How can I grow organic traffic from both traditional and AI search?

The tactics for growing organic traffic are largely settled and widely published. What is not settled for most teams is the sequence — which work to do when, so that the fast returns arrive before the slow ones need defending in a budget review.

This is a twelve-month operating plan built around that constraint.

The honest accounting

Before planning, establish the ceiling. Total impressions multiplied by a realistic best-case click-through rate is the upper bound of what your current query set can produce. If your target exceeds it, the plan needs new coverage or new demand by definition, and knowing that in month one is worth a great deal more than discovering it in month nine.

Then discount for AI Overviews on the affected subset. Pew Research, tracking roughly 68,000 real queries, found users clicked a result in 8% of visits where an AI summary appeared against 15% where one did not. Ahrefs measured a 58% CTR decline for top-ranking pages in February 2026, up from 34.5% in April 2025. Applying historical CTR benchmarks to queries that now carry an overview will overshoot the forecast even if every piece of work lands.

Model the affected subset separately. A blended historical CTR assumption is the most common reason a well-executed plan misses its number.

Quarter 1: find and unblock

The theme is removing constraints rather than adding content.

Weeks 1–2. Baseline both channels. Search Console export, plus twenty buyer questions run against ChatGPT, Gemini and Perplexity twice each. Read the Pages report and fix anything blocking crawling or indexation. Load five key pages with JavaScript disabled and check the CDN for AI crawler blocks.

Weeks 3–6. Rewrite titles and meta descriptions for the ten to twenty highest-impression, lowest-CTR pages. This is the cheapest traffic available anywhere in search — the ranking already exists and the change shows within one to three weeks.

Weeks 7–12. Fix whatever the baseline surfaced. For most sites that is rendering, a large unindexed bucket, or a section that has quietly been excluded.

Why first: everything in later quarters is wasted behind an unfixed constraint, and the title work produces a visible result inside a month, which buys the credibility the slower work needs.

Quarter 2: consolidate and deepen

The theme is doing more with pages you already have.

Consolidate. Run site: searches on your priority terms and merge groups of pages competing for the same query. Redirect the losers into the winner and repoint internal links. Sites regularly gain positions from this, and it costs nothing but editorial nerve.

Deepen the striking-distance set. Export queries ranking 8–20 with meaningful impressions. For each, list the H2s of the top three results and add the subtopics you are missing. Move the answer to the first two sentences of each section. This single edit improves rankings, snippet capture and AI extractability simultaneously.

Delete or merge the dead weight. Pages with negligible impressions over twelve months, no links and no purpose. Thin pages are not neutral — they compete with your good ones and dilute site-level signals.

Why second: this is the highest return per hour available once constraints are cleared, and it operates entirely on demand already proven to exist.

Quarter 3: publish and corroborate

The theme is adding coverage — and only now, because publishing into unfixed fundamentals is how a year gets wasted.

Publish against genuine gaps. Six to ten pages, each targeting a cluster nothing on your site currently addresses. Verify the gap by searching your own domain rather than assuming. Use the prompt gaps from your AI baseline as briefs, since those are questions you have watched an assistant answer without you.

Start corroboration in parallel. Collect the URLs your scans show assistants citing in your category, ranked by frequency. Get accurately described on the ones you are absent from. This is the slowest-compounding work in the plan, which is why it starts in quarter three rather than quarter four.

Why third: new content needs the site to be technically sound and topically consolidated to rank, and corroboration needs enough runway to show results within the year.

Quarter 4: compound and defend

Publish one piece of original data. A survey, a benchmark, an analysis of your own operations. This is the only asset competitors cannot replicate by writing harder, it earns links passively for years, and it is the most reliable route to AI citations because any answer using the claim must attribute it to you.

Defend what you won. Positions gained are positions someone else wants back. Update the pages that earned top positions, watch the queries you own for sustained slips, and refresh anything factual that has aged.

Re-baseline. Run the same twenty prompts and the same Search Console export as quarter one. Twelve months of comparable data is worth considerably more than four quarters of separately-designed reports.

What to stop doing

• Publishing on a fixed cadence regardless of gaps. Volume without a gap is cannibalisation with extra steps.
• Chasing technical scores that do not affect crawling, rendering or indexation.
• Reporting average position. It mixes a page going 90 to 40 with one going 4 to 3. Report position bands.
• Competing on facts that fit in a paragraph. Those pages are being absorbed and the trend does not reverse.
• Running SEO and AI visibility as separate programmes. Two backlogs competing for the same engineering time, making contradictory edits to the same pages.

What to measure each quarter

Five numbers, same five every quarter, reported separately rather than blended: queries in the top ten, non-branded organic clicks, share of voice in AI answers against two named competitors, AI referral sessions with conversion rate, and the count of factual errors assistants state about you.

Keep a dated change log alongside — deploys, migrations, content updates, known algorithm updates. Five lines a month. It turns “why did this move?” from an investigation into a lookup, and it is the difference between a programme that learns and one that repeats itself.

Adjusting for your constraints

The plan assumes a team that can write and can get engineering changes scheduled. Two common constraints change the sequence.

If engineering access is slow. Raise the tier-one fixes as a single bundled ticket in week one rather than filing them individually as you find them — rendering, crawler access and any indexation blocker together, with the business case attached. A rendering change filed as a marketing request competes badly for priority; the same change framed as “our content is invisible to ChatGPT” competes considerably better. Then work quarters two and three in the meantime, since consolidation and editorial restructuring need no engineering at all.

If editorial capacity is the constraint. Do quarter one and the consolidation half of quarter two, then stop adding and start maintaining. A smaller, well-maintained site outperforms a larger neglected one in both channels, and a team that cannot sustain new content should not start producing it. Corroboration work is also a better fit here, because it is outreach rather than writing.

If you are starting from nothing

A site with no organic presence at all should compress quarter one into a fortnight and spend the rest of the year on quarters two and three, because there is little existing performance to optimise and coverage is genuinely the constraint.

Target narrow, specific queries rather than head terms — they are winnable now, they convert better, and the topical authority they build is what eventually makes broader terms reachable. The same applies on the AI side: a new brand will not be named for a broad category question and can absolutely be named for the constrained version of it.

What a good year looks like

For a site with existing organic presence and unfixed fundamentals, substantial growth is genuinely achievable, because quarters one and two operate on demand you are already earning and failing to collect. Most of the year’s gain typically arrives in the first six months from work that adds no new content at all.

For a mature site with clean fundamentals, growth is slower and comes from coverage and authority. Expect a steadier curve and judge it over four quarters rather than two.

On the AI side, a realistic year is: retrievability fixed and verified, twelve months of stored prompt history, mention rate up meaningfully on the constrained questions you can genuinely win, several pages being cited by name, and a corroboration programme with named targets running. What is not realistic in year one is being named on broad category questions dominated by market leaders.

The thing that most reliably separates a good year from a wasted one is not the tactics — it is doing them in an order where each step compounds rather than being spent behind an unfixed constraint.

Sources

• Pew Research Center — approximately 68,000 real search queries; clicks fell from 15% to 8% of visits when an AI summary was present.
• Ahrefs — CTR impact of AI Overviews; 34.5% decline for position one (April 2025), approximately 58% (February 2026).
• Vercel and MERJ — analysis of 500M+ GPTBot fetches finding zero JavaScript execution; same behaviour for ClaudeBot and PerplexityBot.
• Aggarwal, Murahari, Rajpurohit, Kalyan, Narasimhan and Deshpande — “GEO: Generative Engine Optimization”, ACM SIGKDD 2024 (arXiv:2311.09735).

Frequently asked questions