CATEGORY

FP&A Strategy

WRITTEN BY

Fin

Your AI CFO

FP&A Strategy

How CPG CFOs Should Vet FP&A Tools to Replace Excel

How CPG CFOs Should Vet FP&A Tools to Replace Excel

FP&A Strategy

10 minutes

WRITTEN BY

Fin

Your AI CFO

SHARE BUTTONS

WRITTEN BY

Fin

Your AI CFO

SHARE BUTTONS

Your Shopify revenue, Amazon settlements, TikTok fees, and ad spend are probably living in separate tabs right now, and someone on your team is manually pulling them into one P&L every month. That's a data engineering problem, and a BI dashboard won't fix it. Here's how to think through which FP&A tools are actually built for the way omnichannel CPG brands run their finances.

TLDR:

  • A 2024 study found 94% of business spreadsheets contain errors; BI tools only visualize that bad data, they don't fix it

  • Horizontal FP&A tools built for SaaS have no native concept of Amazon settlement fees or Shopify payouts

  • Require channel-level contribution margin by channel, not blended gross margin, before allocating budget

  • A profitable quarter and a cash crisis can happen simultaneously; your FP&A tool must show both

  • Iris Finance takes the consolidation work off your team: it normalizes and matches Shopify, Amazon, and TikTok order-level data in a proprietary warehouse before any analysis runs

Why Excel and BI Dashboards Eventually Break for CPG Finance Teams

Excel works until it doesn't. For CPG finance teams, the break usually happens around the third or fourth channel add: Amazon, TikTok Shop, a wholesale account, a 3PL. The spreadsheet that once took an afternoon now takes a week to close.

The failure modes are specific. Version control breaks first: someone updates the COGS tab, someone else is working off last Tuesday's export, and by Friday you have three "current" models that disagree. Manual data consolidation follows. Pulling Shopify payouts, Amazon settlements, ad spend, and landed cost updates into one P&L is a data engineering problem your analyst is solving by hand every month.

A 2024 Frontiers of Computer Science study found that 94% of business spreadsheets contain errors. The study is a literature review covering roughly 35 years of spreadsheet quality research.

BI dashboards add visualization but don't fix the underlying data problem. A Tableau or Looker dashboard connected to a manually maintained warehouse shows a clean chart built on stale inputs. The contribution margin line is probably wrong, and you won't know it until two weeks after the decision already got made.

FP&A Platform vs. BI Dashboard: What Each One Actually Does

BI tools like Tableau, Looker, and Power BI are visualization layers. They read from a data source and render charts. What they cannot do is write back to a planning model, lock a period for audit purposes, or run a scenario that changes your forecast when you adjust an assumption. They show you what happened. The analysis stops there.

FP&A tools are planning infrastructure. The distinction comes down to whether a tool supports writeback, approvals, and locked periods versus one that relies on manual discipline. An FP&A tool connects your actuals to a live model, runs scenarios against it, tracks plan versus actual on a rolling basis, and maintains an audit trail your accountant can rely on.

For a CPG CFO, the difference shows up at month-end. A BI dashboard tells you revenue was down 12%. An FP&A tool tells you which channel drove it, whether you were on pace to catch it mid-month, and what the downstream cash impact looks like given your current inventory position.

Buying a BI tool to solve a planning problem is a category mismatch. You get a better-looking version of the same gap.

The Three Categories of FP&A Platforms a CPG CFO Will Encounter

Three distinct categories exist, and picking the wrong one means rebuilding six months from now.

Excel-native tools (Vena, Datarails, Cube) keep your analysts working in spreadsheets but add a governed database behind them. Version control improves and consolidation gets faster. The tradeoff: you still need clean data flowing into Excel, and someone still has to maintain those models. For CPG brands adding channels, the input problem stays.

Cloud-native financial modeling tools (Anaplan, Pigment, Workday Adaptive Planning) replace Excel with a dedicated modeling environment. More structure, better scenario handling, stronger audit trails. Implementation tends to take longer and the learning curve is steeper. These are horizontal tools built for general business planning, so they have no out-of-the-box data model for contribution margin by channel or Amazon cohort LTV. A CPG brand asking them those questions is asking the wrong question of the right category.

Vertical-specific FP&A tools are built around the actual data models of a particular industry. For consumer brands, that means order-level revenue reconciliation across Shopify, Amazon, and TikTok; channel-level contribution margin; cohort profitability; and a financial model connected to live transaction data.

Category

Examples

How It Works

Key Tradeoff for CPG

Excel-native

Vena, Datarails, Cube

Adds a governed database behind your existing spreadsheets; analysts stay in Excel

Version control improves, but input data quality problems remain; someone still consolidates channels by hand

Cloud-native horizontal

Anaplan, Pigment, Workday Adaptive Planning

Replaces Excel with a dedicated modeling environment; stronger audit trails and scenario handling

Built for general business planning (headcount, ARR-style modeling); no out-of-the-box data model for Shopify payouts, Amazon settlement fees, or subscription cohorts

Vertical CPG-specific

Iris Finance

Order-level revenue reconciliation across Shopify, Amazon, and TikTok; channel-level contribution margin built into the data model

Purpose-built data model for omnichannel CPG; less suited to a brand with one simple channel and few SKUs

FP&A Platform Categories: What CPG CFOs Need to Know

Which Category Fits Your Team

  • If your finance team lives in Excel and your business is relatively simple, an Excel-native tool extends your runway without forcing a workflow overhaul. A brand under $10M on one or two channels with a small finance team will get version control and faster consolidation without a wholesale workflow change.

  • If you are running three or more channels with subscription complexity and want real-time margin visibility before your books close, you need a vertical tool built around CPG data models. Once you pass $10M across DTC, Amazon, and TikTok Shop, the data reconciliation problem is more than a horizontal tool can handle without custom engineering.

  • Cloud-native horizontal tools sit in the middle and serve neither extreme well for CPG. They add modeling structure, but the moment you need channel-level contribution margin or Amazon settlement reconciliation, you're building workarounds that belong in the data layer, not the planning layer.

Why Horizontal FP&A Tools Often Underserve Omnichannel CPG Brands

Horizontal FP&A tools fail CPG brands at the data layer, not the modeling layer.

Anaplan, Pigment, Abacum, Drivetrain, and Mosaic are horizontal planning tools designed for general business planning, and SaaS-style metrics like burn rate, headcount cost, and ARR are common in their use cases. Out of the box, they have no data model for a Shopify payout, an Amazon settlement fee, or a Recharge subscription cohort, so a CPG brand has to build that layer itself. Those aren't minor gaps. They are the entire job.

The connector problem surfaces first. "Integration" often means a general API pull, not order-level reconciliation. Knowing your Shopify revenue is different from knowing delivered gross profit after merchant fees, discounts, refunds, and shipping overruns. A tool without that distinction hands you a gross sales number and calls it revenue. For an omnichannel CPG brand, that number is wrong before you've run a single scenario.

Contribution margin by channel is where the mismatch becomes unworkable. A SaaS-focused tool has no built-in concept of COGS per SKU, 3PL cost per order, or ad spend allocated against new customer acquisition by channel. You can build custom formulas to approximate it, but that's data engineering inside a planning tool, which defeats the purpose of buying one.

The result is a CFO who bought something that works in demos and breaks in production, usually around month two when the first real variance analysis needs to trace back to an order.

Multi-Channel Financial Planning in a Single View: What to Require

When you are running DTC, Amazon, wholesale, and TikTok simultaneously, the financial question is never "how much did we sell?" It's "which channel made money after every cost is accounted for?" Most tools only answer the first one.

A unified multi-channel view should include channel-level contribution margin, not blended gross margin. Blended numbers hide splits like this hypothetical one: Amazon at 8% contribution margin while your DTC subscription base sits at 34%. You need that split before you allocate budget.

There are a few things to require here:

  • SKU-level profitability has to sit inside the same view. A channel can look healthy while high-velocity SKUs quietly drag margin down through return rates, bundle discounting, or fulfillment costs. Revenue by SKU without profitability by SKU leaves inventory decisions unanswered.

  • Plan versus actual by channel, tracked daily, closes the gap before damage compounds. If TikTok contribution margin falls off pace in week two, you want that flagged then, not on day 16 when the books open.

  • Ad spend has to connect to delivered gross profit, not gross revenue. A tool showing Meta spend against top-line Shopify sales is putting two unrelated numbers next to each other. You need ad spend allocated against new customer acquisition by channel, net of fees, refunds, and COGS, so your CAC reflects what a customer actually cost to acquire and serve.

How to Assess Integration Depth, Not Integration Count

Connector count tells you how many pipes exist. Integration depth tells you what flows through them. Only one of those determines whether your financial data is actually accurate.

Questions worth asking before you accept a demo as proof of capability:

  • Does your Shopify integration pull order-level data or just revenue totals? Order-level means you can trace any P&L line back to an individual transaction. Revenue totals give you a number with no audit trail.

  • Does your Amazon connector match settlement fees, merchant fees, and returns against each order, or does it show gross POs and stop there?

  • How does the tool handle channels with no direct API, like distributor depletion reports from UNFI or KeHE? Manual upload is a legitimate answer. No answer is a data gap you will own every month.

  • Does the TikTok connection capture order-level fees, including FBT fees and affiliate commissions per transaction, or just ad spend totals from the marketing API?

That last point matters. TikTok publishes ad spend through its marketing API, which every tool can read. What almost no tool has is the commerce-side order data with every fee per transaction. If your TikTok Shop is growing and your connector only reads ad spend, your contribution margin is wrong.

Ask vendors to show you, in a live environment, how a return posted 40 days after the original order gets handled. If the answer involves a manual journal entry, that is a maintenance burden you are inheriting indefinitely.

The Data Architecture Question: Layer Above Your Systems or Replacement Warehouse

Every FP&A evaluation eventually hits this question: does the tool sit on top of your existing systems, or does it replace the warehouse underneath them?

A reporting layer reads from whatever you already have. Fast to connect, faster to implement. The problem is that it inherits every data quality issue upstream. If your Shopify export has duplicate orders, your Amazon settlement reconciliation is off, or your ad platform reports spend on different date logic than your ERP books revenue, the reporting layer reflects all of it without correction. Garbage in, cleaner chart out.

A purpose-built warehouse is different. The vendor owns the transformation logic: how a Shopify payout becomes recognized revenue, how an Amazon return 40 days post-order gets allocated, how TikTok fees get stripped out at the order level before margin is calculated. When that logic lives inside the tool, accuracy is the vendor's problem, not yours. That matters when your CFO signs off on a number.

The long-term maintenance cost is where the architecture decision really shows up. A reporting layer is cheap until an API changes. Shopify updates its payout structure, Amazon revises settlement formats, TikTok adds a new fee category, and suddenly your dashboard shows stale or broken data until someone fixes the connection. With a purpose-built warehouse, the vendor absorbs that maintenance burden.

The real audit question: when your dashboard and your accountant disagree, which one wins, and how long does it take to find out?

How to Assess Scenario Modeling for CPG-Specific Decisions

Scenario modeling quality separates faster than almost any other capability during an evaluation. The demo version looks similar across tools. The production version is not.

The first thing to test is isolation. When you adjust a scenario assumption, does the base case stay locked, or does editing one scenario overwrite others? Baseline-dependent modeling means every new scenario risks corrupting the reference you're measuring against. A sandbox approach creates a copy on first edit and leaves the base case untouched. For a CFO who needs to show a board three outcomes without explaining why the numbers shifted, that distinction is not minor.

The second question is whether scenario ranges are calibrated to consumer brand economics or generic percentage inputs. A tool that lets you type "+15% COGS" is a calculator. A tool that shows you what a 15% COGS increase looks like against realistic contribution margin outcomes on either side is a modeling environment. That difference matters for a tariff scenario, a retail launch, or a channel expansion decision where you need a defensible range.

CPG-specific scenarios worth confirming the tool can handle:

  • Tariff impact on landed COGS flowing through to channel-level contribution margin and cash timing

  • New channel launch with CAC ramp assumptions separated from existing channel performance

  • Inventory buy sizing against a forecast that accounts for seasonality and cash position simultaneously

  • Price change impact across channels where fees and fulfillment costs differ

If the vendor's demo scenario is "increase revenue by 10%," that is a SaaS planning tool in a CPG context. Push until they show you a margin scenario that starts at the order level.

Cash Flow Forecasting as a Non-Negotiable Evaluation Criterion

For CPG brands, profitability and cash position tell different stories, often at the same time. You can run a profitable quarter while running out of cash if a large inventory buy lands before your Amazon settlement clears and your biggest retail account pays on net-60 terms. An FP&A tool that shows a healthy P&L without the cash timing underneath it is missing the decision that actually keeps the business running.

A useful 13-week cash flow model requires four things working together:

  • Live bank data so your starting position reflects what actually cleared, not what your ledger shows from two days ago

  • Accounts payable visibility so upcoming vendor obligations are already in the forecast before someone manually enters them

  • Purchase orders flowing in from your inventory system so a large inventory buy (say, a hypothetical $400K) shows up in cash timing the moment the PO is created

  • The ability to model payment allocation across cash and credit while holding above a minimum reserve threshold, because how you sequence payments matters as much as the total outflow

That last point is where most tools fall short. Knowing you owe a hypothetical $600K to three vendors this month is one number. Knowing whether to pay two from operating cash and one from credit is a planning problem. It depends on your Amazon and Shopify settlement timing. Ask vendors whether purchase orders enter the forecast automatically or through manual entry. Manual means your forecast is only as current as whoever last updated the PO log.

AI Capabilities in FP&A Tools: What Separates Useful from Noise

Every FP&A vendor claims AI in 2026. The claim is not the capability.

The practical test is simple: can the AI answer "why did my contribution margin drop 7% last week?" with ranked drivers and attribution by channel, or does it summarize data you already have in a slightly different format? One is analysis. The other is a reformatted dashboard with a chat interface stapled on.

The more important question sits underneath the AI feature itself: what data is the AI actually reading? A model querying raw warehouse exports, inconsistently formatted settlement files, and marketing API pulls with mismatched date logic will produce confident-sounding answers that are wrong. Data quality determines whether outputs from a financial intelligence platform are trustworthy or just fast.

"Don't stick Claude on your data warehouse and expect good results. Zero percent chance. The reason AI fails on financial data is it either has too much contradictory context or not enough."

When the transformation layer handles reconciliation before the AI touches anything, the model works from clean, labeled, ordered inputs instead of a pile of raw exports. That architecture is what makes AI analysis defensible to a board, and actually useful beyond a demo.

Ask vendors two questions: what data layer is the AI operating on, and can you show a live variance analysis in a production account that traces back to an order?

Implementation Reality: Time-to-Value for a CPG Finance Team

The gap between a compelling demo and a working daily P&L is where most FP&A evaluations go wrong. Vendors show you a live account with two years of clean data and a fully balanced P&L. They rarely show you what month one looks like for a brand with a messy COGS sheet, three Shopify stores, and an Amazon account that's been running for four years.

A few questions worth asking before you sign:

  • How long did it take your most comparable customer to get a working daily P&L?

  • What data did the customer have to prepare before setup began?

  • Who owns the COGS upload, the SKU matching, and the chart of accounts mapping?

  • What happens when an integration breaks mid-month during close?

Historical data depth matters more than most FP&A software demos suggest. Cohort forecast accuracy improves meaningfully after four to five months of data and becomes highly reliable after 1.5 to 2 years of history. The more you can supply at onboarding, the faster the forecasting layer becomes reliable.

Ask what "live in weeks" actually means for a brand at your channel count and ERP complexity. For a Shopify-plus-Amazon brand on QuickBooks, standard setup is fast. Add NetSuite, a 3PL, TikTok Shop, and a subscription tool, and the accurate answer from any vendor should be longer.

How Iris Finance Is Built for the CPG Evaluation Criteria Covered Above

Iris Finance is an AI-native FP&A solution built for the evaluation criteria this article covers, not a general planning product retrofitted for CPG. It takes the consolidation, order-level matching, and forecast rebuilding off your finance team's plate.

The data architecture starts with a proprietary warehouse that structures every source before analysis runs. Shopify payouts, Amazon settlements, TikTok order-level fees, subscription data from Recharge or Skio, and ad spend across Meta, Google, and TikTok all flow in and get matched at the order level automatically, with no manual exports or tab-merging on your side. Fin, our AI copilot, reads from that structured layer, which is why it can answer "why did contribution margin drop 7% last week?" with ranked drivers instead of a summary of numbers you already have.

What That Looks Like in Practice

The daily P&L builds itself and refreshes every one to two hours, so nobody on your team pulls exports or rebuilds tabs to get it. Contribution margin by channel is visible before your books close, backed by an order-level audit trail on every line. Scenario modeling uses Bollinger band ranges drawn from anonymized data, so a tariff scenario or channel expansion model is calibrated against what comparable CPG brands have actually achieved.

Every integration, every default metric, and every AI output was designed around how omnichannel consumer brands between $10M and $500M actually run their finances, backed by Iris Finance's recent funding round. It was not built for SaaS finance teams and adapted.

Final Thoughts on Replacing Excel and BI Dashboards for Omnichannel CPG Finance

Most finance teams don't realize how much time they spend maintaining data instead of analyzing it until they stop doing it manually. The evaluation criteria in this article exist because the wrong tool just moves that burden somewhere else. When you're ready to take the data work off your team and see accurate, order-level CPG financials, reach the Iris Finance team.

FAQ

How do I track cohort LTV by channel on Amazon and Shopify without building a custom data warehouse?

You don't need to build your own warehouse. A vertical FP&A tool with native order-level integrations for both channels handles this directly. The critical variable is whether the Amazon connector matches fees, returns, and settlements at the order level or just pulls gross PO data; order-level data is what makes cohort tracking possible, since you need fees, returns, and settlement timing per transaction before you can calculate true acquisition cost and payback by cohort.

How do I know if my TikTok Shop margins are actually profitable after fees and ad spend?

You need order-level fee data from TikTok's commerce API, beyond ad spend from the marketing API: FBT fees, affiliate commissions, and merchant fees stripped out per transaction before contribution margin is calculated. Most tools only read the marketing API, so their TikTok margin numbers are wrong before any scenario runs; if your connector can't show you a per-order breakdown of every fee category, your contribution margin line is overstated.

What is the difference between an FP&A platform and a BI dashboard for CPG financial planning?

A BI dashboard like Tableau or Looker is a visualization layer. It reads from a data source and displays charts, but cannot write back to a planning model, lock a period, or run a scenario when you change an assumption. An FP&A tool connects your actuals to a live model, tracks plan versus actual on a rolling basis, and maintains an audit trail your accountant can rely on; for a CPG brand, the difference shows up at month-end when you need to know which channel drove a 12% revenue decline and what the downstream cash impact looks like.

Should I use an Excel-native tool like Vena or Datarails, a horizontal cloud tool like Anaplan, or a CPG-specific FP&A platform to replace my spreadsheets and BI dashboards?

The answer depends on channel count and data complexity: Excel-native tools extend your runway if your business is relatively straightforward and your team lives in spreadsheets, while horizontal cloud tools like Anaplan are built for general business planning and have no out-of-the-box data model for a Shopify payout, Amazon settlement fee, or subscription cohort. If you're running three or more channels with real-time contribution margin requirements, a vertical tool built around CPG data models is the right category, because the gap in horizontal tools shows up at the data layer, not the modeling layer.

How should we think about large vendor payments when our Amazon settlement timing and Shopify payouts don't align with payment due dates?

This is a sequencing problem, not a cash balance problem. You need a 13-week cash forecast that shows settlement timing from each channel alongside AP due dates, so you can decide which obligations to pay from operating cash and which to route to the credit line while staying above a minimum reserve. The forecast is only reliable if purchase orders enter automatically from your inventory system and bank data reflects what actually cleared, not what your ledger shows from two days ago; manual PO entry means your forecast is only as current as whoever last updated the log.