1. Ward
  2. /
  3. Connectors
  4. /
  5. Looker
  6. /
  7. Shrinkage
BI · Looker

Run shrinkage straight off Looker

Find the leak before it drains you. Ward runs it directly off your Looker feed, read-only, with no changes to your BI.

Request access Looker connector reference
Read-only by default No copy of your data Ward AI drafts, you approve
01 · The connection

How Ward attaches to Looker / Looker Studio.

Ward can query Looker via API or connect directly to the underlying database. Either way, Ward monitors while your team browses.

app.getward.ai Live demo
Acme Retail @Merchandising: VP Analyst claude-sonnet default
A

AI Insights

Agents run against your baselines overnight. These are what they flagged without being asked.

3 flagged 8 queries run 3 sources swept 04:00, acme-retail
Act now labor_efficiency 0.94

Rev/labor-hour −22% vs. cluster, staffing mismatch at 11a–1p peak

Schema Scout · routed to Merchandising Agent Pin Ask Ward Investigate
Review inventory.fresh 0.89

Fresh fill 83%, backroom replenishment lag at 2–4p

Promo Agent · routed via Merchandising Pin Ask Ward Investigate
Watch promo.lift 0.81

BOGO crackers cannibalized Brand Y by 28%, net category +6%

Margin Agent · routed via Finance Pin Ask Ward Investigate
Recommended

Re-baseline Store 37 schedule against true peak, raise replen window to 1p, and review the BOGO before next cycle.

Sources

Connect external systems. Ward AI maps the schema and drafts the cleaning rules.

Search sources…
NameTypeLast sync
sap_pos_transactionsimport2m ago
sap_inventory_shrinkageimport2m ago
sap_labor_schedulingimport14m ago
retail_inventory_weeklyimport1h ago
retail_google_ads_dailyimport1h ago
retail_meta_ads_dailyimport1h ago
retail_ga4_website_dailyimport1h ago

Pipelines

Move data from sources into models on a schedule. Describe the sync; Ward AI builds it.

NameSourceModelStatusSchedule
sync_sap_pos_transactionssap_pos_transactionspos_transactionsenabledhourly
sync_sap_inventory_shrinkagesap_inventory_shrinkageinventory_shrinkageenableddaily
sync_sap_labor_schedulingsap_labor_schedulinglabor_schedulingenableddaily
sync_retail_inventory_weeklyretail_inventory_weeklyinventory_weeklyenabledweekly
sync_retail_google_ads_dailyretail_google_ads_dailygoogle_ads_dailyenableddaily
sync_retail_meta_ads_dailyretail_meta_ads_dailymeta_ads_dailyenableddaily
Sources, then what Ward AI drafted from them, then what it computed. Click between the panels.
What Ward reads
Looker API for query resultsUnderlying database (direct)LookML model metadata

Ward does not replace Looker. Ward watches the same data Looker visualizes and proactively alerts when something changes. Your dashboards stay. Ward adds intelligence.

Scope

Credentials are read-only and locked to the objects you name. Ward cannot reach a table you did not grant it, and write access is opt-in per object rather than implied by the connection.

Residency

Ward queries Looker in place and keeps no second copy. Your region, your retention policy, and no staging warehouse in the middle.

Drift

Every connector carries a schema contract. A source that changes shape fails the contract and stops the pipeline, rather than quietly writing wrong rows until someone notices the numbers moved.

02 · What it computes

Find the leak before it drains you.

Ward identifies abnormal inventory loss patterns and distinguishes between theft, damage, spoilage, and administrative error.

Ward compares expected inventory against actual counts, segments loss by cause category, and flags store-level anomalies against your estate baseline.

  • Cause-level shrinkage attribution
  • Store-vs-estate benchmarking
  • Receiving dock anomaly detection
  • Pattern recognition across time

Definitions resolve through the semantic layer, so a number here reconciles with the dashboard your team already publishes instead of disagreeing with it quietly. How the layer works →

Sample output

Store #37 showing 4.2% shrinkage vs 1.8% estate average. Pattern suggests receiving dock discrepancy, not shoplifting.

US retail shrinkage hit $112.1 billion in 2022, up 19.4% year over year. — National Retail Federation

03 · Scope and audit

What the agent may read, and what it did last Tuesday.

Agents on your own data fail the security review, not the demo. The model is never the problem. The problem is that nobody can say what the agent is allowed to reach, what it already did, or what happens when it is wrong.

01

The agent sees the Looker objects you declared on the source and nothing else. Scope is enforced at query time, not in the client.

02

Read-only until you grant a write action explicitly, per object. A grant leaves a record naming the agent, the caller and the rows it touched.

03

One audit log covers every person and every agent in the tenant, so a reviewer answers one question against one system rather than correlating two.

Governance and policy →
How a scope gets set

Describe the job in plain English. Ward AI proposes the agent and the exact permissions it would need against your Looker objects, and the proposal sits there until a person approves it. Nothing runs on a draft.

Every finding it returns carries the SQL that produced it. If the query is wrong you can see that it is wrong, which is the only way a data team ever trusts one of these.

04 · Where it runs

The same pipeline, in three operating contexts.

What shrinkage detection off Looker looks like once it is live. Retail is where Ward is deployed, not what Ward is.

Grocery

30,000+ SKUs · stores

One store flags elevated shrinkage well above the estate average for two consecutive periods. Traditional LP assumes shoplifting. Ward traces the majority of variance to receiving dock discrepancies in frozen foods, vendor deliveries consistently short against POs. One process fix, mandatory blind receiving, brings the store back in line within six weeks.

The grocery build →

Convenience

3,000+ SKUs · locations

Ward flags multiple locations with a consistent pattern: tobacco void rates spike during a specific overnight shift window. The amounts are small enough to evade threshold-based alerts but consistent enough to represent significant annual loss per store. Ward attributes the pattern to specific shift schedules, and investigation confirms scan avoidance by a ring of night-shift employees across the affected stores.

The convenience build →

Fashion

15,000+ SKUs · locations

Ward flags a cluster of stores where high-value item returns run well above estate average, most without original tags, with the same payment cards appearing across multiple locations. The pattern matches a wardrobing ring. LP adjusts the return policy for flagged categories and sees a significant drop in high-value returns within weeks.

The fashion build →
05 · Who consumes it

You stand it up. They read the output.

Your platform team owns the connection, the definitions and the policy. These are the teams that consume what comes out of it.

Head of LP

Loss Prevention

Cause-level attribution: theft, spoilage, or admin error. No more guessing.

What they see →

CFO

Finance

Shrink trend lines and dollar impact, attributed by category and store.

What they see →

Director Store Ops

Store Operations

Which stores are bleeding, and why, flagged before the next count.

What they see →
06 · What changes

What connecting Looker actually moves.

MeasureEffectWhy
Time to Insight Proactive, no login Explains why metrics moved before anyone checks a dashboard.
Anomaly Detection Inter-refresh coverage Catches deviations between Looker dashboard refresh cycles.
Decision Velocity Root cause attached Every anomaly card includes cause analysis; no drill-down needed.
Data Utilization Unused models activated LookML dimensions and measures queried beyond built dashboards.
Questions

What a technical evaluation asks first.

Ward can query Looker via API or connect directly to the underlying database. Either way, Ward monitors while your team browses.

Ward reads Looker API for query results, Underlying database (direct) and LookML model metadata. Credentials are read-only and scoped to the objects you name, so Ward cannot reach a table you did not grant it.

No. Ward queries Looker in place and holds no second copy of your data. Nothing is exported to run Shrinkage Detection, and your residency and retention policy are unchanged by connecting Ward.

Every connector carries a schema contract. When a source changes shape the contract fails loudly and the pipeline stops, rather than writing wrong rows into your warehouse for a week before anyone notices the numbers moved.

Ward identifies abnormal inventory loss patterns and distinguishes between theft, damage, spoilage, and administrative error. Ward compares expected inventory against actual counts, segments loss by cause category, and flags store-level anomalies against your estate baseline.

Yes. Every figure Ward reports carries the query that produced it, resolved through the semantic layer, so it reconciles with the dashboard your team already publishes rather than disagreeing with it quietly.

Only what you declared. An agent is scoped to named tables and cannot reach past them, it is read-only unless you grant a write action explicitly, and every query, tool call and result is logged against the agent identity that made it.

Typically Head of Loss Prevention, CFO / VP Finance and Director of Store Operations. They read the output; your platform team owns the connection, the definitions and the policy that produced it.

One store flags elevated shrinkage well above the estate average for two consecutive periods. Traditional LP assumes shoplifting. Ward traces the majority of variance to receiving dock discrepancies in frozen foods, vendor deliveries consistently short against POs. One process fix, mandatory blind receiving, brings the store back in line within six weeks.

Explore

Shrinkage Detection on another system.

SAP Oracle Shopify NetSuite Snowflake BigQuery Blue Yonder RELEX Power BI Tableau NCR BigCommerce Lightspeed Epicor

What else Ward computes off Looker.

Stockout Prediction Price Optimization Demand Forecasting Assortment Planning Promo Effectiveness Customer Behavior Fill Rate Monitoring
Looker connector → Shrinkage overview → All 540+ connectors → Every pairing →

Point Ward at your Looker and see what comes back.

Deploying is free. You pay 5% of the model compute it uses, and nothing else. No call required to find out whether this works on your data.

Get a demo → Take the 3-minute assessment

Read-only to start · your LLM keys · SOC 2 Type II underway · or book a call directly

Get started

Find out what your data has been hiding.

Tell us about your operation. We’ll show you the problems Ward catches, and the ones your current tools miss.

Step 1 of 3
What are your goals?
Step 2 of 3
About your operation
Step 3 of 3
Your contact info