Run schedule & cost

Overview

Run schedule & cost (Settings → What we track) is where a project's capturing is decided and accounted for. A schedule asks every active question on every active AI engine on a cadence — daily, weekly, every two weeks, or monthly — so the trend data behind Analytics, Sentiment Analysis and Insights accrues without anyone pressing a button.

The page has one band at the top and six tabs beneath it.

The one-cycle band

The band states, for the project's active schedule (or for one full cycle when nothing is scheduled), in this order:

1. What one cycle is — P questions × M engines × L locations = N answers per cycle. Each factor is printed separately. SERP legs (Google AI Overview / AI Mode) are their own unit and, when on, appear as a separate line: one per question for each search surface turned on that your plan includes, asked in the project's main market rather than once per location. They are never folded into the answer count, and whether a cycle will run (below) counts them, as the capture check does.
2. Whether it will run — for an active schedule, a cycle is judged on whether one cycle fits the daily cap at all; if what is left of the day's allowance when it fires cannot hold it, that fire is refused and retried the next day (a daily schedule waits for its next cycle), and no per-capture ceiling applies to it. For one full cycle priced as a one-off, whether it fits the room your plan's daily cap leaves today and the most answers one capture started from the app may hold. If it does not, the band says so in plain words: "8 answers per cycle · your plan allows 4 a day". Nothing is disabled by this sentence; the capture gate decides when you start a run, and this is the warning before it.
3. What it will draw — for organization owners and admins: what the shape will draw from your organization, never what the platform pays its providers. The band leads with the answers a month at this cadence (for a one-off, the answers in that run) — from 8 October 2026 including the Google AI Overview / AI Mode answers the captures ask, with how many of them those are; where Google AI Overview is captured in India by the project's own daily check, the band counts only the answers this schedule adds to that check and states the check once, on its own line, so two schedules never count it twice — then says how many of them the remaining room in your organization's pooled allowance covers — the allowance is shared by all of your organization's projects and runs in a 30-day period from its start — and what the rest would draw at the pay-as-you-go rate, or that they would not run when pay-as-you-go is off. There is no per-cycle dollar figure. If the draw cannot be worked out, the band says so; it never shows $0 in its place. Members and viewers see the counts and a one-line reason instead of a price. Until 8 October 2026 this line showed an estimate of what the answers are billed to the platform — the platform's own cost, which is no longer shown on customer screens or sent to them by the API.
4. The budget it draws on — the project's budget mode in words (lifetime cap, monthly cap, uncapped), its allocation, what has been used this period, and the next cycle's date and time.
5. Two money facts — "Your plan includes N answers a month at no charge; U used" (a fact about money, not a cap: nothing refuses a capture for exceeding it) and "Used this month: X answers" (observed — no monthly limit exists).

If the band cannot be loaded, it says so and offers a retry. It never shows "no data" for a read that failed.

Tabs

Each tab is linkable (/dashboard/schedules?tab=cycles, ?tab=spend, and so on). The old /dashboard/runs and /dashboard/usage addresses redirect to their tab.

Creating a Schedule

1. Open Run schedule & cost and click New schedule.

2. Enter a descriptive name (e.g., "Weekly Brand Monitoring" or "Daily Competitor Check").

3. Choose a cadence and the day it runs. The price of the cadence you pick is shown beside it, and the next five cycle dates are previewed below, counted exactly as the scheduler counts them:

   • Daily — the highest cadence available; best for time-sensitive campaigns, launches, or crisis monitoring.
   • Weekly — once every 7 days, on the day of the week you pick.
   • Every two weeks — once every 14 days, on the day of the week you pick.
   • Monthly — once a month, on the day of the month you pick (the 29th–31st run on the last day of a shorter month). Pick the 1st to have each cycle cover a calendar month.

   The day defaults to today's. To change it later, edit the schedule in the table; the next cycle is recalculated.

4. Optionally scope the schedule to one prompt group, so a topic runs on its own cadence.

5. Decide whether to run the first capture now. It is off by default: creating a schedule spends nothing until its first cycle unless you turn this on.

First capture behavior: with the switch off, nothing is captured until the first scheduled cycle. With it on, one capture starts as soon as the schedule is created, charged like any other run.

Run once now

Run once now (top of the page) opens the same capture form as New capture — one prompt, a hand-picked set, or every active prompt — priced as a single run, never as a month of them: what that run will draw, i.e. the answers the remaining room in your organization's allowance covers and the rest at the pay-as-you-go rate. The price line re-prices as you change the selection. If the run is larger than your plan's remaining daily room, the form says so beside the button; the button stays enabled and the server decides. Once a capture of several prompts has started — or when you open the form while this project's last manual run is still going — the form shows that run (a smaller run in its own progress card; a larger one, or a run already going when the form opens, as “A capture is running — started HH:MM · N answers · see Cycles”, the time in your own clock) and keeps its start button off for several-prompt captures until the run has ended (a single prompt can still be captured), then says how it ended; if its status cannot be read, several-prompt captures stay off and the form offers “Check again”; a job still waiting for the worker has no run to show yet, but the server still asks you to confirm a repeat.

Running the same thing twice. If this project's last manual run (a bulk capture, a cycle started by hand, or the first capture started with “Run the first capture now” when a schedule is set up) is still running, or started or finished less than an hour ago, starting another asks you to confirm first. The confirmation says when that run started and that it is still running, or how many minutes ago it finished, how many answers it has, and — for organization owners and admins — what it has been billed so far. Confirming starts the run and records on it which run it repeated and who confirmed. Over the API, POST /api/snapshots/capture-bulk and POST /api/runs/trigger answer 409 with code REPEAT_CAPTURE_UNCONFIRMED until the request carries confirmRepeat: true. Scheduled runs are never held back by this, and neither is a single prompt captured on its own (the capture form's single-prompt mode).

What Happens When a Schedule Runs

1. All active prompts are queried — every prompt marked active (or the schedule's group) goes to the configured engines.
2. All active engines are used — each prompt is sent to every engine you have turned on.
3. One answer per combination, per location — 10 active prompts × 3 engines × 1 location is 30 answers a cycle. SERP legs are counted separately.
4. Usage is tracked — token usage and costs are recorded for each capture and roll up on the Spend tab.

Monitoring cycles

The Cycles tab shows each cycle's status in plain words:

• Queued — waiting for the worker
• Running — capturing now
• Complete — every expected answer landed
• Partial — some answers did not land; the reason (credits, a provider outage, a policy pause) prints under the status
• Failed / Cancelled — nothing landed, or the cycle was stopped; the reason prints under the status

Fires that were refused before they became a cycle (a credit or policy gate) are in the Job log on the Schedule tab, marked Blocked with the reason.

Note: Stuck jobs are reclaimed by a workload-aware timeout (up to a ~45-minute ceiling) rather than a fixed cap, so large-but-healthy captures are not killed mid-run while genuinely stuck jobs are still cleared.

Troubleshooting a cycle that stopped short

• Read the reason under the status on the Cycles tab, or in the Job log.
• Check that your engines are still valid and active (Engines & surfaces tab).
• Confirm that you have active prompts in the project (a schedule with no active prompts has nothing to capture).
• Check the Budget tab and the organization credit balance — a used-up budget is the most common reason.

Managing Schedules

• Pause a schedule — stop captures without deleting the schedule; resume it at any time. Resuming recalculates the next cycle. Only active schedules count toward your plan's schedules-per-project limit, so resuming one counts like creating one: it is refused when the project already has as many active schedules as the plan allows.
• More than one schedule on a project — schedules on one project are not combined yet: the charts follow the more frequent one, and on days a schedule limited to a question group runs, they show only that group's questions. The New schedule page and resuming a schedule beside an active one both say so.
• Edit a schedule — change the name or cadence. A cadence change recalculates the next cycle.
• Delete a schedule — permanently remove it. Historical cycles and answers are preserved.

Choosing the Right Cadence

Note: Higher cadences consume more answers and more of your plan's daily cap — the band above the tabs shows the monthly figure for the cadence you run. Health monitoring adapts its thresholds to your shortest active cadence.

Best Practices

• Start with weekly — for most brands, weekly cycles give the best balance of data richness and cost.
• Use daily sparingly — reserve it for short campaigns or crisis monitoring, and switch back afterwards.
• Use groups for different cadences — a daily schedule for a launch group and a weekly one for the rest of the project.
• Read the Cycles tab — a Partial or Failed cycle prints its reason; catch it early.
• Align with your reporting cadence — if you report monthly, make sure the cadence gives you enough cycles to show a trend within that window.