Crux
Book a call

Workflow library

The questions your AI analyst answers for ecommerce teams, grouped by the job they do.

Start here

What can I ask?

What can I ask you about our business?

Data needed

  • Every table your brain publishes

Expected output

  • A tour of what your brain holds, grouped by area, with example questions for each

Diagnose performance

The master workflow. When a number moves, find where the change is, which lever explains it, the main driver within that lever, and what to do about it.

Diagnose performance

Analyse the change in contribution margin after ad spend (CM3) last month against the month before. Break it into three levers: new customers (how many, and at what CAC), repeat (how many orders came from returning customers) and average order value. Which lever explains most of the change? Within it, follow the drivers down, weighting by volume, until one stands out: a channel, a product or a discount. Finish with the one thing we should do about it this week.

Data needed

  • channel_reporting
  • orders

Expected output

  • How much CM3 moved last month, and which of the three levers moved it
  • The driver within that lever: the channel, product or discount behind most of the change
  • One action to take this week

Contribution margin, the short version

Analyse the change in contribution margin last month. Tell me the core driver of the change across these levers: new customers, repeat and average order value. What was the main driver within that lever and what should we do about it?

Data needed

  • channel_reporting
  • orders

Expected output

  • How much contribution margin moved, and which lever explains it
  • The main driver within that lever
  • A recommendation on what to do

Pacing against budget

How are we pacing against budget this month? Show ad spend, net sales and new customers against plan to date, and where we will land at month end at the current run rate. Which line is furthest off plan, and what would close the gap?

Data needed

  • daily_budget_versus_actuals

Expected output

  • Ad spend, net sales and new customers against plan, month to date
  • Where each will land at month end at the current run rate
  • The line furthest off plan, and what would close the gap

Customers

How many new customers you acquire, what each one costs, and how well your traffic converts.

Acquisition leakage audit

Audit our acquisition campaigns over the last 8 weeks. For each campaign with spend, what share of its orders came from existing customers, and how much spend does that represent? Show brand search as its own category with the same analysis: it is mostly bought to defend our name against competitors, and a high existing-customer share there says much of that spend may not be bringing in anyone new. Then show which campaigns have a CAC well above their channel's average and what is driving it. Finish with the audience exclusions and budget changes to make, brand search included.

Data needed

  • channel_reporting, by campaign

Expected output

  • Each acquisition campaign with the share of its orders from existing customers and the spend that represents
  • Brand search as its own category: how much of it reaches existing customers, and how much may be cut
  • The campaigns whose CAC sits well above their channel's average, and why
  • The audience exclusions and budget changes to make

Repeat share by acquisition campaign

For each acquisition campaign last quarter, what share of its orders came from repeat customers? Rank them, highest first.

Data needed

  • channel_reporting

Expected output

  • Acquisition campaigns ranked by the share of their orders from repeat customers

Highest cost per new customer

Which campaigns have the highest cost per new customer, and what is driving it?

Data needed

  • channel_reporting

Expected output

  • Campaigns ranked by cost per new customer
  • What is driving the cost on the most expensive

New customer channel mix

How has our new customer channel mix evolved over the last 6 months?

Data needed

  • channel_reporting

Expected output

  • New customers by channel, month by month, for 6 months
  • The channels gaining share and the ones losing it

New customers and CAC by month

Show new customers, ad spend and CAC by month for the last 12 months, overall and by channel. Where has CAC moved most, and is that because spend changed or because new customers did?

Data needed

  • channel_reporting

Expected output

  • A monthly table of new customers, ad spend and CAC, overall and by channel
  • The channel whose CAC moved most
  • Whether a change in spend or in new customers caused it

Landing pages that leak

Which landing pages had the most sessions over the last 30 days, and how does each convert against our site average? Show the busy pages that convert worst first, and the step in the funnel where each one loses people. Which three should we fix first?

Data needed

  • ecommerce_funnel

Expected output

  • Your busiest landing pages, each with its conversion rate against the site average
  • The funnel step where each one loses people
  • The three pages to fix first

Which Meta ads deserve a closer look

Over the last 30 days, which Meta ads spent the most, and how does each one's CAC compare with the account? Flag ads whose CAC is getting worse as spend rises, and ads with a strong CAC on little spend that deserve more budget.

Data needed

  • meta_performance

Expected output

  • Your top-spending Meta ads, each with its CAC against the account
  • Ads whose CAC is getting worse as spend rises
  • Ads with a strong CAC that deserve more budget

Google spend that wins no customers

Which Google Ads campaigns and ad groups spent the most over the last 30 days with few or no conversions? Rank them by spend, show what share of our Google spend they take, and tell me what to pause or cut.

Data needed

  • google_ads_performance

Expected output

  • Campaigns and ad groups ranked by spend, where conversions are few or none
  • The share of your Google spend they take
  • What to pause or cut

LTV to CAC by acquisition channel

For customers we acquired in the last 12 months, compare 6-month LTV with CAC by acquisition channel. Which channels pay back, which do not, and where should budget move?

Data needed

  • ltv_retention

Expected output

  • 6-month LTV, CAC and the ratio between them, by acquisition channel
  • The channels that pay back and the ones that do not
  • Where to move budget

Frequency

How often customers come back, read in cohorts: groups of customers acquired in the same month.

Best cohorts

Compare the customers we acquired over the last 18 months by how they came to us: acquisition channel, first product bought and discount code used. For each, show 6-month LTV and the share who bought again within 6 months. Which cohorts are best, what do they have in common, and what 3 to 5 learnings should shape how we market to customers?

Data needed

  • ltv_retention
  • customers
  • order_items

Expected output

  • Your cohorts by acquisition channel, first product and discount code, each with 6-month LTV and repeat rate
  • What the best cohorts have in common
  • 3 to 5 learnings, in priority order, for how you market

Cohort LTV and retention

Analyse our customer cohorts over the last 12 months. Which ones had the highest LTV and retention rates? What was the driver for this and what can we take as a learning into how we market to customers going forward?

Data needed

  • ltv_retention
  • customers

Expected output

  • Your cohorts ranked on LTV and retention
  • What drove the best ones
  • Learnings to carry into your marketing

Repurchase timeframe

What's our typical repurchase timeframe, and how does it differ by acquisition product? How has it changed over the last 6 months?

Data needed

  • customers
  • orders
  • order_items

Expected output

  • The typical gap between first and second order, overall and by first product
  • How that gap has changed over 6 months

Lapsed customers worth winning back

Which customers have bought from us at least twice but not in the last 6 months? Group them by how much they have spent and what they first bought, and tell me which groups are worth a win-back campaign.

Data needed

  • customer_profile
  • customers

Expected output

  • Lapsed repeat customers, grouped by what they spent and what they first bought
  • How many customers sit in each group
  • The groups worth a win-back campaign

Email campaigns and flows that earn

Over the last 90 days, which Klaviyo flows and campaigns earned the most revenue per recipient? Show open and click rates beside it, and flag any flow that is earning less than in the 90 days before.

Data needed

  • klaviyo_* tables

Expected output

  • Flows and campaigns ranked by revenue per recipient, with open and click rates
  • Any flow earning less than in the 90 days before

Order Value

What each order is worth and what it earns, driven by product mix and discounting.

Discounting

How has our discounting changed over the last 12 months? By month, show the share of orders discounted, the average discount depth and the total given away. Then compare each discount code group with orders that used no code on order value and CM2 per order, and flag codes used mostly by customers who had bought before. Give me the 3 biggest opportunities to discount less without losing orders.

Data needed

  • orders

Expected output

  • Month by month: the share of orders discounted, the average discount depth and the total given away
  • Each discount code group against no-code orders, on order value and CM2 per order
  • The codes used mostly by returning customers
  • The 3 biggest chances to discount less without losing orders

Discounting strategy, the short version

How has our discounting strategy changed over the last 12 months? What's it costing us and what are some opportunities to optimise it?

Data needed

  • orders

Expected output

  • How your discounting has changed over 12 months
  • What it costs you
  • Ways to optimise it

What drives our AOV

What has been the key driver for our AOV over the last 3 months? Compare discounting, order size and product mix.

Data needed

  • orders
  • order_items

Expected output

  • How much AOV moved over 3 months, split into discounting, order size and product mix
  • Which of the three drove it

Return rate by product

What is our return rate by product over the last 6 months, by units and by value? Show the products that sell in volume and are returned most, and what those returns cost us.

Data needed

  • order_items
  • order_item_returns

Expected output

  • Return rate by product, by units and by value
  • The products that sell in volume and are returned most
  • What those returns cost you

Products that convert once viewed

Over the last 30 days, which products get plenty of views but convert to add-to-cart and purchase below our average? And which convert well but get few views? Tell me what to do with each group.

Data needed

  • product_funnel

Expected output

  • Products with plenty of views that convert below your average
  • Products that convert well but get few views
  • What to do with each group

Stock about to run out

Needs merchandising data

Which variants will run out of stock in the next 4 weeks at the current rate of sale? Rank them by the sales we would lose, and tell me which to reorder first.

Data needed

  • variant_inventory_daily

Expected output

  • Variants that will run out within 4 weeks at the current rate of sale, ranked by the sales at risk
  • Which to reorder first

This prompt needs merchandising data. Ask us about adding it →

Sell-through and markdown by style

Needs merchandising data

For this season's styles, show sell-through to date and the share of units sold at a markdown. Which styles are selling through fastest at full price, and which need a markdown decision now?

Data needed

  • order_items, merchandising columns
  • dim_product_variant

Expected output

  • Sell-through to date and the share of units sold at a markdown, by style
  • The styles selling fastest at full price
  • The styles that need a markdown decision now

This prompt needs merchandising data. Ask us about adding it →

Context layer

Your brain reads what customers buy. With the context layer it also reads what they say: product reviews, Meta ad comments, post-purchase and feedback surveys, competitor reviews and your own meeting notes, held beside the revenue they relate to. These prompts need it installed.

Review themes into actions

Needs the context layer

Pull out the main themes from our product reviews in the last 30 days. Sort them into marketing actions and product improvements, with how often each comes up and a customer quote for each.

Data needed

  • Product reviews

Expected output

  • The main themes in recent reviews, ranked by how often they come up, each with a customer quote
  • Marketing actions
  • Product improvements

This prompt needs the context layer. Ask us about adding it →

Ad hooks from reviews and comments

Needs the context layer

Analyse our product reviews and Meta ad comments from the last 30 days. Filter for positive sentiment and give me 10 hook ideas for Facebook ads, each in our customers' own words, with the review or comment it came from.

Data needed

  • Product reviews
  • Meta ad comments

Expected output

  • 10 Facebook ad hooks in your customers' own words
  • The review or comment each one came from

This prompt needs the context layer. Ask us about adding it →

Objections in ad comments

Needs the context layer

What objections and questions come up most in our Meta ad comments over the last 30 days? For each, suggest the answer to add to the ad copy, the landing page or the FAQ.

Data needed

  • Meta ad comments

Expected output

  • The objections and questions that come up most, ranked
  • The answer to add to the ad, the landing page or the FAQ for each

This prompt needs the context layer. Ask us about adding it →

What customers say brought them

Needs the context layer

Over the last 90 days, compare what customers told us in the post-purchase survey about how they heard of us with the channel we attributed their order to. Where do the two disagree most, and which channels are we under-crediting?

Data needed

  • Post-purchase survey
  • orders

Expected output

  • Survey answers set against the channel each order was attributed to
  • Where the two disagree most
  • The channels you are under-crediting

This prompt needs the context layer. Ask us about adding it →

Complaints beside returns

Needs the context layer

Which products had the most negative reviews in the last 90 days, and what do the complaints say? Set them beside each product's return rate and tell me which fixes would save us the most.

Data needed

  • Product reviews
  • order_items
  • order_item_returns

Expected output

  • The products with the most negative reviews, and what the complaints say
  • Each one's return rate beside it
  • The fixes that would save the most

This prompt needs the context layer. Ask us about adding it →

Where competitors fall short

Needs the context layer

What do customers praise and complain about in our competitors' reviews? Where are they weak and we are strong, and how should that change our messaging?

Data needed

  • Competitor reviews

Expected output

  • What customers praise and complain about in competitors' reviews
  • Where they are weak and you are strong
  • How to change your messaging

This prompt needs the context layer. Ask us about adding it →

Decisions against the numbers

Needs the context layer

Summarise the decisions and actions from our meetings in the last 30 days. For each, tell me whether it has been done, and what the numbers it was meant to move have done since.

Data needed

  • Meeting notes
  • channel_reporting

Expected output

  • The decisions and actions from the last 30 days of meetings
  • Whether each has been done
  • What the numbers each was meant to move have done since

This prompt needs the context layer. Ask us about adding it →

Make it a habit

Set this up as a scheduled report in your own Claude, pointed at your Growth Brain, so the report is waiting for you every Monday. Add the questions you want answered every week.

Weekly performance report

Write our weekly performance report for last week (Monday to Sunday). Compare it with the week before and the same week last year: net sales, orders, new customers, ad spend, MER, CAC and AOV, each with the % change. Then tell me the biggest change, which lever it sits in (new customers, repeat or order value) and the one thing to look at this week. Keep it to one screen.

Data needed

  • channel_reporting

Expected output

  • A one-screen report on last week: net sales, orders, new customers, ad spend, MER, CAC and AOV, against the week before and the same week last year
  • The biggest change, and the lever it sits in
  • The one thing to look at this week