Resource and Capacity Planning Software Problems: The 7 That Cost Real Money, and How to Avoid Them

Why does the allocation model get scoped as a grid of cells?

Because the artefact being replaced is a grid. Rows are people, columns are weeks, and the request is make this but better. A developer who has not shipped in this category will build exactly that, and it will work until the first fortnight of real staffing.

Services staffing is not weekly. Engagements start mid week and slip by days. A person is on two projects at once at different proportions. Part-time staff have contract hours that are not a fraction of a standard week. Contractors have agreement end dates after which capacity should be zero, not reduced. Leave is booked in half days. Offices in different countries have different standard weeks and different holidays, so the same calendar week is a different amount of capacity depending on who is in it. Every one of those is a rounding decision in a cell model, and rounding errors do not cancel out. They accumulate in the direction of optimism, because a planner filling a cell fills it.

The fix is to model an allocation as a dated assignment range with an effort expression, validated against that person's true capacity: contract hours, their office's holiday calendar, approved leave from the human resources (HR) information system (HRIS), and existing commitments. Bookings carry states, tentative, confirmed and locked, and moving a locked allocation requires an approval rather than a keystroke. Every change is logged. That last part matters more than it sounds, because it ends the Monday argument about who changed what, which is a large share of why the meeting runs long.

What goes wrong migrating the staffing spreadsheet?

The workbook is not a data source, it is an archaeological site. Five years of accumulated convention where a colour means something, a merged cell means a different thing, and a name in brackets means tentative unless it is in the third tab where it means someone on notice. Nobody wrote any of it down and the person who established the convention has usually left.

Three things go wrong when teams try. Somebody proposes retyping the current quarter and starting fresh, which loses the history that makes utilization trends and hiring cases possible, and produces a week of chaos during cutover. Somebody else writes a parser that reads the cells and not the formatting, which silently drops the tentative and leave semantics that live entirely in colour. And in every migration the parse surfaces conflicts the spreadsheet was hiding, usually double bookings that existed for months because a formula was overwritten, and the reflex is to treat those as migration bugs rather than as findings.

The fix is to parse the workbook properly, formatting included, map every tab, colour code and merged region to structured allocations, and produce a reconciliation report of everything ambiguous for a human to resolve. Then run both systems in parallel for at least two full resourcing cycles, comparing utilization and bench by office each week, and cut over only after one cycle matches on both sides. Treat the conflicts the parse exposes as the first return on the project, because they are exactly the double bookings that have been costing you client escalations.

Why do CRM, timesheet and HRIS integrations break after launch?

These three carry the value and they degrade in three different ways.

The timesheet connection to Harvest, BigTime or Deltek breaks on identity. A project renamed in the timesheet tool no longer matches the plan, so actuals stop landing against the allocation and planned versus actual quietly corrupts for a quarter before anybody notices, which is precisely the reporting the build was justified on. The customer relationship management (CRM) connection to Salesforce or HubSpot breaks on process: a team adds a pipeline stage or renames one, the probability weighting no longer maps, and shadow projects either stop being created or are created for deals that are not real. The HRIS connection to BambooHR or Workday breaks on lifecycle: a leaver is processed, their capacity should go to zero from their last working day, and instead they sit on the chart as available because the integration only syncs new joiners.

The fix is the same discipline in all three. Validate every inbound payload against a strict schema and fail loudly rather than absorbing whatever arrives. Map on stable identifiers rather than on names, and when an identifier cannot be matched, queue it for a human instead of guessing. Monitor freshness so a feed that stops delivering raises an alert the same day, since a stale plan looks exactly like a quiet week. Sequence the integrations rather than launching four at once, starting with the timesheet system, because planned versus actual is the core loop and the others are worth less without it.

What happens when multi-entity rules and cost-rate access are missing?

Two gaps show up late and cost more than their scope suggests.

The first is multi-entity reality. Each office needs its own holiday calendar, its own working week definition and its own payroll rules. Rate cards need versioning by role, office and effective date. Contractor agreements need end dates that automatically zero out capacity. Contractors invoicing in one currency against projects billed in another need normalisation for a consolidated margin view while local teams keep local numbers. A system built around one calendar and one currency does not extend to a second office, it gets replaced, and the trigger is usually an acquisition that nobody warned the project about.

The second is access to cost rates. This system holds cost rates, which are a salary proxy, alongside personal data for staff across jurisdictions. Delivery leads need availability and skills. They do not need what their colleagues cost. If role based access is bolted on as a screen level permission rather than designed into the schema, it leaks through an export, a report or an interface, and that particular leak is unrecoverable as an internal matter.

The fix is to model entities as first class objects from the start, and to design role based access into the data model so cost rate is a separately governed attribute rather than a column somebody can widen a query to reach. Add single sign on through Okta or Azure AD, an audit log of who viewed rate data, and for UK and EU staff a lawful basis, data minimisation and a deletion workflow for leavers under the General Data Protection Regulation.

Should you build custom or configure what you already own?

If you are under roughly 75 billable people in one office, buy. Float or Resource Guru plus a disciplined weekly call is the correct answer at that size, everyone knows everyone, staffing by name works, and per-seat pricing is a rounding error against the cost of a build. Runn is worth a look if tentative project modelling is your main gap. Kantata suits firms that want an all-in-one professional services suite and will adapt their process to match it, which is a genuine trade rather than a compromise.

The signals to build are concrete and they usually arrive together. A dedicated resource manager spends 15 or more hours a week reconciling systems. A double booking caused a client escalation in the last two quarters. Utilization is known three weeks after the fact. The per-seat bill has crossed roughly $40,000 a year and people still export to a spreadsheet to answer real questions. A second office or an acquisition has just multiplied calendars, currencies and rate cards. At 150 or more billable people with pipeline driven staffing, the spreadsheet plus per-seat tool arrangement leaks more per year in lost utilization and blown fixed-fee margin than the first release costs to build.

How do hidden costs get into the quote?

A focused first release covering the allocation engine with conflict validation, a skills matrix, availability and bench views, and one CRM plus one timesheet integration runs $60,000 to $130,000 over 12 to 16 weeks in our delivery experience. A full platform adding pipeline weighted scenario planning, margin analytics, hiring triggers and multi-entity support runs $150,000 to $400,000 over 6 to 12 months. The overruns come from a short list.

• Integration count above everything else. Salesforce, Harvest, Workday and NetSuite are each their own workstream, so sequence them rather than launching four in parallel and paying for four simultaneous discovery efforts.
• Spreadsheet migration, because five years of workbook archaeology where colours carry meaning takes real effort to parse and a reconciliation pass by someone who knows the conventions.
• Scenario planning depth. Showing weighted pipeline on a timeline is one thing. A full scenario board with hiring triggers and what-if comparison is another, and the two get quoted as if they were the same feature.
• Approval workflows across offices, which look trivial until you discover that each office has a different idea of who can move a locked allocation.
• Parallel running, which needs two to four weeks of real calendar and someone's attention on both systems at once.

What separates a build that works from one that fails here?

Ask to see the allocation data model before signing anything. The honest answers involve dated assignment ranges rather than person-week cells, split allocations across projects, partial full-time equivalent (FTE) capacity, contract end dates and timezone safe week boundaries. A vendor who proposes a calendar table has never watched this category break in production.

Demand integration receipts rather than claims. Named, versioned work against a CRM, against a timesheet system and against an HRIS, and a specific answer to how they reconcile a project renamed in the timesheet tool, because that single incident corrupts planned versus actual for a quarter when it is handled naively.

Require a migration plan for the spreadsheet itself: parsing the workbook including formatting, mapping every tab and colour code into structured allocations, surfacing the conflicts the sheet was hiding, and parallel running both systems for two full resourcing cycles before cutover. A vendor who says they will just re-enter the data is planning a week of chaos on your behalf.

Ship the allocation core first and treat dashboards as phase two. Utilization charts built on a model that cannot represent a Wednesday start are confident and wrong, and once leadership has seen a number they will keep asking for it.

Check the compliance posture and settle ownership. Role based access designed into the schema so delivery leads never see cost rates, single sign on, deletion workflows for leavers, and a contract assigning you full ownership of the source code, the schema and the infrastructure configuration. Digital Heroes assigns all of that on custom builds. Anything less rebuilds the lock-in you are paying to escape.