The answers are in your data.
I start where your data already lives: order history, customer data, stock figures. Off-the-shelf tools show you averages. Real machine learning finds the segments, the risk and the opportunities the averages hide. Click a feature and see how it works.
All figures on this page are examples from a fictional webshop. The real calculations can be tried in the demo, no login needed.
01 · Domain
Customer intelligence
Your order history knows more about your customers than you do. I turn that knowledge into decisions.
Customer segmentation · clustering8,412 customers
Customers analysed8,412re-clustered last night at 03:00
Segments found624 behavioural features
Largest share43%segment 1 · 12% of customers
Revenue share per segment€55,000/month split
Segment 1 · loyal big spenders43%
Segment 2 · steady repeat buyers21%
Segment 3 · bargain hunters9%
Segment 4 · new customers13%
Segment 5 · slipping away8%
Segment 6 · dormant6%
Segments · computed from order historysync to Klaviyo / Mailchimp
Segment 1 · loyal big spenders
1,009 customers · average basket €200 · 43% of revenue
Segment 3 · bargain hunters
2,608 customers · mostly buy with a discount code · 9% of revenue
Segment 5 · slipping away
673 customers · last purchase 94 days ago on average
Found automatically, not hand-made rulesRe-clustered every night
Churn prediction · 30 daysscored last night at 03:00
Customers scored482accounts with active history
In high risk31score above 75%
Top signalLogindeclining usage · weight 0.88
Risk distribution · full basefour tiers
Highest risk · weighted by customer valuewith reasoning
User 2894
logins down 82% in 14 days · 3 open tickets
User 2895
support tickets · declining usage
User 2896
stable activity · using a new feature
Signals: login freq. 0.88 · tickets 0.64 · adoption 0.41Step in while there is time
Lifetime value · 24-month prediction8,412 customers
Average CLV€300all customers · 24 months
Highest segment€1,100segment 1 · ±12%
Lowest segment€85segment 3 · ±22%
Expected value per segmentwith uncertainty
Recommendationdata first
Shift ad budget
from segment 3 to lookalikes of segment 1
Bid cap by value
up to €110 per lead in the top segment
Trained on your own historyUpdated monthly
Recommendations & cross-sell · user 4823segment: cold weather
Order lines analysed41,23024 months of history
Measured uplift+18%vs control group
Basket effect+27%AOV at checkout
Context · what the model knows about the customerlive
Basket: parka winter pro
Viewed ×3: wool scarf
Segment: cold weather
Recommended for user 4823 · actually bought togethersorted by probability
Wool gloves M
bought together by 68% of the segment · 312 joint purchases
Wool scarf
73% of similar customers chose this
Thermal underwear
244 joint purchases · +€35 to the basket
Collaborative + content hybrid · apriori rules on topBlock lists + margin · retrained weekly
Campaign impact · spring campaignmeasured against baseline
Real uplift+34%vs predicted baseline
Cannibalisation9%sales that would have come anyway
Campaign ROI3.1×after discounts and ad spend
Sales per day · baseline dashedcampaign week highlighted
Uplift per segmentwho responded
Baseline: forecast without the campaignReady 48 hours after the campaign ends
Lead scoring · pipelineupdated on every event
Leads scored68this month
Above threshold 7011ready to call
Hottest lead92Jensen ApS · CFO
Call the hottest firstsorted by score
Jensen ApS · CFO
opened 4 emails · visited the pricing page ×2
Pro Shops · CEO
demo request yesterday
Marketing.dk · CMO
downloaded whitepaper
What pulls the score upweights in the model
47 features per leadRetrained daily
02 · Domain
Inventory & sales intelligence
The same engine, pointed at your stock: buy right the first time, and find the money tied up on your shelves.
Demand forecast · weeks 28-33recalculated last night at 03:00
Next 6 weeks+24%YoY · vs the same weeks last year
History used18 monthsper SKU + season
Accuracy85-92%on a 6-week horizon
Sales per week · forecast dashed18 months of data
Recommended purchase orderswindow open now
Parka winter pro
season peak in 6 weeks · lead time 4 weeks · order now
Wool gloves
stable trend · stock lasts 9 weeks
Per SKU · daily granularityUpdated every night
Dead stock · scanrun last night
Capital tied up€11,000in genuinely dead stock
Dead SKUs23of 1,247 in the catalogue
Seasonal items spared6often misclassified as dead
Dead stock · sorted by capital tied uptop 3 of 23
SKU 4412 · winter jacket red XL
220 days without a sale · 48 units in stock
SKU 2091 · sandals brown 42
184 days without a sale · 112 units in stock
SKU 7733 · kids rain set 122
201 days without a sale · 64 units in stock
Seasonal items · NOT deadthe model tells them apart
SKU 8804 · knit beanie grey
quiet now · sold in the same season last year · keep
Tells dead stock from seasonal stockSuggestion: bundle + clearance
Dynamic pricingscan run last night at 02:40
Adjustments ready3for your approval
Margin effect+2.3%estimated if approved
Competitor scanEvery nighton SKUs you select
Elasticity · measured per SKUprice sensitivity
Price suggestionsapproved by you
Parka winter pro
high demand · elasticity 1.4 · €170 to €190
Wool gloves
competitor cut the price by €7
Thermal boots
stable · no change
Your min/max rules per itemNothing changes without approval
Anomaly detection3 series monitored
Events today1,400orders, error rate, response time
Flags today1above the threshold
Thresholdz ≥ 3.0tuned on your history
Error rate · todaydashed line: threshold
Status per seriesright now
!
z: 4.7
Error rate · at 14:23
spike detected · +340% above threshold · alert after 40 sec
✓z: 0.8
Checkout events
normal · p95 ok
✓z: 1.2
API response time
normal · p99 ok
1 alert per 1,400 events · no noiseAlert sent to 2 recipients
Cash flow · 6-month projectionre-forecast daily
Liquidity now€56,000from the accounting system
Avg net/month+€2,000last 6 months
Runway14 monthsat the worst month's burn
Net cash flow per monthjun = forecast
Details · in minus outlast 3 months
Aprilin €52,000 · out €50,000+15,100
Mayin €55,000 · out €51,000+28,280
June · fcin €58,000 · out €54,000+36,100
18 months of history · daily granularityDeviations flagged immediately
Live dashboardlast sync 4 sec ago
Revenue · month€55,000+6% vs previous month
Orders · month1,428approx. 47 per day
Customers in the base8,412from the web shop
Revenue · rolling 30 days+12%
Data sources · monitored3 connected
Shopify
e-conomic
Inventory system
Pulls straight from your systemsNo manual exports
Returns analysis · 90 days4,212 orders analysed
Return rate11.4%480 of 4,212 orders
Tied up in returns€5,100items on their way back
Biggest reasonSize52% of all returns
Return reasons · classified by the modelfrom comments + patterns
Highest return rate · per SKUwith recommendation
SKU 3318 · black dress 36
"too small" in 8 of 10 comments · fix the size guide
SKU 5121 · white shirt M
differs from photo · new product photo suggested
SKU 1092 · slim jeans 32
below average · no action
Reasons learnt from your own returns dataUpdated with every return
Competitor prices · nightly scan3 competitors · 412 SKUs
Competitors monitored3scanned last night at 02:10
Your price positionMidcheapest on 31% of items
Price changes overnight147 items now dearest in the market
Price position · 412 matched SKUsagainst 3 competitors
Changes overnight · worth a lookmatched on EAN
Parka winter pro
competitor A cut €15 · you are now the dearest
Wool gloves
competitor B sold out · room for a price lift
Thermal boots
all three unchanged · position held
Matched on EAN and SKUFeeds straight into dynamic pricing
03 · Domain
Automation & integration
The hours you spend today moving data between systems become minutes.
Workflow · order #4127completed in 8 sec
Orders today47automated · 0 keystrokes
Time per order8 secvs approx. 6 min manually
Saved today4.7 hours6 min × 47 orders
Run · step by steporder #4127 · live
✓2s
Order received
Shopify webhook · validated
✓3s
Invoice created
e-conomic · #inv-4127
✓2s
Shipment created
Shipmondo · GLS parcel label
→1s
Stock updated
3 SKUs written down · real time
Auto-retry on errors · 0 failed steps todayRun 47 times today
System integration · synclast sync 4 sec ago
Systems connected3two-way sync
Uptime · 90 days99.97%queued during outages
Data lost0nothing disappears
Connected systems⇄ two-way · → one-way
Shopify
web shop
e-conomic
accounting
Shipmondo
shipping
Data flow todaylive
OrdersSynced automatically142
CustomersUpdated both ways98
InvoicesCreated automatically67
Shopify online
e-conomic online
Shipmondo online
Changes are queued during outagesRuns 24/7
Auto report · week 27sent Mon 08:00
Weekly revenue€13,000+12% vs week 26
Orders342average basket €40
Conversion3.2%+0.4 pp vs previous week
Generated reports · schedulebuild themselves
Weekly report · week 27
every Monday 08:00 · PDF + Excel · 3 recipients
Monthly report · June 2026
1st working day · all key figures + segments
Quarterly deck · Q2 2026
ready for the board meeting
This week's key figures · gathered from 4 systemsno copy-paste
Revenue€13,000+12%
Orders342+8%
AOV€40+5%
Built Sunday night from source dataTo ceo@ and cfo@ + 1 more
Ticket triage · auto-routingqueue · 23
In queue right now23routed continuously
Accuracy94%on your historical tickets
When in doubtManualbelow 80% confidence
New enquiry · classified live#4823
New ticket #4823
"Can't log in, forgot my password"
Routing · done automatically1 sec after receipt
→1s
Sent to tech-tier-1
SLA 4 hours · priority normal
Fine-tuned on your own tickets6 categories learnt from data
Fraud detection · real time1,437 tx · 30 days
Scored today46all transactions
Flagged1parked, not declined
False positive2.1%learns from every decision
Transactions · sorted by riskthreshold 0.72
!
0.94
Flagged · tx #8923 · €560
unknown IP · deviating pattern · parked for approval
✓0.08
tx #8924 · €85
normal · approved automatically
Why was #8923 flaggedsignal weights
Threshold 0.72 · adjustable to your risk appetiteNo blind declines
04 · Domain
Search & document AI
Smarter ways to find the right thing, and documents that turn themselves into structured data.
Semantic search · live22 ms total
Products in the index1,247your own catalogue
Response time22 mssearch-as-you-type
Matches onMeaningnot just letters
Search right nowrelevance per result
⌕ black winter jacket size menter ↵
Parka winter pro
size M · €170 · black · winter
Down jacket black
size M · €130 · down fill
Pipeline per searchruns partly in parallel
bge-m3 · multilingualDanish + English mixed
Auto-suggest · did you meanas the customer types
Median response time38 msbetween two keystrokes
SourceSearch logsyour most popular
Empty results0always a suggestion
The customer types · with a spelling mistakecorrected live
⌕ winter jaket|
Did you mean: winter jacket?
1,247 results behind the correction
winter jacket size m
432 searches this month
winter jacket black men
318 searches this month
BM25 + embedding hybridCorrected before the customer sees an empty page
Invoice scanning · OCRfaktura_0412.pdf
Invoices today132parsed automatically
Per document1.2 secvs 3-4 min manually
Average confidence98%below 90% flagged for review
Scanning · faktura_0412.pdffields highlighted
Extracted fields · with confidenceready for posting
SupplierJensen ApS99%
Invoice no.2026-041299%
Amount€1,70098%
VAT€33098%
Due date04.08.2026 · handwritten field94%
Straight to your accounting systemAs a draft for approval
Custom OCR · your formattrained on 500 docs
Standard OCR71%on your document format
Custom-trained98%after 500 examples
Uplift+27 ppfield recognition
Your document · fields recognisedlayout matched
Same documents · before and after trainingfield accuracy
In practice
manual post-processing drops from every 3rd to roughly every 50th document
Corrections become training dataGets better in production
Classification & tagginginbox + library
Documents today61sorted on receipt
Files tagged4,200the entire media library
When in doubtManualbelow 90% confidence
Distribution today · per category61 documents
Most recently classified and taggedwith confidence
kontrakt_leverandor.pdf
routed to: Contracts
faktura_q2_ny.pdf
routed to: Bookkeeping
kampagne_maj_04.jpg
tagged: studio · model · parka · black
No manual sorting · new files tagged on uploadThe same model can route emails
"It is a capital mistake to theorize before one has data."
Sherlock Holmes · Arthur Conan Doyle, 1891
[Method] / spot → prototype → build
Three steps. You pay nothing until step three.
And one principle throughout: baseline first. I measure what the task costs in hours and money before I build, and measure again after. That way the effect is a number in your accounts, not a feeling.
Step 1
Spot
€0
A 30-minute call + a written report.
- 3 concrete proposals
- Impact estimate per proposal
- Recommended next step
- Price estimate
Step 2
Prototype
€0
A working demo on your own data.
- Live demo or hosted preview
- Built in 1-4 hours
- Video walkthrough
- No obligation
Step 3
Build
Fixed quote · agreed in advance
A production-ready ML solution.
- Delivered in 1-4 weeks
- Full documentation
- 30-day bug-fix guarantee
- No vendor lock-in
Ready to see what your data is hiding?
You get concrete proposals, whether or not we work together afterwards.