The Inventory Forecasting System Every DTC Brand Needs by Year Two

Most DTC brands run their inventory on a spreadsheet for too long. It works at $500K. It mostly works at $2M. It starts breaking at $5M and it's actively destroying the business by $10M, and by the time founders realize the system is the problem, they've already eaten through six months of growth on stockouts and dead stock.

I've watched this in my portfolio repeatedly. Smart founders, good operators, growing brands — all running inventory on a Google Sheet maintained by one person who hasn't taken a real vacation in two years. Eventually that sheet either breaks or that person leaves, and the brand learns the hard way that inventory was the actual business under the marketing.

Inventory forecasting isn't sexy. It doesn't get celebrated on Twitter the way creative tests and ad spend do. But it's the system that determines whether your growth compounds or eats itself. Here's the framework I push portfolio companies to build before the spreadsheet collapses.

Why the Spreadsheet Stops Working

The spreadsheet works when three conditions are true: you have under 20 SKUs, you have one supplier per SKU, and you have predictable demand. Take any of those away and the system starts producing bad decisions.

The specific failure modes I see repeatedly:

Demand spikes get missed because the forecast was last updated three weeks ago and the founder is too busy to refresh it
Lead times drift without anyone noticing — a supplier slipped from 30 days to 45 days six months ago and the safety stock never adjusted
Bundle SKUs get under-counted because the spreadsheet treats them as their own SKU instead of decomposing them into component pulls
Promotional spikes get over-corrected when one influencer drop spikes demand for two weeks, the forecast extrapolates the spike into a permanent baseline, and the brand orders six months of dead stock
Cross-SKU effects get ignored — when you launch a new color of an existing product, demand for the old colors moves, and the spreadsheet treats them as independent variables

By the time you have 50 SKUs and three suppliers, you can't model these effects in a sheet. You can pretend you can. You can't.

The Three Numbers That Actually Matter

Strip away the complexity and inventory forecasting is really about three numbers per SKU. If you can get these three right, you can forecast accurately. If you can't, no amount of dashboard sophistication will save you.

Demand velocity, lead time, and safety stock — get these right per SKU and the rest of the system follows.

Demand velocity. How many units do you sell per day, on a rolling basis? Not your trailing 12-month average — that's too smooth. Not your last week — that's too noisy. The number you actually want is a weighted blend: heavier weight on the last 30 days, with a longer baseline of the prior 90 days for trend correction. Most brands use the wrong window and either overreact to spikes or miss real growth.

Lead time. How long from purchase order to inventory available on the shelf? This needs to include everything: manufacturing time, transit time, customs delays, QC, receiving. Most brands track only manufacturing time and get burned by the rest. The real lead time is usually 30-50% longer than the number the founder quotes.

Safety stock. How much buffer do you carry to handle demand variability and supply variability? This isn't a single number — it should be calculated per SKU based on demand variance, lead time variance, and the cost of a stockout. High-margin, high-demand SKUs should carry more safety stock. Low-margin, slow-moving SKUs should carry almost none.

A real inventory system tracks these three numbers per SKU, recalculates them weekly, and uses them to generate reorder triggers. That's it. The rest is plumbing.

What to Forecast Weekly vs. Monthly

The cadence question trips up most operators. Some things need to be looked at every week. Some things should be looked at every month. Mixing the two creates noise.

Weekly cadence:

Stock position by SKU — what's on hand, what's on order, what's at the 3PL, what's in transit
Reorder triggers — anything below safety stock plus lead-time demand gets flagged
Spike detection — any SKU running 30%+ above forecast for the trailing two weeks
Stockout risk window — projected days of cover for each SKU at current velocity

Monthly cadence:

Demand velocity recalculation — refresh the weighted demand number per SKU
Lead time recalculation — pull actuals from the last 90 days and update planned lead times
Safety stock recalibration — adjust based on observed demand and supply variance
Dead stock review — anything that hasn't sold in 60 days gets flagged for action
New product launch impact — how did the new SKU affect demand for adjacent SKUs

The weekly view is operational — keep the business running. The monthly view is strategic — fix the model that drives the operational decisions. Most brands collapse them into a single weekly review and end up doing neither well.

When to Build vs. Buy

Eventually every brand asks whether to build their own inventory system or buy one. The honest answer depends on where you are.

Under $2M ARR: Don't build. Don't even buy. Use a tighter spreadsheet with proper formulas, set weekly recurring reviews, and accept that you'll have some stockouts. The cost of building or buying isn't justified at this scale, and the engineering overhead will distract from growth.

$2M to $10M ARR: Buy. There are good off-the-shelf tools — Cogsy, Inventory Planner, Brightpearl, Cin7, NetSuite — that will solve 80% of the problem for a few hundred to a few thousand dollars a month. Pick one that integrates with your Shopify and your 3PL, and accept that it won't be perfect. The cost of building custom at this stage is rarely worth it.

$10M+ ARR: Reevaluate. By this point you have enough SKU complexity, supplier complexity, and channel complexity that the off-the-shelf tools start to feel constraining. Some brands stick with the tools and add custom layers on top. Some brands build proprietary forecasting systems because their specific dynamics justify it. There's no single right answer at this stage — it depends on the business.

The mistake brands make is jumping the gun. Either they buy a $30K/year tool at $1M ARR and realize they don't have the operational maturity to use it, or they try to build something custom before they've figured out the basics on a spreadsheet. Match the system to the stage.

The Human Layer the Software Can't Replace

Even the best inventory software can't make the calls that actually matter. The decisions that determine whether a brand wins on inventory are almost always human:

Whether to bet on a launch. When you're forecasting demand for a new SKU with no history, the software gives you a range. Someone has to decide whether to order 5,000 or 25,000 units, and that decision lives or dies on judgment.
Whether to ride out a stockout or expedite. When you're three weeks from stocking back in and demand spikes, you can airfreight at 3x cost or you can let the SKU go out of stock. The right call depends on margin, brand position, and how the stockout will affect repeat purchase.
Whether to discontinue a SKU. Some SKUs underperform on the spreadsheet but are strategically important for the brand. Software will tell you to kill them. A good operator will know when to override.
Whether to renegotiate with a supplier. When lead times are stretching or quality is slipping, the data tells you something is wrong. Someone has to decide whether to push back on the existing supplier or move to a new one — and that's a relationship call, not a model call.

The brands that win on inventory have software that handles the calculation and operators who handle the judgment. The brands that lose either trust the software completely or refuse to use it at all.

Start Before You Need It

The single most common pattern I see is brands building their inventory system after the first major stockout or the first major dead stock write-down. The lesson is always the same: we should have built this earlier.

If you're under $2M, you don't need much yet — but you should know what numbers you'll need to track when you scale, and start tracking them now in whatever form works. If you're between $2M and $10M, pick a tool and implement it before the spreadsheet kills you. If you're past $10M without a real system, the cost of the next mistake is already higher than the cost of fixing the system.

Inventory is the unglamorous foundation of every consumer brand. Get it right and growth compounds. Get it wrong and growth eats itself. The brands that scale are the ones that build the boring systems before the boring problems become urgent ones.