GenGrowthDaily and weekly KPI context
Latest available clicks, impressions, CTR and impression-weighted average position, with the exact reporting windows, data freshness and GSC verification links.
Daily Search Console triage
Run a Daily GSC Briefing Before You Start SEO Work
Read the latest available Search Console data in a short daily brief, with exact dates and filters. Still-updating data is provisional; change-based recommendations require complete comparison periods.
Requires a Google Search Console connection
GenGrowth requests read-only Search Console access. It cannot change your site, your rankings, or your Google account, and this briefing is recomputed on demand rather than stored.
Connect Search Console
No scheduled jobs, no hidden scoring model, and no saved report history. Every run is rebuilt from the property you pick, and unavailable evidence stays explicit.
Connect Search Console to build today's briefing
Choose one verified property and read its latest available Search Console data. Still-updating dates are labelled provisional; incomplete periods cannot trigger change-based actions. Access is read-only, and the report is not stored on our servers.
Connect Search Console
Read-only access · On-demand report · Disconnect anytime
How the daily briefing is built
- 1
Connect Search Console
Read-only, revocable at any time from your Google account settings.
- 2
Read the latest available reporting dates
The briefing uses the latest available date Google returns, in Pacific Time. It shows current observations while dates are provisional and only compares periods once all required dates are complete.
- 3
Attach query evidence only when it is strong enough
The date series is required. Query and query-page reads are optional attachments: if they are truncated, unavailable, or do not agree on aggregation, the KPI view still renders and the affected actions are withheld.
- 4
Show only the next few things worth checking
At most three evidence-backed changes and at most three next actions, ordered deterministically rather than by an opaque score.
What the briefing gives you each run
Cadence that fits the sample
A daily read when the last full week has enough impressions to support it, and a weekly cadence when it does not.
Six query paths, three page paths, one standing check
Click opportunities, click declines at a held position, moves into and slides out of the actionable band, first-observed query-page pairs and the same pairs seen at six or better, three page-dimension paths, and pages drawing impressions with no clicks — each only when its own evidence floor is met.
Private handoff to the next tool
A same-tab handoff into the related GSC or on-page tool without leaking your property, query, or page into the URL.
Explicit manual checks
Manual Actions and Security Issues remain human checks because the API cannot read them. The briefing reminds you; it does not pretend to observe them.
What one morning's briefing looks like
- The window it read
- Each run reads Search Console's latest available dates, including data Google is still updating. Daily dates use the Pacific calendar. The exact current and previous windows are printed with the reading time; incomplete periods remain provisional and cannot trigger a change comparison.
- A change it will name
- A query holds its average position within half a place while its weekly clicks fall by both 15% and 3. Both floors are cleared, so it is reported as a click decline — explicitly not as a ranking loss, because the position evidence says the ranking did not move.
- A move it refuses to promote
- A query whose average position crosses into the 1–10 band while its prior window carries only 50–99 impressions. It is listed as a provisional position move — the two provisional kinds are both position moves, never clicks — counted in its own list, and it cannot become one of the day's actions. A base that small moves this much on its own.
- What it says when it finds nothing
- “No change cleared the evidence floor in this run. That is not the same as proving nothing changed.” Alongside it, the run prints how many rows each path read, how many it evaluated, and how many produced a candidate.
- What you can take away
- At most three query records, at most two page records, and at most one first appearance at an average position of six or better, which is counted outside the three-action budget so a bad week can never crowd out the one lane that reports something going right. Each card carries the evidence line it was derived from and a button into the tool that can take the question further.
The latest available dates, and when a comparison is valid
Search Console reports daily data on a Pacific-time calendar. The briefing reads the latest available date returned by Google, with no fixed date offset. A later run can therefore contain newer data. The report shows the actual dates, reading time and whether the newest dates are still updating; incomplete data is not treated as settled history.
The latest available date anchors the seven-day window and the seven days immediately before it. Change paths run only when all required comparison dates are present and complete. When recent dates are provisional or missing, the report shows current observations without claiming a decline, growth or a newly appearing result. Missing values remain unavailable.
The trend request reaches the current Pacific date, and the displayed daily window ends on the latest available returned date. The 24-hour chart ends on the latest returned hour, looks back 24 consecutive hours, and displays times in your browser's local time zone. Its newest points can be provisional. The chart switches between 24 hours, 7 days, 28 days and 3 months without another read. Buckets Search Console returned nothing for stay gaps; none is substituted with zero.
Daily or weekly is decided by your sample, not by you
Clicks and impressions are counts, and counts on a small base move by roughly their own square root without anything having happened. A briefing that reads day-to-day noise as news is worse than no briefing.
The floor is 1,000 impressions in the complete week
Below it the briefing downgrades itself to a weekly cadence and says so on the page, with the reason attached. This is a property-level test, not a preference: the same site can be daily one month and weekly the next, and the run states which it is on that day's evidence.
A click path also has to have settled rows
Cadence asks the three click-driven paths — the two query lanes and the page click decline — whether they actually settled any rows, not whether the report as a whole looks healthy. A property where only position paths could run is a weekly property, because an impression-weighted average position is not a click signal.
The mode is named, not implied
A run reports itself as change detection, position observation, current-position watchlist, or unavailable. The last three exist so a property that cannot support change detection still gets an honest description of where it stands, instead of an empty page that reads like nothing happened.
Small rows stay observations
A query whose prior window carries 50–99 impressions can appear as a provisional move. It is listed apart from the changes, cannot produce an action, and prints the impression count that kept it out.
Every path a finding can arrive on
There is no hidden score. Nine paths and one standing check are evaluated independently, each against its own stated floor, and every row in the current window is accounted for as not evaluated, evaluated with no signal, or a candidate.
Two consequences of that table are worth stating rather than discovering. The first-appearance paths cover 8 up to but not including 21, and 6-or-better, so a pair first seen at an average position of exactly seven is evaluated and reported as no signal; the gap is in the specification these thresholds came from and closing it is one constant. And two of them need a confirmed brand list: the click-opportunity path and the zero-click page check both rest on your site's own non-brand rate. Without that list there is no trustworthy brand/non-brand split, so both report themselves as not evaluated instead of quietly measuring your own name.
Every path also prints why it could not run. “The evidence this path stands on could not be read this run” is a different statement from “this path found nothing”, and the briefing never renders the first as the second.
The evaluation paths, as implemented · thresholds are the engine's own constants
| Path | What the evidence must carry | What makes it a finding |
|---|---|---|
| Click opportunity | Confirmed brand terms; a non-brand query with 300+ impressions this window, inside a position band still holding 2,000+ impressions once the query itself is removed | Observed clicks come in at half or less of what your own leave-one-out CTR curve predicts, with at least 5 clicks predicted |
| Stable-position click decline | 100+ impressions in each window and 3+ clicks in the prior one | Clicks fall by both 15% and 3, while average position moves no more than 0.5 |
| Move into the 1–10 band | 100+ impressions in each window, with the prior average position outside 10 | The current window sits inside 1–10 and is at least 1.5 places better |
| Slide from the actionable band | The same query-and-page pair at 100+ impressions in each window, with at least one window inside the top 30 | That pair's average position falls by 3 or more and its clicks fall by 30% or more with it |
| First-observed query and page pair | Page attribution readable in both windows; 100+ impressions now at an average position from 8 up to but not including 21 | The pair is absent from the prior window's visible rows |
| First seen at six or better | The same conditions as the path above | The pair is absent from the prior window and now averages 6 or better |
| Page impressions all but gone | 30+ impressions on that page in the prior window, and whether it appears now is decidable | Impressions fell by 80% or more, and the drop is larger than twice the square root of the prior count |
| Page click decline | 100+ impressions for the page in each window and 3+ clicks in the prior one | Clicks fall by both 15% and 3, and by more than twice the square root of the prior clicks |
| Page first appears in this comparison | 100+ impressions now, and whether the page appeared before is decidable | No prior row, or a row shown zero times, with a current average position inside 30 |
| Pages drawing impressions and no clicks | A measured average position of 10 or better, and your site's own non-brand rate predicting at least 3 clicks for this page's impressions | The page drew none. This is a standing state, not a change, and is listed as a check |
A short list, and nothing padded to fill it
Three query records, two page records
Page rows and query rows are two separate populations with separate budgets. Taking page records out of what the query records left over would let the number of query candidates decide whether a page measurement was visible — one population ranked by the size of another.
One first appearance, outside the budget
Every other path reports something that went wrong or is being left on the table. Sharing one budget meant an active property with three declines a day would never once be told that something arrived in a leading position, so that lane carries its own slot.
One row per query, earliest path first
When several paths hit the same query, the highest-ranked one is shown and the rest are counted, not stacked. The ranking is stated once as a product decision rather than left to the order the code happens to run in.
The count of what was not shown
Candidates that formed but did not fit are counted and reported. If that split could not be read, the run says the count is unavailable — never zero.
What it will not say
That an absent query has no impressions
Search Console anonymizes low-volume queries for privacy, withholding them from the query report. Each run reports what share of your property's impressions and clicks never appear in the visible query rows, or states that the share is unavailable. Absence from the sample is reported as not observed — never as zero, never as newly indexed.
That average position is a rank
It is weighted by impressions, so it describes where your exposures happened, not one fixed place in the results. A query whose impressions shift onto a worse-ranking page shows a falling average while nothing about any single result moved — which is exactly why the position-decline path requires the clicks to fall with it.
That query counts and page counts can be added
They are different populations read from different Search Console dimensions, and Google drops rows differently in each. Summing a query-and-page read to reconstruct page totals loses precisely the rows the page dimension exists to recover.
That it checked your manual actions
Manual Actions and Security Issues have no public read API. The briefing links both reports and lets you record on the page that you looked; it never presents that mark as an observation of its own.
That it kept the report
There is no scheduled job, no saved history and no stored copy. The report and both manual-check marks live in this page's state — refreshing, rerunning, changing property or closing the tab clears them. Search Console keeps the history even though this tool does not.
That the run had unlimited budget
One request holds a 45-second whole-request budget across a required date read and its optional attachments, and the Search Console read allowance is shared hourly across every Search Console tool here. When an attachment cannot be read in time the KPI view still renders and the paths that depended on it report as unavailable.
Where each finding hands you off
A briefing that ends in a paragraph of advice is a briefing you have to re-derive tomorrow. Each card opens the tool that can actually take the question further, and states plainly what that tool will and will not use.
Click gaps go to the Opportunity Finder
That tool ranks click opportunities across the whole property from your own CTR curve. The card says outright that it will not use this particular query or page — the gap in front of you is already measured, and the next tool answers a wider question.
Declines go to Traffic Drop Diagnosis
Query click and position declines open the property-level diagnosis, and so do both declining page lanes — a page whose impressions collapsed and a page whose clicks fell. The card states that the next tool diagnoses the property rather than the row, so nobody expects a per-query answer it was never built to give.
First appearances and zero-click pages go to the On-Page Checker
Of the three page lanes only the first-appearance one lands here, alongside the query-level first appearances and the pages drawing no clicks. The page URL, and the target query where one exists, arrive prefilled so you can see what the public HTML actually supports. The briefing has judged the search result, not the page.
Your property never travels in a URL
Handoffs are carried privately inside the tab. If that private handoff cannot be stored, navigation stops and the briefing says so, rather than falling back to putting your property, query and page into a query string.
Tools the briefing hands off to
- GSC Opportunity Finder
Ranks click opportunities across the whole property against your own CTR curve.
- Traffic Drop Diagnosis
For when the decline is property-wide rather than one row.
- On-Page SEO Checker
Checks what one page's public HTML supports, with or without a target query.
- Competitor Keyword Gap
Where the briefing runs out of first-party evidence: terms competitors rank for and you were not observed on.
Further reading
- Striking-distance keywords
What to do with the rankings you already have, which is what most of these lanes surface.
- When several pages rank for the same keyword
Why a query's average position can fall while no single result moved.
- What a public SEO audit can and cannot see
Where first-party Search Console data stops being optional.
FAQ
Does this replace checking Search Console directly?
No. It shortens the first pass. The briefing is meant to answer what changed and where to look next; Search Console remains the place you confirm the detail and continue the investigation.
Why can a report show KPIs but no actions?
Because the KPI view comes from the required date series, while actions need stronger query and query-page evidence. If that evidence is missing, truncated, or contradictory, the briefing shows the KPI context and withholds the action rather than guessing.
Why does it sometimes switch from daily to weekly?
Small samples create ordinary movement that looks dramatic when read day by day. Below the weekly impression floor, the briefing downgrades itself to a weekly cadence so the output does not overstate noise as a daily signal.
Do you store the report or my Search Console data?
No. The Google authorization itself can persist in your browser so you do not reconnect every visit, but the report is recomputed on request and not saved as a server-side history.