Zero to 243 Top-10 Keywords in 6.7 Months

This is a de-identified record of one B2B export site. The original is an independent site in the industrial UAV industry, selling to overseas buyers. No client identity, no dashboard screenshots, no raw exports — only de-identified numbers, the actual categories of work, and the judgments I can stand behind.

Here is the least flattering sentence in it, up front: 0 → 243 is a phase result produced by planning, content, and the outside world together. The only driver I can independently confirm is one news cycle. Separating "what I did" from "what the market, the algorithm, and a news event did" is the part of this case worth writing down.

Background

A new site with a zero baseline, in industrial UAVs for overseas procurement. Our side built the site, so technical and design fixes went straight to an internal team instead of into a client backlog. Numbers were pulled about 6.7 months after launch.

The headline numbers

Metric	Value
Top-10 keywords at start	0
Top-10 keywords at pull	~243
Observation window	~6.7 months
Phase status	On track (P2)

Numbers come from a phase review record, pulled through 2026-05.

What the planning phase actually was

Keywords: subtract first, then pick the core

The client exported a large keyword list from another platform. The problem with that list was not size — it was that many terms had neither real search volume nor any competitive reference. Building on it as-is would have set the wrong direction with great confidence.

What I did was clean first, then narrow:

• Ran the client's list through Google Keyword Planner to attach volume and competition, and dropped the terms with no volume and no discernible intent.
• Picked five core keywords out of what survived, expanded around those, and kept the expansions in separate sheets.

So this was less "keyword grouping" than keyword convergence: strip the noise, then concentrate attention on a handful of terms that actually carry commercial intent, instead of spreading across several hundred terms on day one.

Competitor analysis: the client's list, plus what the SERP said

Competitors came from two places. The client named a few sites they considered strong in the industry; I kept the ones that held up, then added the sites that actually appeared alongside us in the SERP for real keywords, cross-checked with Semrush data.

What I looked at:

• Organic vs. branded traffic share — is SEO carrying them, or is the brand?
• Backlink counts. Being straightforward about it: this one did not go deep.
• Traffic by country, and the trend over the previous two years.
• One core keyword torn down end to end, paired with page-level analysis.
• Behavioral signals: visits, average session duration.

I also logged how their product pages handled specifications and imagery, as reference material for our own content later.

Site diagnosis: small problems, all of them fixed

Because our own team built the site, this round turned up detail-level issues rather than structural ones — TDK, H1s, mobile basics. Each went to the technical or design team and was fixed. Nothing structural was left outstanding.

Product detail pages: no template work on this project

This needs saying plainly, because a case study that claims every module was delivered is not a case study. This project had no dedicated product-page template work. The client's product pages were already reasonably complete: titles, short descriptions, meta, image alt text, plus a structure covering Product Features and Advantages, Product Parameters, Application Areas / Scenarios, and Download. Rebuilding a template on top of that would not have paid for itself, so I only flagged detail-level additions in the competitor analysis and did not count "product page template" as a deliverable here.

What the operating phase looked like in the data

The operating phase ran on GSC and GA4: watch the change, turn it into content and page recommendations. Below are de-identified ranges and trends — no raw exports, no domain, no URLs.

The part worth telling is a contrast. First the impressions timeline (trailing 12 months, daily):

• January: single digits to a few dozen impressions a day (~3–46). Startup noise for a new site.
• Late February: the climb starts — daily impressions go from dozens to low hundreds (~80 → 150).
• End of February into March: daily impressions jump to roughly 1,000, and hold around 900–1,900 through March.

That curve tracks the geopolitical news cycle of February–March 2026 (the US–Iran conflict). A few timely pieces landed inside a burst of concentrated search demand.

Overlay clicks, and the story stops being flattering:

• While impressions exploded, clicks barely moved — daily clicks sat at 0–4 through the spike, and CTR was diluted to near zero.
• The pages producing the impressions were the news pieces: thousands of impressions each, single-digit or zero clicks.
• The clicks came from the homepage (CTR ~15%, average position ~4–5) and branded queries (CTR ~70%+) — the entrances with commercial or brand intent, not the news content.

Geography told the same story twice. Impressions concentrated in news-reading markets like the US and Brazil; clicks came from Turkey, Pakistan, the UAE, Egypt, and India — markets closer to actual procurement. On devices, desktop led both impressions and clicks.

The inflection point

The sharpest growth is late February into March 2026, when daily impressions moved from tens to thousands. The timing matches the news cycle closely. A large share of the Top-10 keyword growth came from that same wave of informational long-tail queries — a lot of "iran … drone … 2026"-shaped questions — rather than from procurement terms.

The one judgment I can confirm

Split the credit and exactly one item is mine to claim with confidence: the content landed inside a news window, which produced a phase-level amplification at the impressions layer.

The boundary matters as much as the claim. That wave converted almost nothing into clicks, and it carried no procurement intent. What did carry clicks was the homepage and branded queries — the entrances that should be solid regardless. A news cycle amplifies being seen, not being bought. Being able to separate those two is worth more than the 0 → 243 headline.

On attribution

0 → 243 reflects planning, operations, and the environment working together: content consistency, site fundamentals, client responsiveness, industry demand, and the algorithm all moved the number, and a news cycle sat on top of them. This case claims no solo credit and validates nothing at the conversion layer — in fact the data says this round of growth stayed at the impressions layer. Clicks and inquiries are a separate fight.

I treat restrained attribution as a professional skill rather than a disclaimer. Knowing which part was mine, which part was the environment's, and which part of a pretty metric is hollow is what keeps the next project from mistaking unrepeatable luck for a repeatable method.

What transfers to other projects

Strip out this specific client and the sequence moves to most new B2B export sites:

1. Don't start a new site by spreading keywords. Start by giving the client's exported list a health check — attach volume and competition, cut the terms with neither volume nor intent.
2. Concentrate on a few core terms and expand from there, kept in separate sheets. More controllable than several hundred terms on day one.
3. Take the client's competitor list, then cross-check it against the SERP and a data tool. Neither source alone is trustworthy.
4. On a site your own team builds, the advantage is fix latency — diagnosis is worth more when the fix ships this week.
5. In review, keep impressions, clicks, and intent separate. Impressions up ≠ clicks up ≠ purchase intent. One timely news piece can carry thousands of impressions and almost no clicks.
6. Mark external factors — news cycles, seasonality, algorithm updates — as their own line. Don't book them under your own work.

The module structure behind this project is written up separately in an 8-module B2B SEO delivery framework.

Boundaries and risks

• No client identity, domain, product URLs, or dashboard screenshots.
• Operating-phase data is given as de-identified ranges and trends (from a trailing-12-month GSC export) — no raw data, domain, or URLs.
• Attribution stays restrained: the one independently confirmable driver is an external news cycle, and its effect stayed at the impressions layer without converting to clicks or commercial intent.
• This case validates nothing about inquiries, conversion, or downstream business results.