Loading tutorials…
Loading tutorials…
Moby is Triple Whale's pitch for the AI era — natural-language queries against your DTC data. It's genuinely useful for specific tasks and overhyped for others. Here's the setup + the honest framework for when Moby beats opening the dashboard.
Who this is forTriple Whale users on Scale tier or above where Moby is included. Especially useful if you find yourself building ad-hoc queries weekly and would rather ask in plain English than build a custom dashboard widget.
What you'll need
Step 1
Moby is gated to Triple Whale Scale tier and above. Inspire and Grow do not include it.
Settings → Plan → confirm your tier is Scale or Enterprise.
In the Triple Whale sidebar, look for the "Moby" or chat-bubble icon. If it's greyed out or missing, your tier doesn't include Moby.
If you're on Grow and want Moby: contact Triple Whale sales for an upgrade quote. Don't upgrade ONLY for Moby — make sure you'll use the other Scale features (cohorts, multi-pixel attribution, creative analytics) too.
If Moby is included, click the icon to open the chat interface.
Step 2
Start with a "context-setting" conversation. Tell Moby what your store is, AOV, COGS, channels, season. This improves later answers.
Open Moby chat. First prompt: "I run a [category] DTC store on Shopify. Average order value is $[X]. Main channels are [Meta, Google, TikTok]. My fiscal year starts [month]. Margin target is [Y]%. Customer base is mostly [demographic]."
Moby will acknowledge. This context informs how Moby interprets your future questions (e.g., "is my ROAS good" → Moby will reference your margin target).
Pin this context conversation as a "system" message if Moby supports it (Settings → AI → Persistent Context). Otherwise, re-paste the context at the start of each session.
Step 3
Ask Moby 5 questions where you already know the answer (from dashboards). Verify Moby gets them right before trusting on harder questions.
Q1: "What was my total revenue last 7 days?" — verify against Shopify Admin → Reports.
Q2: "What was my blended ROAS last 30 days under TW Pixel attribution?" — verify against your Morning Coffee dashboard.
Q3: "Which channel drove the most revenue last 14 days?" — verify against Channel Review dashboard.
Q4: "What's my top 5 selling products by revenue last 30 days?" — verify against Product Performance dashboard.
Q5: "What's my new customer rate last 30 days?" — verify against Morning Coffee.
If all 5 match within 2%: Moby is grounded in your data correctly. Trust on harder questions.
If any are off by 5%+: Moby is hallucinating or pulling from wrong filters. Don't trust complex queries until calibration matches.
Step 4
Moby's sweet spot: one-off questions that would take 20 minutes to build as a dashboard widget. Ask in plain English.
Good Moby questions: "Which Meta ad sets had ROAS above 3x last 14 days but spent under $200?" / "Which products in my top 20 by revenue have the worst gross margin?" / "How did blended ROAS change after I launched the new TikTok campaign on May 12?"
Bad Moby questions: "Should I increase budget on Meta?" (too vague — Moby will give generic advice). "What's my best ad?" (best by what metric?). "Why is my ROAS declining?" (root-cause analysis Moby can't actually do without explicit hypothesis testing).
Pattern: specific, time-bounded, metric-defined questions get good answers. Vague strategic questions get vague answers.
Step 5
Moby can summarize your week into a paragraph. Useful for status updates to founders or board members.
Prompt: "Summarize my last 7 days of performance vs the prior 7 days. Include revenue, ROAS, top product, top channel. Format as a single paragraph for a weekly update to my CEO."
Moby will produce a clean paragraph with the key numbers and a one-line interpretation.
Useful for: weekly board updates, agency status emails, "what happened this week" Slack posts.
Always verify the numbers in the summary before sending — Moby can transpose digits or misreport a delta.
Step 6
If a prompt produces a useful answer, save it. Reusing the same prompt structure gets more consistent answers.
Moby → Saved Prompts (if available in your tier).
Save prompts like: "Weekly board summary," "Top 10 ads under-spending review," "Cohort vs last quarter comparison."
Re-run them weekly with date filters updated. Templates produce consistent output formats — easier to compare week-to-week.
Step 7
Document the questions Moby answers vs the questions you still need dashboards or analysts for. This is the most important step.
In your team's Triple Whale playbook (Notion, Slab, whatever), write down:
Moby is for: ad-hoc questions, weekly summaries, ad-set drilldowns, quick metric checks.
Moby is NOT for: budget allocation decisions (use dashboards), strategy questions (use a human), cohort interpretation (requires context Moby doesn't have), forecasting (Moby will guess and sound right).
Document this. Share with anyone else on your team using Triple Whale. Without explicit boundaries, Moby gets misused and team starts mistrusting all Triple Whale data.
Common mistakes
Trusting Moby without calibration
What goes wrong: You ask Moby a question, get a confident answer, act on it. The answer was wrong because Moby pulled from the wrong date range or misinterpreted your data. You make a budget decision on hallucinated numbers.
How to avoid: Always run the 5 calibration questions first. Re-calibrate monthly. If calibration fails, treat Moby outputs as suspect until investigated.
Using Moby for vague strategic questions
What goes wrong: You ask "should I scale Meta?" — Moby gives generic advice that sounds smart but is detached from your actual situation. You feel like you got an answer; you didn't.
How to avoid: Only use Moby for specific, time-bounded, metric-defined questions. Strategic questions need a human with context.
Treating Moby as a replacement for dashboards
What goes wrong: You stop building dashboards because 'Moby can just answer.' Then queries vary slightly week to week and you get inconsistent reporting. Team loses trust in the data.
How to avoid: Dashboards for recurring questions (what changes monthly: nothing). Moby for novel questions (this is new this week). Different tools for different jobs.
Not setting context at session start
What goes wrong: Moby answers without knowing your AOV, margin target, or season. Generic answers that don't reflect your business reality.
How to avoid: First message of every session: 2-3 sentence context. Better: configure persistent context in Moby settings.
Asking Moby to forecast or predict
What goes wrong: Moby will produce a confident forecast. Forecasts are wrong because Moby doesn't actually model your business — it pattern-matches on past data. You make Q4 commitments based on the forecast and miss them by 30%.
How to avoid: Use Moby for descriptive ("what happened") and diagnostic ("which segment caused that") questions. Don't use it for predictive ("what will happen") or prescriptive ("what should I do") questions.
Not verifying numbers in Moby outputs
What goes wrong: Moby reports "revenue up 14% week over week" — actual is 7%. You forward the summary to your board. Embarrassing correction later.
How to avoid: Always verify the key numbers in any Moby summary against the underlying dashboard before sharing externally.
Recap
Done — what's next
How to set up Triple Whale for your Shopify store
Read the next tutorial
Hand it off
Moby is a tool. Like any tool, value depends on the operator. Most DTC owners who try Moby for two weeks either over-trust it (bad decisions) or give up (wasted potential). A vetted DTC specialist who uses Moby weekly knows the prompt patterns that work and the questions to bypass it on — typically $400-800/mo for ongoing weekly review that leverages Moby + dashboards together.
See specialist rates
Only if you'll also use the other Scale features (cohorts, multi-pixel attribution, creative analytics). Moby alone is not worth the $300-400/mo tier jump from Grow. But Scale tier as a whole is justified at $500K+/mo Shopify revenue.
Sometimes. LLMs can transpose digits, misinterpret date ranges, or pull from the wrong filter. This is why calibration questions matter. Don't trust any Moby number you haven't verified against a dashboard or report.
Moby's interface is conversational — it doesn't produce SQL or widget configs you can save. For programmatic queries, use Triple Whale's API or build widgets manually. Moby is for one-off conversational analysis.
Moby is connected to YOUR Triple Whale data. ChatGPT is not. Moby can answer questions about your specific revenue, customers, and channels. ChatGPT can only answer generic DTC questions. Different jobs.
No. Moby only has access to your own Triple Whale data. For competitor benchmarks, use industry reports or Similarweb-style external data.
Triple Whale ships Moby model improvements quarterly-ish. The big jumps usually come with major LLM provider updates (e.g., GPT-4 → GPT-5). Your trust calibration should re-run monthly to confirm Moby still grounds correctly.
Triple Whale
Most founders sign up for Triple Whale, click 'Connect Shopify,' and assume the rest will figure itself out. Three weeks later they're paying $200/mo for a dashboard nobody opens because attribution looks weird. Here's the setup that actually delivers value from day one.
Triple Whale
Triple Whale ships with 40+ default widgets and most owners build dashboards with all of them — then never open the dashboard again. Here's how specialists build dashboards that get read every morning.
Triple Whale
Blended ROAS tells you which channels drove revenue last month. Cohort analysis tells you which channels drove customers who are still spending today. That's the difference between scaling profitably and scaling broke.
Triple Whale
Triple Whale's attribution model picker is buried in Settings and most users never touch it. The default is fine for some stores and dangerously wrong for others. Here's the decision framework that drives 80% of the value you get out of the platform.
Triple Whale
DIY DTC attribution is a great idea — until it isn't. This is the honest framework: when the cost of self-managing Triple Whale (or any attribution platform) exceeds the cost of hiring, and how to tell which side you're on.