Knowledge Base
Quick answers on forecasting concepts and on using PtahCast day to day. For the reasoning behind any of this, the blog has the long version.
Forecasting Concepts
What the terms on your forecast and flow metrics pages actually mean.
What is Monte Carlo forecasting?
A simulation technique that runs the remaining backlog thousands of times, each time resampling from your team's actual historical cycle times and throughput. Instead of one predicted date, you get a full distribution of possible completion dates and how likely each one is. See the full explanation.
What do P50, P85, and P95 mean?
They're percentiles from the simulated distribution. P50 (Aggressive) is beaten by half of all simulated futures — a coin flip, good for internal planning, risky to quote a client. P85 (Commercial) is beaten by 85% of them, the usual default for a client-facing date. P95 (Safe) is beaten by 95%, for engagements where being late is expensive or reputationally costly. See the full framework for picking one.
What is a confidence level, and why does it change over time?
Every forecast is labeled Low, Medium, or High. Low means it's running on a synthetic seed with no completed tickets yet. Medium means it's a blend of synthetic and early real data. High means it's built entirely from this team's own completed work. The label rises automatically as more relevant tickets close — see the full breakdown.
What is WIP (Work in Progress), and why does it matter?
WIP is the number of tickets in progress at once. By Little's Law, cycle time rises directly with WIP for a given throughput — the more that's open at the same time, the longer each individual ticket takes. It's one of the strongest predictors of whether a forecasted date holds. See why WIP matters.
What's the difference between cycle time and throughput?
Cycle time is how long a single ticket takes from start to done. Throughput is how many tickets finish per unit of time. A forecast uses both — throughput to simulate how many tickets close in a given period, and cycle time to understand how flow is behaving. See reading your flow metrics.
What's the difference between cycle time and lead time?
Lead time is measured from when a ticket is requested or enters the backlog to when it's delivered — the total customer wait, queue included. Cycle time only starts once work actually begins, so it excludes backlog dwell time entirely. PtahCast forecasts off cycle time, since a Monte Carlo simulation of the remaining backlog needs to reflect how fast the team processes work once picked up, not how long tickets historically sat waiting. See the full breakdown of where the two get confused.
Why does PtahCast use cycle time instead of story points?
Story points are a pre-work relative judgment — useful for sizing conversations, not built to predict a calendar date. Tickets with the same point value routinely take wildly different actual time. Cycle time is measured after the fact, so it carries the real variance of how work actually moves through a team. See the full comparison.
What's signal and what's noise on a cycle time scatter plot?
Most of the spread on a cycle time scatter is ordinary variance, not a data problem — different tickets take different amounts of time for normal reasons. A genuine outlier is rarer, and has a specific, identifiable cause (scope that crept mid-ticket, an unscoped dependency). See how to tell the difference.
What is the cold-start problem, and how does PtahCast handle it?
A brand-new engagement has zero completed tickets and zero throughput history to forecast from. PtahCast seeds a synthetic Poisson distribution as an honestly-labeled placeholder (Low confidence), which is gradually replaced by real data as tickets close. See how the seeding works.
What is an accountability log?
A permanent, dated record of every forecast made and whether it held up. It's what turns a Monte Carlo forecast from a one-off chart into something a client can check against reality over time. See the full explanation.
How is this different from a Gantt chart?
A Gantt chart treats every task duration as fixed and known, producing a single deterministic end date with no visible uncertainty. A Monte Carlo forecast simulates thousands of possible futures from real historical variance and shows a distribution instead of one number. See what each one assumes.
Why do price and delivery date need separate estimates?
Most quotes set a price and a date off the same rough estimate, so when that estimate is wrong, both promises fail together. Price should still come from a cost estimate; the date should come from real throughput run through a forecast. See the full argument.
What's the gambler's fallacy, and how does it show up in estimation?
A string of missed deadlines doesn't make the next one "due" to land on time — each ticket's odds are governed by the team's underlying rate, not by what just happened. See how to actually correct for it.
Using PtahCast
Practical questions about running engagements day to day.
Do I need a lot of historical data before PtahCast is useful?
No — it works from your first completed tickets. Forecasts get tighter and more confident as your board accumulates history, but you get a usable, honestly-labeled range from day one.
How do I import my Jira or Trello backlog?
From a board's settings, connect it to a Jira Cloud project or a Trello board using an API token (no OAuth app approval to wait on). You'll pick which of your columns each external status maps to — for example, Jira's "In Review" and "QA" can map to two separate PtahCast columns, or you can leave a status unmapped so tickets in it are simply skipped. Once mapped, click "Sync now" to pull tickets in. It's configured per board, so each client engagement connects to its own Jira project or Trello board independently.
Does PtahCast sync back to Jira or Trello?
No — import is one-directional. PtahCast reads tickets from Jira/Trello and creates or updates them locally; it never writes back to the external system. Moving, editing, or commenting on a ticket inside PtahCast has no effect on the Jira issue or Trello card it came from. Sync also only runs when you click "Sync now" — there's no automatic background polling.
What does my kanban board show that a forecast doesn't, and vice versa?
Your board shows state — what's where, and how much is piled up. It doesn't project that state forward into a probability of hitting a date. That's the specific gap a Monte Carlo forecast fills, using the same completed-ticket history your board already has. See the full breakdown.
Can I re-run a forecast?
Yes, any time your backlog or team changes. A new simulation takes seconds and updates your Aggressive, Commercial, and Safe dates immediately.
What happens if I miss a forecasted date?
It shows up in your accountability log. That's the point — PtahCast keeps a running, honest record of forecasts versus what actually happened, rather than hiding the misses.
What do my clients actually see?
On Agency and Scale plans, clients see a white-labeled portal on your own branding, showing the forecast dates, confidence level, and overall progress — not internal ticket names, team member identities, or raw cycle time data. See what's shown and what's left out.
Can my clients see PtahCast branding?
On Starter and Growth, client portals are PtahCast-branded. On Agency and Scale, portals are fully white-labeled with your logo, your color, and your own custom domain.
How do I write a delivery clause around a PtahCast forecast?
A range-based clause has three parts: a commercial (P85) date, an explicit confidence disclaimer, and a reforecast checkpoint. See starting-point language for an SOW.
How does Cost and margin work?
Set a currency, a weekly burn rate, an overhead lump sum, and the sale value you agreed with the client on a board's settings page, and its Commercial (P85) forecast becomes a projected cost, a margin or loss, and a suggested sale price. Since 12 September 2026 the weekly rate is tracked as a history, so a rate change never rewrites the cost already banked at the old rate. See the complete guide.
How does pricing work — per client or per seat?
PtahCast's plans scale with how many client engagements you're running (3, 10, or 25 on Starter, Growth, and Agency), not how many people you invite. Team size can grow freely without changing the bill. See the reasoning.
Do I need a credit card to start the free trial?
No. The 30-day trial starts with just an email — no credit card required. You'll only be asked for payment details if you decide to continue on a paid plan afterward.
Still have a question?
Email us and we'll get back to you directly.