Ahrefs found that the top-ranking page loses roughly 58% of its clicks when an AI Overview appears in the SERP. That single data point, cited by Corey Morris of Search Engine Journal, anchors a six-step content roadmap process his agency runs for clients — a process built on the premise that ranking position alone no longer tells you what a term is worth.

The roadmap targets AI Overviews specifically, which Morris calls the biggest current intersection between SEO and Google, though he notes the same discipline extends to LLM visibility in ChatGPT and Perplexity, with a promised follow-up article on that front.

Start from the business objective

Step 1 is not a tool task. Every keyword, topic, and concept must map to a business outcome — a lead conversion, ecommerce revenue, or another trackable metric. Morris argues visibility and traffic are rarely deep enough goals, except for media companies monetizing ad impressions. A third-party logistics provider, his running example, defines success as content investment that produces website lead form conversions, booked business by the sales team, and realized revenue over the relationship. Skip this step, he warns, and you risk producing plenty of content that never delivers the ROI stakeholders expect.

Build the keyword universe and run a content gap analysis

Step 2 uses Ahrefs' Content Gap report under Competitive Analysis. Enter your domain, add two to four competitors (the tool allows up to 10), and export keywords competitors rank for that you don't, filtered by keyword difficulty, volume, and ranking position ranges. That export becomes the input for clustering later.

Two human judgment calls matter here. First, select search competitors — the sites you actually compete against in the SERP — not the product competitors your sales and product teams talk about internally. Those lists often differ. Second, the raw export is not a to-do list: a large share of it will be branded, off-intent, or irrelevant, and filtering that out is a manual decision.

The 3PL example illustrates the sweet-spot logic. The head term "3pl" sits at KD 47 with 29,000 monthly searches — tempting but broad and difficult. "3pl fulfillment" is KD 11 at 2,600 searches per month, and "3pl companies" is KD 18 at 5,000, both with more defined B2B intent. Morris doesn't argue against big goals, but recommends winning the winnable terms before attacking the top industry term.

Read the SERP, not just the keyword

This is the AI-search step, and it's where the 58% click-loss figure does its work. Volume and KD tell you whether you can rank; they don't tell you what that ranking is worth. In an AI-era SERP, the first organic spot can sit below an AI Overview, ads, and other features, so the same position can deliver a fraction of its historical clicks.

The Ahrefs workflow: in Keywords Explorer, apply the filter SERP features > Include > AI Overview to isolate terms that trigger one. In Site Explorer > Organic keywords, filter SERP features > Current > Include target in > AI Overview to find where your pages are already cited — or not — inside AI Overviews. Then click into the SERP overview for priority terms to see the full feature set and who's being cited, recording it in a SERP features column on the roadmap.

Crucially, an AI Overview is not an automatic skip. Each term gets its own decision: for informational, top-of-funnel terms where an AI Overview dominates, you may write to be cited rather than to earn the click, or focus on the organic result below it. In the logistics roadmap, an AI Overview appeared on nearly every B2B term examined — "3pl companies," "3pl fulfillment," "third-party logistics companies" — and that reality reshaped the entire content plan.

Cluster and group

A keyword list isn't a plan. Morris argues topical authority in AI systems comes from building around clusters, not one-off pages chasing single keywords. Ahrefs can cluster by Parent Topic — keywords sharing the same top-ranking page — or by shared terms, and clusters map to specific destinations: hub pages, service pages, or articles.

Again the tool needs supervision. Ahrefs clusters by ranking overlap, but you must cluster by business meaning. In the 3PL roadmap, terms rolled into a Foundation > 3PL cluster ("3pl companies," "3pl services," "3pl warehouse") and a separate Compliance > Amazon FBA Preparation cluster ("amazon 3pl," "amazon fba prep center"). The tool may see them as adjacent; Morris sees two different buyer intents requiring different pages — the FBA cluster being bottom-of-funnel and closer to purchase.

Prioritize on four factors

Step 5 multiplies difficulty, potential, intent, and AI Overview reality. No tool scores this combination, Morris emphasizes — it's the human layer. Score clusters against the Step 1 objective first, with proximity to a lead or revenue beating raw volume. Tier the roadmap with a Foundation set built first — winnable plus high intent — and flag terms where you're optimizing for citation rather than clicks, so the content brief sets correct expectations.

In the 3PL plan, the team deprioritized the broad, informational, AI-Overview-dominated "3pl" in favor of "3pl fulfillment" and "3pl companies" for Foundation, and pushed the Amazon FBA prep cluster up the list because its intent sits closer to buying despite fewer monthly searches. Volume became the last tiebreaker, not the first filter.

Write for citation, not just position

The final step turns each cluster into a brief with primary keyword, supporting terms, intent stage, SERP features note, target page, and angle. Morris's writing guidance is concrete: lead each section with a direct two- to three-sentence answer, then support it; mirror real search phrasing in headings; and include real expertise, concrete detail, named entities, and data — the things fast, cheap, AI-only content can't fake. Schema and internal links from supporting pieces to the hub complete the technical layer.

Baseline before the next shift

Morris closes with two directives. Capture your current SERP feature and click data now, before AI features shift again, so you can measure progress from a fixed point — the click math already looks different than two years ago and will change again. And don't rebuild your whole content operation at once: test the method on one cluster, compare performance against content built the old way, and let that result decide how broadly to roll out the framework. The metric to watch going forward is whether citations inside AI Overviews start registering in your analytics as a distinct, attributable visibility channel.