Cadence and cost

This page explains why Starter runs weekly and Growth runs daily, why we won't sell daily Starter at any price, and why the tier limits look the way they do.

If you've ever wondered why a tool that looks simple has the pricing it does, this is the page.

The cost per prompt

Every time we run one of your prompts, we hit four AI engines in parallel:

ChatGPT (OpenAI API)
Claude (Anthropic API)
Gemini (Google API)
Perplexity (Perplexity API)

Each engine charges per request. The exact pricing varies by provider and model, and providers update their rates. The current effective cost per single-prompt single-engine run is around $0.02-$0.04, varying by model and response length.

Multiply by four engines, add the parsing, classification, and storage overhead, and we land at roughly $0.10 per prompt-run. That number is the foundation of every tier limit.

That cost isn't a guess. It's the average over the last 90 days of production traffic. It bounces up and down by 10-15% as providers tweak pricing, but $0.10 is the durable number we plan against.

What weekly vs daily means in cost terms

A prompt running weekly costs us about $0.10 × 52 weeks = $5.20/year to operate. A prompt running daily costs us about $0.10 × 365 days = $36.50/year.

Per-tier math:

Starter ($19/mo = $228/yr): 3 prompts weekly = ~$15.60/yr in engine cost. After payment processing, support, and infrastructure, this tier is break-even at best.
Growth ($59/mo = $708/yr): 15 prompts daily = ~$547.50/yr in engine cost. Margin exists but it's not huge.
Agency ($179/mo = $2,148/yr): 50 prompts daily = ~$1,825/yr in engine cost. Margin is meaningful here, and that funds the priority support, white-label PDFs, and longer retention.

If we sold Starter with daily refresh, the math would be: 3 prompts daily = ~$109.50/yr in engine cost. Against $228/yr in revenue, that's still positive — but once you add payment processing (Stripe takes about $20/yr per subscription), customer support time (one ticket per year easily costs $50 in labor), and infrastructure overhead, the tier moves from thin-margin to loss-making within a year. We've run the numbers more than once. It doesn't work.

So the founder's actual position on tier design is: Starter is priced to break even and let people testify whether the tool earns its keep. Growth is the workhorse — most customers should be on Growth. Agency funds the long-tail features (white-label, priority support, longer retention) that don't pay for themselves on smaller tiers.

Why we're transparent about this

Most SaaS pricing pages either explain margins in abstract terms or stay quiet about cost structure entirely. We do it differently because the alternative — pretending Starter is a great deal you should stay on forever — leads to one of two bad outcomes:

You churn after a year because the value feels capped, and you blame the product.
You stay too long, get incomplete data, and conclude the whole tool category doesn't work.

Both are worse for you than us being clear: Starter is for validation. Growth is for actual ongoing use. If three prompts a week is the data you need, fine — but most teams need more.

What this means for your data quality

Weekly refresh on Starter has real implications:

You get 1/7 the data points of a daily-refresh customer. Trend lines stabilize slower. Detecting a real movement takes more weeks.
Sudden engine shifts can hide. If ChatGPT updates its default model on a Tuesday and your run is the following Monday, you see the post-change state, but you won't have a clean before-and-after.
PR campaign attribution is harder. If you publish a major piece on day 1 and want to see Perplexity uptake, weekly cadence shows you a coarse signal at day 7. Daily would show you day-by-day movement.

None of this is fatal. Plenty of solo founders run on Starter and get value. But if you're trying to attribute specific actions to specific data movements, daily cadence is meaningfully more useful.

Manual runs and your slot count

On every tier, you can trigger a manual Run now on any prompt. Manual runs don't count against your tier's automated cadence — they're additional.

But manual runs do consume budget. Each one costs us the same $0.10. So there's a fair-use cap:

Starter: 10 manual runs per month per prompt (30 total across your 3 prompts)
Growth: 20 manual runs per month per prompt
Agency: Unlimited manual runs (within reason — if the system detects automated abuse, we throttle)

If you hit the cap on Starter, you'll see the cap in the prompt UI. The cap resets on the 1st of each month UTC.

Hard caps that protect you

Two guardrails keep cost predictable for you and survivable for us:

Tier prompt-run cap: The number of prompt-runs you can spend in a week is fixed by your tier (3 on Starter, 105 on Growth, 350 on Agency — that's prompts × cadence). Once your weekly bucket is used up, scheduled runs pause until the next week. Manual runs count against the same bucket.
Monthly cost circuit-breaker: If your account's engine spend in a calendar month crosses a hard dollar threshold ($10 on Starter, $50 on Growth, $200 on Agency by default), the system stops scheduling new runs and we'll reach out. The thresholds are tunable per-account if you have a legitimate reason to go higher.

These guardrails exist for two reasons: protecting you from surprise charges, and protecting our cost structure from edge cases. Neither has bitten a real customer yet, but the breakers are wired.

When to upgrade

Three signals you should move from Starter to Growth:

You've used all 3 prompt slots for at least a month and you keep wanting to add more.
Your weekly cadence is hiding movement — you can tell because you find yourself clicking Run now more than once a week to fill in gaps.
You're using BD GEO Tracker for client work, even informally. Growth's daily cadence and longer retention pay for themselves in client conversations.

Two signals you should move from Growth to Agency:

You're tracking more than 3 brands — Growth caps at 3.
You want white-label PDFs for client deliverables.

We don't push upgrades. The dashboard shows your current tier and what's available, and we'll email when you're approaching a limit. We won't auto-upgrade you.