The tool recommends. Somebody still decides.
Planning software and managed planning get shopped as if they were rivals. They are not. One sells arithmetic at scale. The other sells the thing the arithmetic has always depended on: inputs kept true, and a decision made every week, on time, by name.
What the software genuinely does well.
Per-SKU forecasting across a thousand rows, reorder math that never mistypes, alerts that never sleep. A good planning tool is a spreadsheet that stopped being fragile, and for the arithmetic itself the category delivers. The failure lives somewhere else.
Three failures no algorithm can catch.
Each of these has its own field note in the blog, written from the operational side of live accounts.
Inputs decay
The supplier moved the date in a chat thread. The freight quote assumed a sailing that shifted. The real world updates itself weeks before anyone updates the software, and every recommendation in between is precise about a fiction. The field note →
Systems disagree
Platform, marketplace, 3PL: three counts, and they drift apart a little every week. The forecast cannot tell which number is lying, so it builds confidently on whichever one it was fed. The field note →
Alerts outnumber decisions
Forty flags is not a plan. A plan is quantities, dates and a deposit approval, made before the order-by date passes. Flags that nobody converts scroll off the screen and come back as a stockout with your name on it. The field note →
The documented version of an input nobody questioned: a just-placed purchase order, caught and stopped at roughly 400 days of coverage. The model approved it. The inputs justified it. Neither was true. The catch →
What a managed layer actually is.
One build that models your catalog: every SKU, every stock location, every lead time as a path rather than a guess. Then a weekly rhythm: the plan lands before a 30-minute call, with suggested POs, transfer recommendations and risk flags, and the decisions get made with dates on them. A senior lead owns the account. Specialists keep the numbers current in between. The full mechanics →
Signs software alone is enoughkeep the subscription, skip us
- Somebody on your team already owns the weekly numbers, by name
- Your counts reconcile across systems and stay reconciled
- Under 50 SKUs, or restock is days rather than months
- The last stockout was a demand shock, not a missed date
Signs the gap is ownershipthe part we sell
- Recommendations pile up unactioned between fires
- The tool, the marketplace and the 3PL tell three stories
- Order-by dates pass quietly and surface as emergencies
- The founder is still the planner at 11pm
A software vendor cannot publish the left column without hurting itself. We can publish both, because the fit filter is the product working.
Different line items for different jobs.
A subscription prices arithmetic. A managed layer prices responsibility: the build is $1,500 once, Core is $1,250 a month, Partner is $2,250, published in full with the scope table on the services page. The comparison that actually matters is not tool against service. It is the cost of the decision going unowned for another quarter, and you can put your own numbers on that in two minutes.
The three questions this page gets.
Should we cancel our planning tool if we hire you?
Not on our account. Some client accounts plan on their own SoStocked or Inventory Planner subscription and the cadence runs beside it. Others run on the workbook the build installs and skip the subscription. The deciding factor is usually catalog size and how much of the tool you actually use, and we will tell you which side you land on in the first call.
Is this just a fractional hire with extra steps?
A hire gives you one person and their memory. The Engine gives you a documented system run to a weekly rhythm: a senior lead owns your account, specialists keep the numbers current, and every decision is written down as it is made. It survives any one person having a week off, which memory does not.
How fast does the managed layer start paying attention?
The build phase starts on data access, and the documented record is a working multi-node master sheet inside 72 hours: every SKU, every stock location. The weekly rhythm starts as soon as the model is signed off.
SoStocked is a product of Carbon6. Inventory Planner is a product of Sage. Names appear for factual comparison only. No affiliation or endorsement is implied, in either direction.
Talk it through against your catalog.
Thirty minutes against your own catalog settles most of this page. Bring a stock report, leave with the 90-day risk read, whichever shelf you buy from afterwards.