a repeatable system that compounds learning over time. The cadence typically includes detecting performance changes, flagging opportunities, and approving the next batch of experiments on a weekly cycle.