DANI · AI Data Assistant

An analyst,
not a chatbot.

DANI is MFour's AI data assistant inside Studio. Ask in plain English and it analyzes your connected behavioral and survey data — consumer journeys, cross-shopping, audiences, competitive — answered in seconds. No data wrangling, no code.

Ask DANI See how DANI thinks›

Consumer JourneyCross ShoppingLeakageLLM Convo Signals

Illustrative example · not real client data

What share of our buyers also shop at Costco?

37.4% of your buyers also shopped Costco in the last 90 days.

2.1M

Costco shoppers

37.4%

of your buyers

Weighted to
U.S. population

Moderate integrityFidelity 0.71/ 1.00

Low <0.50Moderate 0.50–0.79High ≥0.80

How I calculated thisobserved n = 2,140

Behavioral depth25%

Temporal persistence20%

Cross-modal agreement20%

Panel coverage20%

External validation15%

Census-balanced weightingAbout the Fidelity Score ›

A moderate score means DANI is flagging thinner data — protecting you from over-indexing, not hiding a weak answer.

Ask DANI about consumer journeys, receipts, attitudes…

The connected consumer journey

One consumer. One journey. Every step connected.

DANI follows what a real consumer actually does — searches, researches, visits, and buys — and connects every moment back to a single identity.

Step 01 · 9:14 AM

Searches for a solution

Asks ChatGPT: “best running shoes for flat feet?”

Captured · ChatGPT

Step 02 · that evening

Researches the options

Compares 12 review sites and 3 brand apps

Captured · App & Web

Step 03 · Saturday

Visits the store

Walks into 2 retailers — GPS-verified

Captured · Location

Step 04 · checkout

Makes the purchase

Buys — $142 basket, item-level receipt

Captured · Purchase

DANI · one identity

Every behavior tied to one ID — so search, browse, visit, and buy all belong to the same person.

Search · research · visit · purchase — connected to one identity, reasoned over by one analyst

Not just answers. DANI builds the whole briefing.

Ask a quick question and get an answer in seconds... or ask DANI for a report and it works autonomously in the background — for hours — to deliver a complete, multi-section, boardroom-ready report.

Step 1 · DANI builds it

One brief. An entire boardroom report.

Describe the report you want. DANI runs on its own.

"Build me a full cross-shopping report on our category — every competitor, every leakage path, ready to present."

DANI · working autonomously

~ hours

Querying every connected stream

Cross-referencing receipts, visits & surveys

Drafting 14 sections + charts

→ delivereda 14-section report, boardroom-ready

6 report templates · pick one & brief it

Consumer Journey

Behavioral path-to-purchase

12 sec

Cross-Shopping

Category share drain & leakage to competitors

14 sec

Audience Profiling

First-party audience intelligence & personas

27 sec

Customer Leakage

Where lost visits go & how to win them back

10 sec

LLM Conversation Insights

How AI reshapes discovery & decisions

15 sec

Competitive Head-to-Head

Competitive market intelligence & scorecards

12 sec

delivers

Step 2 · The artifact

A boardroom-ready report.

DANI · Executive Briefing12 sections

Consumer Journey — Kroger

Behavioral path-to-purchase from first-party panel data

Top-line verdict

Kroger owns the routine trip — but hyperlocal loyalty masks a fragmented wallet and a hidden cliff-lapse risk.

Table of contents

01Executive Summary & Top-Line Verdict
02The 5-Stage Shopper Journey
03Audience Tiering (Heavy / Medium / Light)
04Cross-Shop Dynamics
05Same-Day Mission Analysis
06Loyalty Segmentation & Lapse Risk
＋6 more sections

Charts & scorecards

Fidelity Score on every insight

0.91

Real templates, real sections — generated by DANI, not a slide team.

Built for trust

DANI gives you an answer you can defend.

Access to all behavior data

DANI reasons over MFour's full behavioral signal — location, app & web, purchase, and ChatGPT search — connected to survey responses on a single identity.

Explore the data streams ›

Weighted to the US population

Every result is census-balanced and weighted to the US population, so what you see is representative — not just raw panel counts.

A Fidelity Score on every insight

DANI attaches a Fidelity Score — a 0–1 integrity band built from five weighted components — to each number, so you know exactly how much to trust it before you act.

Ask anything

Ask in plain English. DANI writes the SQL — and runs the analysis.

Ask any question in plain English. DANI writes the query and runs the analysis across every connected stream in OmniTraffic® — no analysts, no ticket queue.

You ask — in plain English

"Find consumers who asked ChatGPT about running shoes, walked into a footwear store within 7 days, then bought a competitor brand — and cluster them into personas."

Ask the way you'd ask a colleague — DANI knows where every signal lives.

DANI · generated query

RAG → SQL

SELECT p.panelist_id, p.age, p.dma_name

FROM chatgpt_convos c

JOIN brick_mortar_visits v USING (panelist_id)

JOIN purchases pu USING (panelist_id)

WHERE c.topic ILIKE '%running shoes%'

AND v.category = 'Footwear'

AND pu.brand <> 'Nike'

AND v.visited_at BETWEEN c.occurred_at AND c.occurred_at + INTERVAL '7 days';

-- DANI then clusters the cohort (k-means) into 3 personas

→ result1,284 consumers · clustered into 3 behavioral personas

Any analysis, on demand

Clustering & segmentation

Group consumers into behavioral personas automatically.

Pattern & anomaly detection

Surface unusual shifts, spikes, and outliers in behavior.

Cohort & retention

Track defined groups across time and channels.

Trend & forecasting

Project where a metric or behavior is heading.

Correlation & drivers

Find which signals actually move the outcome.

Funnel & path analysis

Map the real sequence from signal to conversion.

Market-basket & affinity

See what consumers buy, visit, and use together.

Significance testing

Confidence and base sizes attached to every cut.

New · ChatGPT dataset

Your customers are making buying decisions with AI.

See every opt-in ChatGPT conversation — the brands AI names, joined to demographics on PANELIST_ID, and exactly who’s asking.

Explore the ChatGPT data

ChatGPT

Captured

Best running shoes for flat feet?For overpronation, I’d go with the Brooks Adrenaline GTS — the stability keeps flat arches aligned.

Go deeper

Understand the data behind the answer.

First-party, 100% opt-in behavior — LLMs, location, app, web, media, purchase, and social.

F · 34 · TOLEDO DMA · 11 datasets

Every signal, one identityExplore

F · 34 · TOLEDO DMA · 11 datasets

Explore the data streams

DANI for Survey Analysis

Kill the crosstabs. Go from days to minutes.

DANI has access to every survey you've ever run with MFour. Ask a question in plain English and get the full analysis — tables, charts, key takeaways, exec summary — no manual cross-tabs or pivot tables.

The old way

Manual crosstabs

Row	W1	W2	W3	W4	W5	A	B	C	D	E
18-24	38.2	51.3	64.4	77.5	30.6	43.7	56.8	69.9	22.0	35.1
25-34	45.3	58.4	71.5	24.6	37.7	50.8	63.9	76.0	29.1	42.2
35-44	52.4	65.5	18.6	31.7	44.8	57.9	70.0	23.1	36.2	49.3
45-54	59.5	72.6	25.7	38.8	51.9	64.0	77.1	30.2	43.3	56.4
55+	66.6	19.7	32.8	45.9	58.0	71.1	24.2	37.3	50.4	63.5
Male	73.7	26.8	39.9	52.0	65.1	18.2	31.3	44.4	57.5	70.6
Female	20.8	33.9	46.0	59.1	72.2	25.3	38.4	51.5	64.6	77.7
NE	27.9	40.0	53.1	66.2	19.3	32.4	45.5	58.6	71.7	24.8
MW	34.0	47.1	60.2	73.3	26.4	39.5	52.6	65.7	18.8	31.9
South	41.1	54.2	67.3	20.4	33.5	46.6	59.7	72.8	25.9	38.0
West	48.2	61.3	74.4	27.5	40.6	53.7	66.8	19.9	32.0	45.1
Urban	55.3	68.4	21.5	34.6	47.7	60.8	73.9	26.0	39.1	52.2
Total	62.4	75.5	28.6	41.7	54.8	67.9	20.0	33.1	46.2	59.3

3–4days

Hand-built cross-tabs, pivot tables, and a dashboard rebuild — per wave. So historical trending effectively never happened.

With DANI

One prompt

"Our Q2 brand tracker just closed — give me the overall top-line results."

Maya · Consumer & Shopper Insights

DANI · Q2 brand tracker (n = 1,012)

Aided awareness71%

Consideration54%

Preference38%

Past-30d purchase29%

Key Takeaway: Awareness is healthy at 71%, but the drop to 29% past-30-day purchase points to a conversion gap, not an awareness gap.

Export to deckAdd to report

1minute

Developer API

Wire DANI
Into Everything

Developers can access DANI™ programmatically via HTTP API or MCP server. Plug DANI into your AI agent stack and query millions of online and offline behavioral events happening daily at scale.

API — Query DANI for insights on connected consumer journeys thru simple HTTP API calls

MCP Server — Drop DANI into any AI agent workflow as a first-class MCP tool that does more than fetch data but responds with analysis and insights

Insights at Scale — From ad-hoc queries to production pipelines — same API, any volume

POSTapi.mfour.com/v1/dani/query

{
  "query": "Top eCommerce apps used within 15 min
          of arriving at Costco"
}

Response

{ "response": { "answer": "Top eCommerce apps near Costco...", "charts": [...],
  "tables": [...], "citations": [...] }, "metadata": { "tokens": 181259 } }

Full example

# Query consumer behavior signals
curl -X POST https://api.mfour.com/v1/dani/query \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "query": "Top eCommerce apps used within
              15 min of arriving at Costco",
    "filters": {
      "demographic": {
        "age": "25-34",
        "gender": "female"
      },
      "timeframe": "last_30_days"
    }
  }'

# Response
{
  "response": {
    "answer": "Top eCommerce apps used within 15 min...",
    "charts": [
      { "type": "bar", "title": "Top Apps by Sessions", "spec": "..." }
    ],
    "tables": [
      { "title": "App Breakdown", "columns": ["App", "Sessions"], "data": [...] }
    ],
    "citations": [ { "id": "cs_78e4c67b...", "source": "Behavioral Data" } ],
    "data_context": {
      "sample_size": 15420, "coverage_multiplier": "2.3x",
      "modalities_used": ["app_usage", "location"]
    }
  },
  "metadata": {
    "tokens": 181259,
    "tools_used": ["Survey_Data", "data_to_chart"],
    "query_id": "01bd7890-0406-d123-0022-b787aa5678ef"
  }
}

An analyst,not a chatbot.