Executive overview
Operational health across pipeline, service and productivity.
Pipeline value$6.46Mdown 14.1% on last month
Win rate31.4%up 2.8 points
SLA attainment88.6%up 1.5 points
Activities per rep41down 4.5%
Revenue against target
10 monthsClosed-won revenue by month against a $1.9M monthly target. Three months under, seven over.
Deal source mix
Share of pipeline$6.46Mtracked
Outbound44%
Inbound28%
Referral16%
Partner8%
Event4%
Stalled deals
5 stalledArdent Capital, renewalNils Aaberg · 333 days in Appointment scheduled86x normal
Belclare Foods, new businessIngrid Solberg · 309 days in Appointment scheduled80x normal
Corvin Logistics, new businessChiara Moretti · 302 days in Contract sent60x normal
Dunmore Retail, renewalBram de Vries · 287 days in Contract sent57x normal
Aging is measured against the normal pace of that stage, not a flat number of days. A deal is stalled relative to how that stage usually behaves.
Recent activity
Ingrid SolbergClosed Eastvale Health, $146,000 · todayWon
Nils AabergLogged 6 calls · todayActivity
Bram de VriesMoved Corvin Logistics to Contract sent · yesterdayStage
Chiara MorettiHeld 3 meetings · yesterdayActivity
What moved this week
Rule-derivedWin rate is up 3.1 pointsAlmost entirely Enterprise, which closed faster than its own trailing averageUp
SLA attainment fell 1.4 pointsHigh-priority tickets in the support queue account for the whole dropDown
Presentation scheduled is 12% slowerAgainst the trailing quarter. It is the only stage moving the wrong wayWatch
Five deals passed 280 days in stageAll five sit with two reps and two stages, so it is not a general slowdownWatch
These are threshold rules over the same figures above, not a model. Each one names the metric it came from so you can disagree with it.
Every figure on this screen resolves to a definition in the metric catalogue. If you disagree with a number, the argument is about the definition, and it happens once.
Pipeline
Funnel health, stage velocity and rep performance.
Open value$4.61Mdown 8.5%
Weighted value$1.84Mdown 13.4%
Average sales cycle38.4 daysup 2.7%
Average stage velocity5.6 daysup 0.6%
Stage funnel
| Stage | Deals | Value | Average days | Conversion |
|---|---|---|---|---|
| Appointment scheduled | 13 | $1,641,000 | 3.9 | 68% |
| Qualified to buy | 15 | $1,537,000 | 7.2 | 74% |
| Presentation scheduled | 15 | $1,436,000 | 7.8 | 61% |
| Contract sent | 17 | $1,842,000 | 5.0 | 58% |
| Closed won | 44 | $4,709,000 | 1.9 | - |
Average days is time in stage, not time since creation. A stage that is quick and converts badly is a different problem from one that is slow.
Quota attainment
This quarterDeals by owner
Won, open and lostStalled deals
5 stalledArdent Capital, renewalNils Aaberg · 333 days in Appointment scheduled · $184,000
Belclare Foods, new businessIngrid Solberg · 309 days in Appointment scheduled · $96,400
Corvin Logistics, new businessChiara Moretti · 302 days in Contract sent · $142,000
Service and SLA
Response times, attainment and backlog health.
First response4.2 hoursdown 9.1%
Resolution time1.8 daysup 5.0%
SLA attainment88.6%up 0.2 points
Open backlog112up 9.7%
Tickets created against resolved
14 daysCreated per day. Resolution has not kept pace since day nine, which is where the backlog came from.
Attainment by priority
Resolved within targetUrgent is the worst-served priority, which is the opposite of what the target intends.
Backlog by age
Open ticketsBreaching next
3 at riskExport stuck processingUrgent · Dunmore Retail · 2 hours to targetAt risk
Integration webhook not firingHigh · Corvin Logistics · 5 hours to targetAt risk
Cannot sign in to the dashboardHigh · Belclare Foods · 8 hours to targetAt risk
Attainment is measured against the target that applied when the ticket was raised. Changing a target does not retroactively improve last quarter.
Data quality
Completeness, staleness and ownership gaps across records.
Overall completeness97.5%up 0.4 points
Records needing attention20across deals and tickets
Stale open deals60no activity in 30 days
Orphaned owner records5no matching active user
Field completeness
Required fields populatedOrphaned owners
7 recordsowner-deactivated-3841No matching active user · 5 records
owner-deactivated-2907No matching active user · 2 records
A deactivated owner leaves records behind. Reporting that quietly drops them is worse than reporting that shows them as unassigned.
Records needing attention
| Record | Type | Owner | Issue | Last touched |
|---|---|---|---|---|
| Cannot sign in to the dashboard | Ticket | Ingrid Solberg | Missing priority | 3 days ago |
| Dashboard loading slowly | Ticket | Bram de Vries | Missing priority | 4 days ago |
| Export stuck processing | Ticket | Chiara Moretti | Missing priority | Yesterday |
| Fenwick Group, new business | Deal | Chiara Moretti | Missing amount | 11 days ago |
| Aviato Systems, new business | Deal | Unassigned | Orphaned owner | 38 days ago |
| Tyrne Corporation, renewal | Deal | Unassigned | Orphaned owner | 42 days ago |
This screen exists because most reporting disputes turn out to be data problems. Naming the record and the missing field ends the argument faster than a chart does.
Metric catalogue
What each number means, who owns it and when it changed.
StatusAllAgreedIn reviewDeprecated
Metric catalogue
6 metrics| Metric | Definition | Owner | Status | Last changed |
|---|---|---|---|---|
| Pipeline value | Sum of open deal amounts, excluding Closed lost | Revenue operations | Agreed | 14 Jun 2026 |
| Win rate | Closed won divided by Closed won plus Closed lost, by close date | Revenue operations | Agreed | 02 Mar 2026 |
| SLA attainment | Tickets resolved within the target that applied when raised | Service operations | Agreed | 19 May 2026 |
| Sales cycle | Days from deal creation to Closed won, median not mean | Revenue operations | In review | 28 Aug 2026 |
| Activities per rep | Calls, emails and meetings logged, per active rep per working day | Revenue operations | Agreed | 11 Jan 2026 |
| Qualified pipeline | Superseded by Weighted value | Revenue operations | Deprecated | 05 Feb 2026 |
A deprecated metric keeps its definition and its history. Deleting it makes every chart that used it unreadable a year later.
Alerts
Thresholds, who is notified and what is firing.
StatusAllFiringAcknowledgedResolvedMuted
Firing now
2 firingUrgent SLA attainment below 70%Currently 58.8% · firing for 2 days · service leads notified
Billing export failedNightly run at 02:00 · firing for 6 hours · revenue operations notified
| Alert | Metric | Condition | Notify | Status |
|---|---|---|---|---|
| Urgent SLA attainment | SLA attainment | Below 70% for 1 day | Service leads | Firing |
| Source failure | Source health | Any failed run | Revenue operations | Firing |
| Pipeline coverage | Weighted value | Below 3x quota | Sales managers | Acknowledged |
| Backlog growth | Open backlog | Up more than 15% week on week | Service leads | Resolved |
| Completeness | Field completeness | Below 95% on any required field | Revenue operations | Muted |
Scheduled delivery
Reports that go out on their own, with what they contained.
A schedule sends the report as it stood at send time, with the filters saved into it.
Scheduled delivery
| Report | Recipients | Frequency | Format | Last sent | Status |
|---|---|---|---|---|---|
| Executive overview | Leadership, 6 people | Monday 08:00 | Monday | Active | |
| Pipeline by team | Sales managers, 4 people | Daily 07:30 | CSV and PDF | Today | Active |
| SLA attainment | Service leads, 3 people | Monday 09:00 | Monday | Active | |
| Data quality exceptions | Revenue operations, 2 people | Daily 06:00 | CSV | Today | Active |
| Quarterly board pack | Board, 7 people | Quarterly | 01 Jul | Paused |
A scheduled send is recorded with the figures it contained. When somebody quotes last Monday’s number back at you, you can see exactly what they were sent.
Data sources
Where the numbers come from and how fresh they are.
Sources65 healthy
Last full sync11 minutes agoCRM
Failed runs today1billing export
Rows loaded today1.84Macross all sources
Data sources
1 failing| Source | Type | Schedule | Last run | Rows | Status |
|---|---|---|---|---|---|
| CRM | API | Every 15 minutes | 11 minutes ago | 412,800 | Healthy |
| Service desk | API | Every 15 minutes | 9 minutes ago | 186,200 | Healthy |
| Billing export | SFTP file | Nightly, 02:00 | 6 hours ago | 0 | Failed |
| Product telemetry | Warehouse | Hourly | 38 minutes ago | 1,102,400 | Healthy |
| Directory | API | Nightly, 01:00 | 7 hours ago | 1,284 | Healthy |
| Marketing | API | Every 6 hours | 2 hours ago | 142,600 | Stale |
A failed source shows on every screen that depends on it, rather than quietly reporting yesterday as if it were today.
Freshness is stamped on every screen. "Data as of" is not decoration: a dashboard that cannot tell you how old it is should not be trusted.
Governance
Access, rules and the definition change log.
Access
AdminEvery team, every source, the catalogue and the alerts2 people
ManagerTheir team’s pipeline, service and quality9 people
RepTheir own deals and tickets, and the catalogue34 people
Rules
Row-level scopeApplied on the query, not the interfaceOn
Definition change requires reviewTwo approvalsOn
Freshness stamp on every screenPer sourceOn
Export includes the filters appliedIn the file headerOn
Audit log
Append only| When | Who | What changed | Detail |
|---|---|---|---|
| Yesterday | Ruth Vasari | Metric definition | Sales cycle moved from mean to median |
| 3 days ago | Chiara Moretti | Alert | Completeness alert muted until 30 September |
| Last week | Ruth Vasari | Access | Mid-market manager scope extended to Service |
| Last week | Ruth Vasari | Source | Marketing sync moved from hourly to six-hourly |
Definition changes are the ones worth logging. A chart that changed shape last quarter usually changed because somebody edited the definition.
On-time delivery
One metric, its definition and everything that trusts it.
CertifiedDefinition changed 6 weeks agoMET-0142 · On-time delivery · owner Astrid Vang · used on 18 dashboards
On-time delivery
Rose from 88% to 94% in July. Nothing about the operation changed; the definition started measuring against the revised promised date rather than the original one.
Current value94.1%target is 92%
On the old definition88.4%below target
Dashboards using it183 of them external
History restatedNothe step is still in the chart
What changed
Before 14 JulyMeasured against the original promised date88.4%
After 14 JulyMeasured against the latest revised date94.1%
Approved byNobody · changed in a pipeline editUnapproved
AnnouncedNeverSilent
A six-point improvement that came from a definition change, published to three external dashboards, and nobody was told. That is the failure, not the definition.
Which is right
Against the original dateWhat the customer was promisedHonest
Against the revised dateWhat we last committed toAlso valid
Both, side by sideThe gap is the useful numberBest
Silently switchingNeitherWhat happened
Either definition is defensible. Changing from one to the other without restating the history is what turns a measurement into a claim.
What depends on this
Operations dashboardInternal · 41 weekly viewers
Board packMonthly · shows the step as improvementMisleading
Customer service reports3 customers · contractualExternal
Bonus calculationOperations team · threshold 92%Paid on it
A definition change moved a bonus threshold from missed to met. Nobody chose that and it is not reversible quietly.
Define a metric
What it counts, who owns it, and when it may change.
The metric
What certification means
Definition in words and in codeBoth, and they must agreeRequired
A named ownerA person, not a teamRequired
Change requires approvalAnd a restated historyRequired
Consumers are notifiedEvery dashboard, every reportRequired
Provisional metrics need none of this and are marked provisional wherever they appear. The label is the whole mechanism.
When the definition changes
ConsumersNotify and require acknowledgementNotifyDo not notify
Do nothing is available and is never the right answer for a certified metric. It is listed so that choosing it is an act rather than an oversight.
A metric with no owner cannot be certified.
Lineage
Where a number comes from, and what breaks if a source does.
On-time delivery4 sources, 18 consumersTraced from source tables to the board pack
Trace
1 stale source| Layer | Object | Owner | Last changed | Freshness | Status |
|---|---|---|---|---|---|
| Source | orders (warehouse) | Data platform | 14 Jul | 2 hours | Healthy |
| Source | deliveries (warehouse) | Data platform | 14 Jul | 2 hours | Healthy |
| Source | promised_dates (manual upload) | Operations | 22 Aug | 16 days | Stale |
| Transform | fct_delivery_performance | Astrid Vang | 14 Jul | 2 hours | Changed |
| Metric | On-time delivery | Astrid Vang | 14 Jul | 2 hours | Certified |
| Consumer | Board pack, 3 customer reports | Various | - | Monthly | External |
One source is a manual upload that is sixteen days stale, and it feeds three external customer reports. Lineage is only interesting when it shows you that.
The transform and the metric changed on the same day the definition did. That is how the change is traceable at all, and it was found six weeks later by somebody asking why the chart had a step in it.
Data quality
What is missing, late or contradictory.
Tests running284across 62 tables
Failing186 for over a week
Tables with no tests22of 84
Certified metrics affected4of 34
Failing tests
18 failing| Test | Table | Expected | Found | Failing for | Affects |
|---|---|---|---|---|---|
| Not null | promised_dates.date | 100% | 94.2% | 16 days | On-time delivery |
| Freshness | promised_dates | Daily | 16 days old | 16 days | On-time delivery |
| Unique | orders.order_id | 100% | 99.98% | 3 days | 4 metrics |
| Accepted values | deliveries.status | 6 values | 9 values | 11 days | 2 metrics |
| Row count | returns | ±20% | −62% | 2 days | Nothing yet |
The first two are the same upload failing in two ways, and it is the source behind the metric on three external customer reports.
Twenty-two tables have no tests at all. A passing test suite over two thirds of the warehouse is a statement about the two thirds, and it is usually read as a statement about all of it.
Dashboard adoption
What is built, what is opened, and what is trusted.
Dashboards188across 9 teams
Not opened in 90 days9651% of them
Using uncertified metrics6233%
Duplicate definitions14of the same measure
By dashboard
96 dormant| Dashboard | Team | Viewers | Last opened | Metrics | Verdict |
|---|---|---|---|---|---|
| Operations weekly | Operations | 41 | Today | All certified | Keep |
| Board pack | Executive | 9 | 2 days ago | All certified | Keep |
| Delivery deep dive | Operations | 2 | 4 months ago | 3 uncertified | Retire |
| Regional split v2 | Commercial | 0 | Never | 5 uncertified | Retire |
| Costs, old | Finance | 1 | 7 months ago | Duplicates ledger | Retire |
Ninety-six dashboards have not been opened in three months and each one is a query running on a schedule against tables somebody is maintaining.
Fourteen dashboards define the same measure fourteen slightly different ways. That is not a tooling problem; it is what happens when building a new one is easier than finding the existing one.
Reports: activity
Volume and throughput, by owner and by domain.
Certified metrics34of 188 defined
Opened this month188against 96 closed
Median age-target is,
Owned by nobody41oldest is 4 months
By metric owner
This quarter| Metric owner | Open | Closed | Median days | Overdue |
|---|---|---|---|---|
| Astrid Vang | 12 | 18 | 2 | 1 |
| Marek Sobol | 9 | 14 | 4 | 2 |
| Iona Ferris | 8 | 11 | 3 | 0 |
| Petra Nagy | 5 | 9 | 6 | 3 |
| Unassigned | 41 | 0 | - | 41 |
Forty-one metrics have no owner and none of them can be certified. That is the largest row in the table and the only one nobody is accountable for.
Volume over time
12 weeksQueries run per week. Half of them serve dashboards nobody has opened in three months, and every one of those is a table somebody is maintaining.
By domain
This quarterReports: performance
Against target, with the gap named.
Against target18%of defined metrics are certified
Best metric ownerFinance41% certified
WorstCommercial6% certified, most dashboards
Trend+4 ptson last quarter
Against target by domain
100 is the targetThe team with the fewest certified metrics has the most dashboards. Building is easier than agreeing, and the ranking here is what that produces.
What moved
Rule-derivedA certified metric changed definition silentlySix-point step, three external reports, no restatementDown
Certified metrics up from 24 to 34Mostly finance and operationsUp
96 dashboards unopened for 90 daysEach still running on a scheduleWatch
Threshold rules over the figures above, not a model. Each line names the number it came from so you can disagree with it.
Reports: exceptions
What breached, what is late and what nobody owns.
Breached18data quality tests failing
At risk41inside 48 hours
Unassigned41no owner at all
Reopened14closed then raised again
Exceptions
18 open| Object | Problem | Failing for | Affects | Owner |
|---|---|---|---|---|
| promised_dates upload | 16 days stale, 6% null | 16 days | 3 external reports | Operations |
| On-time delivery | Definition changed, history not restated | 6 weeks | Board pack, a bonus | Astrid Vang |
| 41 metrics | No owner, cannot be certified | 4 months | 62 dashboards | Nobody |
| 22 tables | No data quality tests at all | Always | Unknown | Data platform |
| 14 definitions | Same measure, different answers | Ongoing | 14 dashboards | Various |
The first two rows are one incident: a stale manual upload behind a metric whose definition changed in the same fortnight, feeding reports outside this company.
A definition change moved a bonus threshold from missed to met and nobody chose that. Restating the history is the expensive fix; notifying consumers at the time was the cheap one, and it is a checkbox on the definition screen.
People
Who has an account, and what they own.
People
34 accounts| Name | Role | Domain | Last active | Status |
|---|---|---|---|---|
| Vera Lindqvist | Administrator | Data platform | 11 minutes ago | Active |
| Astrid Vang | Analytics lead | Operations | 25 minutes ago | Active |
| Marek Sobol | Analyst | Commercial | 1 hour ago | Active |
| Iona Ferris | Analyst | Finance | 2 hours ago | Active |
| Petra Nagy | Analyst | People | Yesterday | Active |
| Osian Pryce | Data engineer | Data platform | 40 minutes ago | Active |
Deactivating somebody prompts you to reassign what they own rather than doing it silently, because an unowned record is how a register rots.
Roles
What each role is scoped to and what it may approve.
Recent changes
Certified metric definitions locked behind approvalVera Lindqvist · 6 SeptemberLogged
Analysts allowed to define provisional metrics freelyVera Lindqvist · 2 MayLogged
Role changes sit in the same audit log as record changes, because both change what somebody can do.
Why custom roles exist
- An analyst can define anything as provisional, and it is labelled everywhere
- Only an analytics lead can certify, and certification is what a board pack requires
- A data engineer can change a transform and cannot change what a metric means
- Nobody can change a certified definition without notifying every consumer
Roles
5 roles| Role | Type | People | Scope | Can approve to | Status |
|---|---|---|---|---|---|
| Administrator | Built in | 2 | Everything | No limit | Active |
| Analytics lead | Built in | 3 | All metrics | Certifies metrics | Active |
| Analyst | Built in | 14 | Own domain | Defines provisional | Active |
| Data engineer | Built in | 6 | Pipelines and tables | Changes transforms | Active |
| Viewer | Custom | 9 | Dashboards only | - | Active |
A role carries a scope and an approval ceiling. Both are enforced on the server; hiding a button is not access control.
Permissions
Every capability against every role, as enforced.
What each role can do
10 capabilities| Capability | Administrator | Analytics lead | Analyst | Data engineer | Viewer |
|---|---|---|---|---|---|
| View a dashboard | Yes | Yes | Yes | Yes | Permitted |
| Define a provisional metric | Yes | Yes | Yes | No | No |
| Certify a metric | Yes | Yes | No | No | No |
| Change a certified definition | No | With approval | No | No | No |
| Change one without notifying | No | No | No | No | No |
| Edit a transform | Yes | Yes | Own domain | Yes | No |
| Publish externally | Yes | Certified only | No | No | No |
| Retire a dashboard | Yes | Yes | Own | No | No |
| Add a data quality test | Yes | Yes | Yes | Yes | No |
| Read the change log | Yes | Yes | Yes | Yes | Yes |
Enforced per query rather than by hiding controls. A scoped "yes" means the row exists but is limited, which is why some actions need a second person.
Only certified metrics can be published outside the company. The on-time delivery figure met that bar and still changed meaning silently, which is why the row below it exists.
Notifications
What the system sends, to whom, and when.
What is sent
Certified metric definition changesTo every dashboard owner using itOn
A source fails freshnessTo the source owner and affected metric ownersOn
Metric published externally changes value over 3%To the analytics leadOn
Dashboard unopened for 90 daysTo its owner, monthlyOn
Metric has no ownerTo the analytics lead, weeklyOn
Every message names the record it is about, so a reply lands somewhere.
Quiet hours
No routine notifications20:00 to 07:00, local to the personOn
Urgent overridesA change to a metric on an external customer reportOn
Weekly digestMonday 08:00On
A notification that waits until morning is one that did not work, so the override is explicit rather than a setting nobody finds.
Notification rules are yours to set. What is here is a starting point drawn from how the queue actually behaves.
Integrations
Connected systems and their sync history.
Connected
5 connectedData warehouseTables, transforms and lineageConnected
Source systemsOrders, deliveries, ledger and payrollConnected
Dashboard toolWhat exists, who opens it, and what it usesConnected
Version controlTransform history and who changed whatConnected
Manual uploadsSpreadsheets that feed certified metricsConnected
What breaks without them
- Without version control a definition change is invisible, which is how a six-point step went six weeks without an explanation
- Without dashboard telemetry there is no way to retire anything safely
- Manual uploads are connected and are still the weakest source, because nothing makes a person upload on time
- Numbers keep appearing; nobody can say where they came from
Sync history
Last 24 hours| When | Source | Records | Result |
|---|---|---|---|
| Every 2 hours | Data warehouse | 84 tables | 18 tests failing |
| Continuous | Source systems | 9 systems | Live |
| Daily | Dashboard tool | 188 dashboards | 96 dormant |
| Expected daily | Manual uploads | promised_dates | 16 days late |
A sync that half-succeeded is listed as such rather than as a failure, because the records that did land still changed something.
Import and export
Bulk load and extract, with per-row validation.
Import
Validated before anything is written. A file with one bad row imports the rest and tells you which line failed and why.
Export
RangeEverythingThis yearFiltered view
Recent imports and exports
Last 30 days| When | What | Rows | Result | By |
|---|---|---|---|---|
| 4 Sep 08:00 | Import | 188 | Complete | Vera Lindqvist |
| 1 Sep 18:00 | Export | 34 | Complete | Astrid Vang |
| 22 Aug 09:20 | Import | 284 | Complete | Osian Pryce |
Every import is kept with its source file, so a bad load can be traced rather than argued about.