How the employee score is worked out

Every Monday morning BudLogix scores each person on the floor for the week that ended on Sunday, out of 100. Each person is scored on the parts their position is measured on, plus their evaluation, which every position has. What a position is measured on is set on the position; see Choose what a position is measured on.

Who is scored

People who work the floor: active, on staff, with a Dutchie login and a till role. Owners and anyone in the Not compared group are left out. See Comparison groups and measured figures.

The parts

Each part scores 0 to 100. Each reads as far back as it needs to count enough to trust.

• Evaluation (every position) - for each criterion, the newest manager rating: the last quarterly review, or a check-in written after it. The average level maps 1 to 0 and 5 to 100. Self-evaluations never count. Carried until the next rating. See Check in on an employee.
• Selling (4 weeks) - three measures averaged: the average order against what the store sells on the same shifts, cross-sell (baskets with two or more product families), and upsell (each basket against the same customer's usual basket with someone else). Average order and upsell only move the score when they are outside the range chance explains at this volume; inside it they read In line with the store and score 75.
• New-customer perks (8 weeks) - the share of new-customer visits that owed a perk and got it. The standard is every one, so 7 of 10 scores 70.
• Loss prevention (12 weeks) - sales that broke a discount rule, per 1,000 lines rung. No incidents scores 100 and the figure it is compared with scores 75. One sale line is one incident however many rules it broke. Critical counts 3, high 1 and medium 0.5. A sale a manager accepted, and a new-customer perk given on its own visit, do not count. An explanation that is overdue counts one more, and keeps counting until it is answered, however old the sale.
• Cash accuracy (4 weeks) - the share of shifts that closed balanced or over, by the same rule as Reconciliation: within $1.00 is balanced, and a drawer that comes up over is not a loss. A short shift is not clean, and neither is one whose cash and debit offset - the drawer was short even though debit covered it. The worst shortfall shown is the cash missing. Shifts not counted yet are left out.
• Evaluations on time (carried) - of the people this person evaluates, the share whose quarterly evaluation was submitted within 30 days of the quarter closing (Q3 by 30 October). A draft is not submitted. The latest quarter whose due date has passed holds until the next one; due dates start with Q3 2026.
• Exception follow-up (live) - explained exceptions on sales by the people this person evaluates, given an accepted or not-accepted decision within 7 days of the explanation. One still undecided past its 7 days counts every week until it is decided; decided ones count for 8 weeks.
• Receiving accuracy (8 weeks) - delivery lines this person counted in BudLogix against the manifest, and the differences they left unresolved, per 100 lines. Lower is better.
• Stock records (8 weeks) - Dutchie inventory adjustments this person made, and the share METRC rejected, per 100. An adjustment made for an entry error is shown but never counted against anyone: it corrects an earlier mistake, often someone else's.
• Product records (due this week) - a duty, shared by everyone whose position is scored on it: lots on the Missing THC in Dutchie and Expiring inventory lists fixed within their window. A lot missing THC that METRC has a result for has 1 day, one where somebody reads the label 3 days, and a Dutchie expiration date that disagrees with METRC 2 days; the windows are set under Employee scores. A lot counts in the week its window ends, and 9 of 10 on time scores 90. A lot somebody acknowledged is excused and listed with who acknowledged it; a lot that left the shelf, or moved from reading the label to Copy in from METRC, is left out. Nothing due that week leaves the part out.
• Invoices paid on time (8 weeks) - vendor invoices this person marked paid on or before their due date.
• Punctuality (4 weeks) - published shifts this person clocked in to on the store's time clock no more than 5 minutes after the start. A shift with no clock-in counts as not on time. A store that has never used the clock is not scored on it.

Parts that say "the standard is every one" are read against that rule. The others are compared with other people, below.

Who probably fixed a lot

Dutchie does not record who changed a lab result or an expiration date. So each lot's link in the Missing THC and expiry emails and reports goes through BudLogix, which notes who opened it, then opens the same page in Dutchie. An emailed link carries a key for the person it was sent to, so nobody signs in; opens by email security scanners, and opens within a minute of the email arriving, are ignored. When a lot is fixed, the last person who opened its link in the day before is shown as having probably fixed it, and a lot nobody opened first reads "Fixed directly in Dutchie". Open the part on the employee record to see the count per person and every lot behind the score. It is shown, never scored.

Who a part compares someone with

Loss prevention, cross-sell, receiving accuracy and stock records are read against one figure that blends two things:

• the standard - what is normal at this store for the group, set under Employee scores;
• the colleagues - the median of the other people in the same comparison group whose position is scored on that part and who had enough activity. Receiving accuracy takes colleagues from the whole shop: it is scored on whoever receives, whatever their title. Loss prevention never leaves the group: managers ring Manager's Discount as part of the job, and putting that beside a budtender's figure would read a role as conduct.

The more colleagues there are, the more they count: with n colleagues they count n / (n + k), and the standard counts the rest. k is the number of colleagues it takes before they count as much as the standard, also set under Employee scores. Somebody alone in their group is read against the standard alone, so nobody waits weeks for a score. The part says what it was read against, for example "against the Budtenders standard (2.0) and 3 colleagues; colleagues count 43%".

Under the part, the scorecard also says whether the week was better or worse than the person's own usual, once they have 3 weeks. That line is a trend only and never moves the score.

Not enough activity

A part with too little to judge is left out, never counted as zero or as full marks: fewer than 200 orders, 8 new-customer visits, 300 lines rung, 4 shifts, 5 explanations due a decision, 40 delivery lines counted, 10 adjustments or 10 invoices in its window. Product records has no minimum: one lot due is enough. The score is the weighted average of the parts that are left. There is no score until at least 3 parts have enough activity, or all of them when the position is measured on fewer than 3. A new hire reads None yet.

Figures

Figures count how much somebody did - orders completed, shifts overseen, cash drops, items received, floor moves and so on - over the last four weeks. More is not better, so they are shown under the parts and beside the evaluation and never scored.

Weights and comparison

The parts count equally unless an org admin changes it under Employee scores. Equal weights do not mean equal effect: a part people are far apart on decides more of the difference between two scores than one they are close on. That page shows how much each part really moves scores, and can suggest weights that balance it. The group's median and middle half on the scorecard are worked out over everyone in the group with a score, and are withheld below three people.

Each week's score is stored with every figure that produced it, what the position was measured on, the weights and the rules used, so an old score can be explained exactly. A check-in, an accepted exception or a change to what a position is measured on scores the latest week again straight away. Scores are never used when hiring.

Related

• Employee record
• Position
• Roster
• My scorecard
• Respond to an exception