Flat traffic with a frozen budget feels like a dead end. Jonathan Berthold, VP of Revenue at Ziff Davis, says most marketers are sitting on an easier fix: stop chasing informational clicks and test where buyers actually decide.
Start with the stalled funnel, not more traffic
Berthold’s first move is a full site analysis to understand why growth stopped. Then he targets commercial-intent pages. In his view, top-of-funnel informational traffic is being siphoned by AI Overviews and often doesn’t convert. So the test is either to optimize existing pages for transactional queries or create new content to fill a topic gap competitors haven’t covered.
Without budget, he uses AI to research what ranks for bottom-of-funnel terms and where competitors are weak. He calls bottom-of-funnel targeting “100% the easiest low lift opportunity” many SEOs ignore.
Pitch the learning, not the gamble
Approval is a communication problem. Berthold frames experiments as a hypothesis that will teach the team something, with a long-term payoff. A pitch becomes: if we learn X, Y and Z, we can earn this much later. That’s easier for a P&L-driven leadership team to approve than “let’s test and hope.”
Keep tests too small to tank the site
A sitewide SEO experiment can go wrong and drag down pages that already perform. Berthold’s rule is to start on a small section, iterate, and expand only when results repeat. Paid media taught the same lesson: tripling budget in one day “ruins all the optimization,” he says.
- SEO: judge directionally after about 30 days; 60 days is the max.
- Paid media: review every two weeks, or after spending 2–3x target CPA.
- Scale: expand only after a pattern replicates, not after one lucky result.
The test most teams skip happens after the click
Berthold sees the same blind spot in SEO and paid: marketers optimize for traffic and forget the landing experience. Any page that already earns organic or paid traffic is a candidate for split testing and funnel optimization. He says this should be always on, because small conversion lifts at each stage add up.
Turn AI citations into an outreach list
His own experiment uses AI citation data to find link opportunities. Many teams track which sites get cited in AI answers and note when competitors appear. Fewer actually pitch those publications. Pages that cite competitors for bottom-funnel terms often rank themselves and get cited by LLMs, so earning a mention compounds.
Intent beats a volume tool
Berthold treats search volume as directional, not decisive. In a small B2B market, ten searches a month can still produce a half-million-dollar contract if two of those searchers land and convert. A high-intent query like “I need a plumber to fix my toilet” may show tiny volume but urgent need. As AI search fragments queries, he expects the volume number to become less representative.
The practical sequence: run small tests on high-intent pages, measure conversions, and use that result to sell the next experiment. A conversion-tied test gives leadership a number they can act on.
Source: Search Engine Journal
