How your work flows: process mining on your jobs

What "How your work flows" shows

Most operators can tell you a job took too long. Almost none can tell you exactly where. "How your work flows" is Verinode's process-mining read on your own book: it takes the lifecycle dates already sitting on your jobs, supplements, recruiting, safety, purchasing, leads, and reviews, and turns them into stage-by-stage timing, so you can see which step is actually eating the calendar instead of guessing.

Verinode does not run your operations or decide what to fix. It reads the dates you already have, does the math honestly (real sample sizes, no invented medians, dates that don't add up excluded rather than averaged in), and shows you where the time goes. You decide what to act on.

Where to find it

Open Processes from the sidebar at iq.verinode.ai/processes. Processes is a Premier-tier feature: if your membership isn't there yet, the page shows a preview summarizing what you have so far and a prompt to upgrade for the full fractional-COO process coach. If the section itself hasn't been switched on for your account yet, you'll see a short activation screen first.

Once the page loads, the rows stack top to bottom:

1. The hero band: how many SOPs you've written and how many service lines still have none.
2. Take Action: process decisions waiting on you, plus quick-launch tiles for writing an SOP or running a QA audit.
3. Explore: four KPI tiles (My SOPs, To Confirm, your state's standards, Lines Without An SOP) plus any patterns Verinode has spotted that need your confirmation.
4. How your work flows: the row this article covers.
5. Most recent: your latest SOPs, spotted patterns, and standards, newest first.

Each tile in "How your work flows" scrolls horizontally. Click any tile and it opens straight into that metric's own detail card inside the Flow tab of the Processes card slider, the same slider you reach by opening any SOP, pattern, or standard from the page (Findings, Flow, All Processes, By Standard, Coverage, and Benchmarks are the tabs across the top).

The roster: one tile per process

The row is built as a process roster, not a list of raw stage names: one tile per process you actually run, not a jumble of "Assigned to Started" next to "Received to Interviewed." The tile shown for a process is its slowest stage by median duration (its bottleneck), because that's the highest-leverage place to look first. Verinode currently mines:

• Jobs: your core job lifecycle, Assigned → Started → Completed → Billed → Paid.
• Exterior: the same lifecycle stages, sliced to just your exterior and roofing jobs, so a roofing-heavy book sees its own timing rather than a blend with water and fire work.
• Supplements: Submitted → Responded, the carrier turnaround on a supplement package.
• Recruiting: Received → Interviewed → Hired, your applicant pipeline.
• Safety: Occurred → Action Closed, how long a safety incident stays open before its corrective action closes.
• Purchasing cycle: Ordered → Invoiced (and Received/Paid where those dates exist), your procurement cycle on material invoices.
• Lead response: Received → First response → Converted, how fast a new lead gets worked and whether it becomes a job.
• Review response: Posted → Responded, how quickly you reply to a customer review.
• Drying: a different shape from the rest (see below), because it isn't a stage-to-stage lifecycle, it's your median days-to-dry.

The Jobs tile: discovered flow or bottleneck stage

The Jobs tile gets the richest treatment. When Verinode has enough case-tied activity on your jobs (not just the five milestone dates, but the actual sequence of steps your work went through, including any that repeat), it shows the discovered flow: the real order your jobs travel, mined from what actually happened rather than assumed from a template. The tile reads the slowest gap in that discovered flow, for example "Jobs · 6d · Slowest: Billed → Paid."

Until that richer read is available, the Jobs tile falls back to the same milestone-based bottleneck every other process tile uses: whichever of Assigned → Started, Started → Completed, Completed → Billed, or Billed → Paid takes the longest on your book.

Reading a tile

Every tile in the row shows:

• The process name as its label (Jobs, Recruiting, Drying, and so on).
• A day count as the headline value, the median time that stage takes.
• A sub-line, either "Slowest: [stage name]" when that transition is the process's bottleneck, or the stage name alone when it's the only transition with enough history.
• A small chart underneath: a marker bar showing your median against the peer median where a peer read exists, a ring for job-path share, dots or segments for count-based tiles like drying logs.
• A color: copper-red when you're running slower on that stage than operators like you, green when you're faster, steel blue when there isn't yet a peer group to compare against. The color never turns red just because a stage happens to be a process's internal slowest, only when it's actually behind peers.
• A peer delta, e.g. "+2d vs Peer," once Verinode has enough peer operators reporting the same transition to compare you against. That comparison switches on gradually as more operators in the network contribute the same data, not on a fixed schedule; until then the tile shows your own number alone rather than a fake comparison.

The Drying tile

Drying isn't a stage-to-stage lifecycle like the others, so its tile reads differently. Once you have enough dried jobs on record, it shows your median days to dry and, where available, median equipment-days per job, plus how many jobs are still drying. Below the sample floor, it shows the raw count of drying logs captured so far instead of a median, because a median from a couple of jobs would be a number wearing a real one's costume.

The empty state

If nothing in the row has cleared the bar to produce a metric yet, "How your work flows" shows one line instead of tiles:

"Add your job history and we'll show how long each step takes, and where jobs slow down."

That's not a broken row, it means none of your registered processes have enough dated history yet. As job dates, supplement dates, recruiting dates, and the rest flow in, tiles appear one process at a time.

Opening a tile's Flow detail

Clicking any tile in the row opens the Flow tab of the Processes slider, drilled straight into that tile's own detail card. Every metric in the row (and several more that don't make the curated home roster) lives in this tab, grouped as Your jobs (every job-lifecycle transition, job-path route, and claim journey), Pace and outcomes, Drying performance, and one group per other registered process (Recruiting, Supplements, Safety, and so on). Swiping from a drilled card pages you through the full list in that same order.

What the detail card shows depends on what kind of metric you tapped:

A stage bottleneck (Jobs, Exterior, Supplements, Recruiting, Safety, Purchasing, Leads, Reviews)

The card leads with the day count and a plain sentence: "Going from [stage] to [stage] takes about N days on a typical job," naming it as your slowest stage when it is one, and stating how many days faster or slower than "operators like you" when a peer comparison exists. Below that:

• "Where your time goes" breaks out every stage transition on that process side by side as bars, so you can see the bottleneck against its siblings, not in isolation.
• "How you compare" shows your median next to the peer median, plus the range most peer operators fall in (their 25th-to-75th-percentile band), when enough peer data exists.
• A closing note states how many of your own cases the number is based on, and that half clear this stage within the median while the slow ones stretch out further, calling out that this is the highest-leverage place to speed up when it's your bottleneck.

A job path

For "Job Path" tiles, the card leads with what share of your completed jobs followed that exact route (e.g. "38% of completed jobs · This is how most of your jobs run"), and, where enough of those jobs have a full cycle, how long that route typically takes start to paid. Below it, "The rest took other routes" bars out where the remaining jobs went. A closing note is explicit that different routes aren't mistakes: a deposit-first job or a same-day mobilization simply runs a different clock than the standard flow, and the whole read is scoped to roughly the last twelve months of completed work.

Pace and outcomes

Some processes clear a stricter bar: enough jobs with both a stage duration and an outcome (margin or customer satisfaction) to say something about how pace relates to results. The card leads with the point spread between your fastest and slowest jobs on that stage (for example, "+4.2 points margin"), then breaks the two bands out as bars with their own sample counts. The card is explicit that this is a pattern in your own book, not proof that one causes the other: "the slow ones are worth a look at what they share."

Drying performance

The overall Drying card leads with your median days-to-dry and equipment-days, then breaks losses out by IICRC S500 water class (Class 1 through 4) so you can see that a Class 4 loss legitimately takes longer to dry than a Class 1. Below the sample floor, it shows your log count and how many jobs have finished drying instead of a fake median. Tapping into a single class tile shows just that class's typical dry time and how many jobs it's based on.

Claim journey

Tapping a "Claim Journey" tile (a recent job with at least two lifecycle dates on record) shows a short read: which stage the job has reached and how many of its five lifecycle stages it's moved through, with an Open job button that takes you straight to that job's own detail card to trace it end to end.

Discovered job flow

When the richer discovered flow is available for Jobs, its detail card leads with the real sequence most of your jobs travel (e.g. "Assigned → Started → Completed → Billed → Paid") and names the single gap that eats the most time, in plain language ("the most time goes into getting paid, about 6 days"). "Where your time goes" breaks every gap in that discovered path into bars. "How you compare" appears when a peer read exists for the slowest gap. "Other routes jobs took" shows the next few most common sequences your jobs actually followed. If any step in your flow tends to repeat before a job moves on, a note calls that out too, a step that comes back usually means a handoff is bouncing.

Data-quality guardrails behind every number

A few things happen before any duration reaches a tile or a detail card, so a single bad date entry can't distort what you see:

• Out-of-order dates never enter a median. If a job's Paid date somehow lands before its Billed date in a way that isn't a legitimate business pattern, that pair is excluded from the stage's duration math rather than silently averaged in.
• Extreme outliers are capped, not deleted. A single date typo (a year-off entry turning a 30-day wait into a 395-day one) is capped to a statistically reasonable ceiling before medians are calculated, rather than dropped from the sample or left to distort the number. This only kicks in once a stage has enough cases to define a sensible range, and it never touches your real long-tail jobs, the legitimately slow reconstructions.
• Job paths are classified, not judged. A payment dated before the bill often means a deposit or advance and is labeled "Deposit-First." A missing start date often means same-day mobilization, labeled "Same-Day Start." Only orderings that genuinely cannot happen on the ground, like a payment or completion dated before the job was even assigned, are flagged as impossible, and those feed a separate data-quality fix-it prompt elsewhere in Processes rather than quietly polluting your timing numbers.

Honesty gates: what won't show up, and why

Every number in this row passes through the same checks before it reaches a tile:

1. Minimum sample. A stage transition needs enough in-order cases before a median is published. Below that, the process has no tile yet, not a fake one.
2. Peer floor. Peer comparisons ("+2d vs Peer") only appear once a sufficient cohort of comparable operators is contributing the same stage. Until then you see your own number alone.
3. Winsorized outliers. One date typo can't distort a small median; extreme values are capped, never simply dropped, so the case still counts toward your sample.
4. Correlation labeled as correlation. Pace-and-outcome findings are always phrased as patterns, never as "X causes Y."

Inside the Flow tab, once at least one process is mining, the tab appends honest notes about what's still missing rather than leaving you to guess: which of your other registered processes will appear once their data flows in, and whether peer comparisons for a transition you do have data on simply haven't unlocked yet because the peer group hasn't formed. Neither note is an error, they're Verinode being honest about where the data (yours or the network's) still needs to catch up.

How to use it

1. 1Scan the roster for the reddest (ember-colored) tile: that is the process running slower than operators like you, or if no peer read exists yet, the stage where your own book spends the most time.
2. 2Click it to open the Flow tab detail card and read the verdict line first: it names the exact gap in days and, when available, how you compare to operators like you.
3. 3Check "Where your time goes" to see whether the bottleneck is isolated or whether every stage in that process runs long.
4. 4If a peer comparison is showing, read the typical range (the fast-to-slow spread most peers fall in) alongside the median, not just the single number, before deciding the gap is worth acting on.
5. 5If a related decision exists (a slower stage than your SOP or your peers), it's already waiting in the Take Action row above this one, first, not buried at the bottom of the page.

Best-practice example

Say your Jobs tile reads "6d · Slowest: Billed → Paid," colored copper-red with a peer delta of "+2d vs Peer." Open it. The detail card confirms billing-to-payment is your slowest stage and that you're running two days slower than operators like you, with most of them collecting in a noticeably tighter window. The "Where your time goes" bars show every other stage in your lifecycle is tight by comparison, so this is the one gap actually worth working. From here you might cross-check the Cash Flow view to see whether that same billing-to-payment gap shows up as aging receivables, or open the decision workspace if Verinode has already turned the pattern into a decision worth acting on.

Related reading

• The decision workspace: how a slower-than-peers stage becomes a decision you can act on
• Understanding your margin: what net margin means when a pace-and-outcomes finding references it
• How benchmarks work: the general mechanics behind every peer comparison on the platform
• Reading a benchmark: how to interpret a median, a percentile range, and a peer delta anywhere in Verinode
• Acting on decisions: what happens after you open a decision tied to a slow stage

Data sources