TrainingPlan.dev: 884,799 Search Impressions in 239 Days

The result

TrainingPlan.dev began recording Google Search impressions on January 9, 2026. By September 4—an inclusive 239-day window—the reconciled Search Console record contained 884,799 impressions and 9,061 clicks.

The latest complete 28-day period in that record produced 294,127 impressions and 3,998 clicks. Against the immediately preceding complete 28 days, impressions rose 94.7% and clicks rose 113.3%.

Measure	Observed value	Evidence boundary
Study window	Jan. 9–Sep. 4, 2026	First nonzero Search Console day through last reconciled complete day
Inclusive days	239	Calendar count
Total impressions	884,799	Reconciled first-party Search Console exports
Total clicks	9,061	Reconciled first-party Search Console exports
Latest 28-day impressions	294,127	Complete adjacent-window comparison
Latest 28-day clicks	3,998	Complete adjacent-window comparison
Impression growth	+94.7%	Versus preceding complete 28 days
Click growth	+113.3%	Versus preceding complete 28 days

These numbers establish that substantial Google discovery occurred. They do not establish revenue, AI citation share, or the causal contribution of one page template, schema property, content change, or technical intervention.

Public evidence package: download the monthly and fixed-window summary CSV or inspect the machine-readable method and claim boundaries. Raw query and page exports remain private.

The backlink statement, precisely

Bob did not conduct backlink outreach, buy links, run a guest-post program, or execute a digital-PR or authority campaign for TrainingPlan.dev during this period.

That is not the same as saying the domain had zero links. Useful public websites may acquire links naturally, automated sites may create references, and backlink indexes are incomplete. The supportable claim is no deliberate link-building campaign.

What TrainingPlan.dev is

TrainingPlan.dev is a broad public system for endurance-training information, plans, calculators, product data, research, and decision support. It was built as a new domain rather than migrated from an established high-authority publication.

At a live inspection on September 10, 2026, its sitemap exposed:

• 589 article URLs.
• 77 tool URLs.
• 678 total URLs.
• 706 image entries.

The footprint matters because the result did not come from a single viral article. It came from a portfolio of editorial and interactive resources connected through a consistent publishing system. Quantity alone is not the explanation: an August audit found 245 of 664 then-sitemapped URLs had no impressions in the current 28-day window, which created an explicit consolidation and improvement backlog.

The operating thesis

The site was built around five mutually reinforcing jobs:

1. Complete a real task. Articles answer specific training questions; tools perform calculations or help readers decide.
2. Expose the answer cleanly. Public pages use static, semantic HTML with crawlable links and clear metadata.
3. Build connected coverage. Topic hubs, author pages, research, related articles, and tools give readers meaningful paths rather than isolated URLs.
4. Make trust inspectable. Visible authorship, methodology, sources, structured data, and update processes support—not replace—the content.
5. Use observed search demand to improve. Search Console data identifies emerging questions, weak pages, and clusters needing better internal connections.

The thesis is consistent with ordinary search guidance. It should not be interpreted as a proprietary ranking formula.

Timeline of the work

January to early August: build the useful surface

The early system established training information, editorial templates, programmatic data, calculators, internal links, metadata, and a growing set of pages. Search Console began showing nonzero impressions on January 9.

By August 5, the internal audit found both promise and debt:

• 1,931 conversational query rows had generated 8,162 impressions and 66 clicks.
• 245 of 664 sitemap URLs had no impressions in the current 28 days.
• 113 posts still carried July dates, highlighting freshness and maintenance inconsistencies.

These were useful diagnostics, not celebratory vanity metrics. They showed where search demand was appearing and where pages had not yet earned discovery.

August: technical, trust, and content-system work

The implementation record included:

• Canonical, robots, H1, sitemap, author, methodology, entity, and structured-data review.
• Internal-link and topic-cohort improvements.
• Image, metadata, and content-template work.
• AI-referral attribution and crawler-access checks.
• Edge caching and production-boundary validation through Cloudflare.
• Original research and more visible editorial methods.

The final release passed 48 of 48 bot-access checks spanning representative Google, Bing, OpenAI, Anthropic, and Perplexity request paths.

August to September: measure fixed cohorts

A defined cohort of 343 changed pages grew week over week from 592 to 770 clicks (+30.1%) and from 41,690 to 55,330 impressions (+32.7%). Within that cohort, 318 shoe and editorial pages grew from 538 to 725 clicks (+34.8%).

The changed shoe cohort’s weekly gain also became less concentrated: Gel Contend pages accounted for 16 of its 187 net additional clicks. That diversification is more encouraging than a curve driven by one page.

The timing does not prove causation. Some pages had only four finalized days after an observed recrawl, multiple interventions overlapped, the domain was already growing, content aged, demand may have changed, and Google’s systems changed independently.

Why the architecture was bot-friendly

The primary public content could be requested without a login and read without relying on client-side application state. Standard anchors, canonical URLs, sitemaps, author pages, and methodology pages made the content graph traversable. The production stack used edge delivery and explicit response validation.

Crawler friendliness did not mean trusting arbitrary traffic. The work distinguished public documents from expensive or private endpoints and tested representative documented bot paths through the production boundary.

llms.txt and related discovery surfaces existed, but this case attributes no causal ranking or citation effect to them. They are inexpensive indexes, not the center of the strategy.

Why useful tools mattered

TrainingPlan.dev did more than publish prose. Its public tools let readers calculate, compare, plan, and make decisions. A tool can earn repeat use and references because it completes a job that a generic article only describes.

The important lesson is not “publish 77 calculators.” Each tool needs a distinct user task, visible explanation, validated inputs, useful output, a stable URL, and maintenance. A broken or interchangeable tool is scaled clutter.

How the evidence was reconciled

The fixed cumulative total combines:

1. An all-time Search Console export through August 5.
2. Non-overlapping daily rows for August 6 through September 4.
3. Adjacent complete 28-day windows for the growth comparison.

The analysis avoids adding overlapping export windows. Query and page exports are retained privately because they expose sensitive performance detail. The public evidence package should eventually add sanitized daily aggregates, calculation formulas, source-file hashes, and screenshots of the fixed Search Console windows so another analyst can verify the arithmetic without receiving sensitive query data.

The current public aggregate package also exposes the monthly time series:

Period	Clicks	Impressions
Jan. 9–31	57	21,966
February	184	32,297
March	226	52,717
April	467	84,979
May	682	101,745
June	1,111	104,566
July	1,543	137,448
August	4,050	295,351
Sep. 1–4, partial	741	53,730

The monthly values sum to the fixed 884,799-impression and 9,061-click total. They are descriptive periods, not isolated experiments.

Jan21,966

Feb32,297

Mar52,717

Apr84,979

May101,745

Jun104,566

Jul137,448

Aug295,351

Sep*53,730

What likely contributed

The evidence supports these as plausible contributors:

• A large and growing surface of genuinely distinct training questions and tools.
• Static, accessible delivery and crawlable site structure.
• Search-demand feedback used to revise page cohorts.
• Clear authorship, methodology, research, and source signals.
• Internal connections between articles, data, and interactive resources.
• A new domain aging and accumulating ordinary search-system observations.

“Likely” is deliberate. This was not a factorial experiment, and the interventions were neither randomized nor isolated.

What the case does not prove

It does not prove that:

• Publishing hundreds of pages causes growth.
• A specific schema type raises rankings.
• llms.txt earns AI citations.
• Cloudflare caching raises search position.
• No links existed or links did not contribute.
• The August implementation alone caused the September curve.
• Search impressions produced a corresponding level of revenue.
• TrainingPlan.dev has leading visibility in AI answer products.
• The result will transfer to another domain, market, or content team.

The result is strong evidence that Bob can build the technical and editorial substrate that SEO and AEO depend on. Direct answer-engine performance requires its own fixed-prompt and citation evidence.

The reusable playbook

1. Create one complete useful loop

Choose one meaningful audience problem. Publish the explanation, decision aid or tool, evidence, author and method, internal paths, and measurement instrumentation together. Prove the publishing system before scaling it.

2. Instrument before celebrating

Record the first nonzero date, page/query baselines, sitemap size, conversion actions, crawler access, and deployment history. Decide comparison windows before reading the result.

3. Scale distinct jobs, not keywords

Every new URL should have a unique audience, question, evidence base, or tool outcome. Consolidate pages that converge on the same answer.

4. Audit production like a crawler

Test status, redirects, body content, canonical, robots, structured data, links, sitemap membership, cache behavior, and WAF response from the live boundary.

5. Treat zero-impression URLs as research questions

A page with no impressions may be new, inaccessible, redundant, low demand, poor quality, weakly linked, or simply outside the measurement window. Diagnose before deleting; consolidate when it does not deserve a distinct URL.

6. Report cohorts and limitations

Fixed cohorts make changes easier to interpret. Adjacent complete windows reduce partial-period errors. They do not remove seasonality, algorithm changes, maturation, or overlapping work.

Next measurements

The case study should be updated against the fixed September 4 baseline at 90 days and one year. The next release should include:

• Sanitized daily search aggregates and source hashes.
• A dated URL and deployment timeline.
• Link-profile observations from named data providers, without claiming completeness.
• Indexed-versus-sitemapped cohorts.
• A fixed-prompt AI mention and citation baseline.
• Qualified conversion measures where privacy and volume allow reporting.
• Negative results and pages removed or consolidated.

That follow-up will tell us whether the curve persisted and whether the site developed measurable answer-engine visibility beyond organic search.

The practical conclusion

TrainingPlan.dev demonstrates that a brand-new domain can earn substantial Google discovery while investing in useful content, free tools, technical access, internal structure, trust surfaces, and evidence-led iteration—without running a deliberate link-building campaign.

It does not provide a shortcut. It provides a well-instrumented example of the long way done systematically.