In short: A run turns your project's keywords into a page plan. You see a dollar estimate on your own provider accounts first, and nothing is fetched until a person approves it. Re-runs of stored keywords are free on SerpDad, and results fetched in the last 30 days are reused at no provider cost.
What a run is
A run is one pass over a project: import, estimate, approval, fetch, grouping, checks and review. Every run plans the project's whole keyword list, not only the new keywords. That is what keeps one owner page per keyword. A project has one active run at a time.
Every project starts with a workflow called Keyword pile to page plan. The New run button and the Import page both start it. See Jobs and workflows.
The steps of Keyword pile to page plan
| Step | What happens | Provider cost |
|---|---|---|
| Import keywords | Adds the keywords and skips junk | None |
| Fetch SERPs | Fetches the top 10 Google results for each keyword | DataForSEO, on your key |
| Search volume | Fills monthly search volumes, one task per 1,000 keywords | DataForSEO, on your key |
| Cluster by SERP overlap | Puts keywords on one page when their top 10 results share enough pages | None |
| Score scope and format | Decides whether each page stands alone or becomes a section, and which format fits | Your AI provider |
| Route around owned pages | Marks a page as Optimize when your site already has a page for it | None |
| Park weak pages | Parks single keywords with no volume and thin pages nothing overlaps | None |
| Labs secondary keywords | Adds secondary keywords to each standalone page (20 per page by default) | DataForSEO, on your key |
| Plan check | Confirms every keyword has exactly one place, then swaps the new plan in | None |
| Human review | Lists the calls that need a person. The run does not wait for answers. | None |
| Generate brief | Writes briefs for the top 5 pages by search volume | Your AI provider |
| Notify by email | Emails the person who approved the run | None |
The cost gate
The last import step is the cost gate. It shows, before anything is spent:
- The keywords in this run against the 200,000 limit, for example "2,094 of 200,000 keywords in this run".
- Exact lines for the top 10 fetch and search volume, at your DataForSEO account's own prices.
- Ranges for secondary keywords, page classification and briefs. These depend on how many pages the plan finds.
- Your DataForSEO balance, your monthly spend cap after this run and the owner approval line.
- How many keywords reuse results already fetched in this workspace, and how old the oldest is.
Reused results and skipped junk show "No cost".
The refresh toggle
By default, keywords with a result fetched in this workspace in the last 30 days keep that result. That costs nothing on the fetch line. Turn on refresh to fetch every keyword again. A plan can mix results read on different dates. The gate says so, and every receipt shows each result's fetch date.
When the gate refuses or waits
| Situation | What you see |
|---|---|
| Keys missing or not working | "Connect your keys to see the cost", with a link |
| Over 200,000 keywords | The run is refused, with how many keywords to remove |
| Free first run already used and no plan | "Your second run needs a plan" |
| Estimate above the owner approval line | A Manager sees "Request approval". The run waits for an Owner or Admin. |
| Would pass the monthly spend cap | The run is refused and names the cap |
| Balance below the exact part of the estimate | A warning. The run can still start, and it pauses if the balance drops below $2.00. |
Run your plan
- Open Import in the project and add keywords. See Import keywords.
- Read the estimate.
- Select Approve run. The button shows the estimated total. SerpDad records who approved it, when, and the estimate.
- Follow progress on the run page, or leave. Each step shows its counts, receipt and cost.
- When the run completes, answer any review calls and read the plan.
How keywords become pages
SerpDad decides from counts on the fetched results. It never guesses from what the words mean.
- Keywords are sorted by importance, using search volume and commercial value.
- The most important keyword not yet on a page starts a new page. It becomes the page's focus keyword.
- Every other keyword that shares at least the project's shared-results line with that focus keyword joins the page. The default line is 3 of the top 10 results.
- Keywords that joined a page never pull in more keywords.
- Each keyword gets exactly one owner page. A keyword that ties exactly between two pages stays on the suggested page and goes to Review.
URLs that rank for a very large share of the project's keywords are left out of the count.
Page or section
Each search result is classified as a dedicated page (the whole page is about the phrase) or a section (the phrase is one part of a bigger page). A page stands alone when enough of its top 10 results are dedicated pages. The default is 5 of 10. Otherwise it becomes an H2 section inside a related page. The AI classifies results. It never invents them.
Format
The format is the most common page type among the top 10: list, how-to, comparison or landing. When the results lean clearly one way, SerpDad settles the format itself. A true split goes to Review.
Shape the plan
In the plan, the Preset and granularity control changes how broad pages are on the next re-run.
| Preset | Shared results that put two keywords on one page | Effect |
|---|---|---|
| Fewer | 2 of 10 | Fewer, broader pages |
| Balanced | 3 of 10 | The default |
| More | 4 of 10 | More, narrower pages |
You can also set a custom line from 1 to 10. A free preview shows the page count before you re-run. Format overrides, the Locations hub and hub names are covered in Read the plan.
Re-runs
Re-running a project's stored keywords has no SerpDad charge and adds nothing to keywords stored. You pay your providers only for results older than 30 days, refreshed results and new work.
Human work survives re-runs: review answers, moves, pins, focus choices, briefs, verified claims and signed articles.
Compare a re-run before applying it
Turn on staging for a project to check a re-run before it replaces the live plan. Staging is off by default.
- Run the project again.
- Open the comparison. It shows the plan sum before and after, verdict changes, which pages changed and why, and every piece of human work kept.
- Select Approve and apply Run N to swap it in, or Discard Run N to throw it away.
A Manager, Admin or Owner applies a re-run.
When a run stops
- Key problem: the run pauses for up to 72 hours. Fix the key and select "Check key again". See Provider keys.
- AI provider not answering: SerpDad retries for 24 hours before the run fails.
- Failed run: the old plan stays live, and fetched results are kept for the next run. The failed run shows what it cost and a retry with its estimated extra cost.