Why this matters
The advanced SEO level forecast clusters with the traffic estimator. A templated page set asks two harder questions first: does each page deserve to exist, and does the forecast rest on the site's own evidence? The guide to programmatic SEO without thin pages sets the test each page must pass, the states a row can end in and the staged rollout. The SEO forecasting guide builds a forecast from a BaselineAn estimate of what a metric would have done without a given change, campaign or promotion, against which the effect is measured., net opportunity, the site's own click-through and ramp-up, and ranges. Breed pages are a classic Long-Tail KeywordA specific, usually longer query searched rarely on its own but often as a group. play, which tempts teams to build one per spreadsheet row.
The dog food maker and its breed pages
Thurnvoss Petfood is an invented maker of dry and fresh dog food sold online by subscription. Its sign-up flow runs a feeding calculator, so its records hold anonymised portion and weight data by breed and life stage. The agency proposes a page for every breed and life stage on a hosted page builder at a subdomain, each with an introduction generated by a language model, the same feeding advice and a sign-up button, all launched together. The board wants a forecast first.
Thurnvoss Petfood: the proposal, its demand and its data
The proposal: 340 breeds × puppy, adult and senior = 1,020 pages, launched together in January.
| Combination group | Combinations | Keyword tool searches a month | Search Console impressions a month today | Subscriber dogs with portion data per combination, last 12 months |
|---|---|---|---|---|
| Popular breeds, puppy and adult | 50 | 50,000 | 8,600 | 120 or more |
| Popular breeds, senior | 25 | 2,000 | 250 | 30 to 90 |
| Other common breeds, puppy and adult | 120 | 15,000 | 1,700 | 30 to 119 |
| Other common breeds, senior | 60 | 600 | 30 | Under 30 |
| Rare breeds, all life stages | 765 | 2,900 | 120 | Under 30 |
- A sample of results pages: breed puppy and adult queries show breed feeding guides with portion tables and pet food brands' breed pages; breed senior queries mostly show general senior feeding articles; rare breed queries show breed clubs and size-group advice.
- Data per combination: typical weight range, daily portion by weight band and activity for each recipe, puppy weight by month of age, the share of owners recording a sensitive stomach and the recipes most chosen. The data owner sets a minimum of 30 dogs per combination before any figure is published.
- Existing feeding section: 14 guides, including general puppy and senior feeding guides. Non-brand clicks last 12 months: 96,000. Year on year on guides not refreshed: −4%. Seasonal index from 1.3 in January, when new puppies arrive, to 0.8 in August. The puppy guide already earns about 600 clicks a month from popular-breed puppy queries.
- Thurnvoss's own non-brand click-through, last six months: positions 1 to 3, 14%; 4 to 6, 6%; 7 to 10, 2.5%. Where Thurnvoss ranks in the top three, the keyword tool's figures average 1.25 times the impressions Search Console records.
- Ramp-up record for guides published in the last two years: almost nothing for two months, about half their settled clicks in months three to five, settled from month six.
- Organic clicks to sessions: 0.9. Trial starts per feeding guide session: 1.2% on average, 0.8% to 1.5% across the last four quarters. Trials becoming paid subscriptions: 45%.
- Platform: each new shop template costs a developer sprint. The developer says a reverse proxy can serve the page
builder under a
/breeds/path on the main domain with about a sprint of work and testing.
Demand, unique data and honest ranges
Demand comes from searches, not the dataset: Search VolumeThe estimated number of times a query is searched in a period, usually per month. estimates are calibrated against the site's own impressions, and the results page shows whether a template fits the need. Substance comes from data that differs between pages, with a published minimum deciding which rows become pages. The forecast varies only the uncertain assumptions and counts only clicks the site does not already earn.
Task
- Test the pattern for demand: which combination groups are distinct needs a template can satisfy?
- Apply the data minimum and give every row a state: indexable page, page kept out of the index, or no page with a broader page serving it.
- Design the template and the test every page must pass, and decide the fate of the generated introductions and of rows that thin out after launch.
- Forecast the programme from the baseline with net opportunity, Thurnvoss's own click-through and ramp-up, low, central and high cases and outcomes, for a settled month and for year one.
- Decide between a subfolder and a subdomain, with the reasons written down.
- Stage the rollout: tranches, sitemap and links, what is read before each tranche, and forecast against actual.
Worked answer
Demand. Breed puppy and adult combinations are distinct needs: people search for them, Thurnvoss already earns impressions for them, and the results reward breed pages with portion tables. Breed senior queries are answered by general senior articles, so a senior page per breed fights the results page. Rare breed demand is spread thinly and served by size-group advice.
Data and states. Popular and other common breeds pass the minimum for puppy and adult, so those rows become indexable pages. Popular seniors pass it too, but with no distinct need their figures become a section on the breed's adult page, linking to the senior guide. Rare breeds and the remaining seniors get no page: they are pooled into size-group feeding pages for small, medium, large and giant breeds, where pooled data clears the minimum.
Template. The page leads with the breed's portion table by weight band and activity, then weight by month of age or the adult weight range, the sensitive-stomach share, the recipes chosen, the number of dogs behind the figures, the date refreshed and the size-group comparison. The method sits once on a methodology page, and the next step is the feeding calculator pre-filled with the breed. The generated introductions go: they restate the row, and the guide notes that Google's spam policies name generated pages without added value as scaled content abuse. Near-identical pages for similar queries funnelling to one sign-up button are the Doorway PageA page or site created to rank for specific, similar queries that leads visitors to the same destination or to a less useful intermediate page. pattern, so render three random pages with the breed covered before build; a reader must be able to tell them apart. A row that falls below the minimum later folds into its size-group page with a redirect.
Forecast. Divide the tool figures by the calibration factor, replace the illustrative curve in the SEO traffic estimator with Thurnvoss's own click-through, and subtract the clicks the puppy guide already earns. Positions move a band either way across the cases, and the low case takes the bottom of the conversion range. Year one is booked by tranche on the ramp-up record, and the monthly sheet applies the seasonal index.
Thurnvoss Petfood: the forecast worked through
Calibrated demand for indexable rows: popular 50,000 ÷ 1.25 = 40,000; other common 15,000 ÷ 1.25 = 12,000.
| Case | Position band | Popular | Other common | Gross clicks a month | Less clicks moved from the puppy guide | Net new clicks |
|---|---|---|---|---|---|---|
| Low | 7 to 10, 2.5% | 1,000 | 300 | 1,300 | 600 | 700 |
| Central | 4 to 6, 6% | 2,400 | 720 | 3,120 | 600 | 2,520 |
| High | 1 to 3, 14% | 5,600 | 1,680 | 7,280 | 600 | 6,680 |
- Settled month, central: 2,520 × 0.9 = 2,268 sessions × 1.2% ≈ 27 trials × 45% ≈ 12 subscriptions. Low: 700 × 0.9 × 0.8% ≈ 5 trials, about 2 subscriptions. High: 6,680 × 0.9 × 1.2% ≈ 72 trials, about 32 subscriptions.
- Baseline, central: 96,000 ÷ 12 × 0.96 ≈ 7,680 clicks in an average month from existing guides, so about 10,200 in a settled month with the programme.
- Year one, central, tranches in January (20 popular pages), April (30 popular) and June (120 other common). Full-month equivalents within the year: 8.5, 5.5 and 3.5. Clicks per page a month: popular 48, other common 6. Gross: 20 × 48 × 8.5 + 30 × 48 × 5.5 + 120 × 6 × 3.5 = 8,160 + 7,920 + 2,520 = 18,600. Moved from the puppy guide: 600 × (20 × 8.5 + 30 × 5.5) ÷ 50 = 4,020. Net: about 14,580, against 30,240 if twelve settled months were booked.
Where it lives. A subfolder, served through the proxy. The pages support sign-ups and the feeding guides, so they need the main site's links, navigation and reputation, and one Search Console and analytics property. The platform was the only argument for a subdomain, and the proxy removes it without a later migration. The subfolder and subdomain comparison asks for canonicals, caching and redirects to be tested through the proxy before launch.
Rollout. The pilot is the strongest popular-breed pages, read by hand. Before each tranche, Search Console must show them crawled, indexed and earning impressions for breed queries, and analytics must show visitors using the portion table and calculator; pages discovered but not indexed, or indexed and ignored, stop the next tranche until the template or data is fixed. Each row's state lives in the dataset, the sitemap lists only indexable pages, and an A to Z hub and the size-group pages link to every breed page, each linking to its sibling life stage and similar-sized breeds. Forecast and actual are compared monthly by line and the forecast revised quarterly.
Common mistakes
- Sizing the programme by the breed list. The senior and rare rows are combinations, not needs.
- Treating the minimum as optional. Below it, pages are names swapped into the same text.
- Forecasting from raw tool figures and a published curve. Calibrate demand and use your own click-through.
- Booking settled clicks from launch. On the ramp-up record, year one roughly halves.
- Choosing the subdomain because the builder defaults to it. Ask whether the proxy is feasible first.
- Launching every page at once. Nothing can be learned, and crawlers face a backlog.
The capstone applies the same tests to a section of fitting guides.
