Consulting Firm Software: Fixing the Margin Leak Between Time, Staffing, and Billing

Why resourcing and margin software makes or breaks a consulting firm

The sixth business day of the month. Your finance lead has three browser tabs open: Harvest for time entries, Float for who was supposed to be on what, and a Google Sheet called "Margin_Master_v11_FINAL_use_this" that a controller who left in 2023 originally built. She is trying to answer one question the managing partner asked in a Slack message at 8pm: did the Meridian engagement make money?

It takes her four hours. Two consultants logged 60 hours to a project code that was closed in the prior phase. One senior manager billed at a rate that was renegotiated mid-engagement, and Harvest still has the old rate. A subcontractor's invoice is sitting in QuickBooks unmatched to any project. The Float plan says a designer was allocated 20 hours a week, but she actually spent that time on a pursuit that never converted, which nobody coded anywhere. The answer comes back: 34% gross margin. Nobody in the room believes it, and by the time anyone could verify it, the engagement is over and the pricing lesson is lost.

The dollar leak is not theoretical. At a 60 consultant firm billing an average of $180 an hour at 70% utilization, one point of realization is roughly $150,000 a year. In the firms we work with, realization leakage surfaces about six weeks after it happens, which means they are pricing the next three proposals on numbers that were already wrong. Meanwhile the resourcing manager runs a Tuesday staffing call off a spreadsheet exported from Float, and every partner in the room lobbies for the same two people, because the tool tracks who is allocated but has no idea who is actually good at healthcare payer work.

Problem 1: time, rates, and revenue live in three systems that never agree

Harvest tracks time against a project code. Your MSA has a rate card with four levels, a 10% volume discount above 500 hours, and a blended rate exception for the client's subsidiary. QuickBooks or NetSuite has the invoice. Nothing connects the three. So when a partner discounts a senior manager's rate from $325 to $290 in an email during a scoping call, that change lives in an email. Time keeps flowing at $325. You invoice at $290. The variance shows up as a mystery in the WIP account and gets written off at quarter close as "billing adjustment."

Off-the-shelf cannot fix this because Harvest models a rate as a single number on a person or a project. It does not model a rate card as a contractual object with effective dates, role levels, escalators, and exception clauses. Neither does Float. Even the mid-market suites like Kantata or Deltek treat rate complexity as a configuration problem you solve with custom fields, which means the logic ends up in someone's head and a spreadsheet anyway.

A custom build makes the rate card a first-class entity. You model the engagement contract: role levels, rates, effective date ranges, volume tiers, annual escalators, expense caps, subcontractor pass-through markup. Every time entry resolves its billable value against the contract at the moment it is entered, not at invoice time. When a partner changes a rate, they change it in one place and the system re-prices open WIP and flags what already invoiced at the old rate. Realization variance becomes visible on day two of the engagement instead of day 60. This is a useful place for AI: an extraction model reads the signed MSA and SOW PDFs and drafts the rate card object, role mappings, and cap clauses for a human to approve in about ten minutes instead of an analyst spending two hours keying it in and mis-keying one field.

Problem 2: staffing decisions are made on availability, not on fit or margin

Float and Resource Guru answer "is this person free." That is a calendar question. The actual question a resourcing manager is answering is: given that this engagement is priced at a 42% target margin, who can I put on it who has payer-side claims experience, is not about to roll off onto the Kellerman pursuit, whose cost rate keeps us above target, and who has not been on the road four weeks running and is about to quit?

None of the off-the-shelf tools hold the data to answer that, because they do not model a consultant as anything richer than a name with a capacity number and maybe a tag. They have no cost rate, no skill taxonomy tied to your practice areas, no travel load, no engagement history, no bench cost.

What a custom build does: model people with cost rate, target rate, skill and industry tags, certifications with expiry, travel days trailing 90, and prior engagement history. Model the demand side with pipeline probability from your CRM (Customer Relationship Management). Then the staffing screen shows candidate consultants ranked with the margin impact of each choice rendered live. Put the $95-an-hour cost senior on it instead of the $140 principal and the engagement margin recalculates in front of the partner. This is where a forecasting model pays off. Concretely: forecast bench four to eight weeks out by joining committed roll-off dates against weighted pipeline, and flag the specific people who will be unassigned. Not a dashboard nobody opens, a Monday morning message to the resourcing manager naming five people and the number of unallocated weeks. Every week of bench for a $140 cost consultant is roughly $5,600 out the door.

Problem 3: timesheet compliance is a nagging problem, and it corrupts everything downstream

Consultants hate timesheets. So they fill them Friday at 6pm from memory, in four-hour blocks, against whichever code they remember. Your utilization number, your margin number, and your next proposal's estimate all rest on data a tired person guessed at. Harvest sends a reminder email. That is the entire intervention available to you.

Off-the-shelf tools cannot go further because they have no signal about what the consultant actually did. They are a form.

Custom builds get to use the signal you already own. Pull calendar events, project channel activity, and document edits, then pre-populate a draft timesheet the consultant confirms or corrects in 90 seconds on their phone. This is a legitimate AI use case and one of the highest-ROI things we build in this category: an assistant that proposes "Tuesday: 3.5h Meridian discovery workshop, 2h Meridian deliverable prep, 1.5h internal" based on real calendar and activity signal, and the consultant taps accept. Compliance goes from Friday-night fiction to same-day fact. Then add validation the incumbents cannot: block time against closed phases, warn when someone books past the SOW hours cap, require a note when an entry exceeds 10 hours. The whole margin stack gets more trustworthy because the input got more trustworthy.

Problem 4: nobody can see engagement economics until it is too late to act

The classic failure: a fixed-fee $400,000 transformation engagement priced at 45% margin. By week nine, scope has quietly drifted, two extra consultants are on it, and the real margin is 12%. The partner finds out at month close in week twelve. There is no intervention left. You either eat it or have an ugly conversation with a client who was never warned.

Harvest can show hours against a budget. It cannot tell you burn rate against the delivery plan by workstream, cannot separate scope drift from estimation error, and cannot alert on trajectory.

A custom build makes the engagement plan a structured object: phases, workstreams, deliverables, planned hours by role level, planned costs including travel and subcontractors. Actuals flow against that structure daily. The system computes projected margin at completion using current burn rate, not just consumed budget, and alerts the engagement partner when projected margin drops more than five points below the sold margin. Week four, not week twelve. That is enough runway to have a change order conversation while the client still remembers agreeing to the extra scope. On the same data you get a change-order draft: when the system detects unplanned deliverable work, it assembles the evidence trail of hours and dates so the partner is not reconstructing an argument from Slack history.

Problem 5: subcontractors and pass-through costs are a black hole

The firms we build for above 50 people typically run 10% to 25% of delivery through subcontractors and independent consultants. Those people are not in Harvest as employees. Their invoices arrive as PDFs by email, get keyed into QuickBooks by an ops person, and land in a GL account that may or may not carry a project dimension. Your engagement margin is wrong by whatever that number is until close.

Off-the-shelf project tools have no concept of a subcontractor with a cost rate, a purchase order cap, and a markup on pass-through. You end up with a shadow spreadsheet, and the shadow spreadsheet is where firms lose the most money in this category.

Custom: subcontractors enter time in the same system against the same project codes, with their cost rate and PO cap attached. The system accrues their cost to the engagement in real time and blocks time entry past the PO cap. Document extraction on inbound subcontractor invoices matches lines to accrued time and flags variances above a threshold you set, say $500 or 5%. Expenses and travel carry the same treatment against client-specific expense policies, because most MSAs cap travel at a percentage of fees and nobody tracks it until the client disputes an invoice.

Cost and timeline bands

These are Digital Heroes numbers from delivery across 2,000-plus projects, not an industry survey. A focused first release for a consulting firm, meaning contract and rate card modeling, time capture with the assisted entry, staffing with margin impact, and live engagement economics with alerting, typically runs $60k to $130k and ships in 12 to 16 weeks. A full platform, adding resource optimization, revenue recognition, client portals, pipeline-integrated demand forecasting, and multi-entity consolidation, runs $150k to $400k phased over 6 to 12 months.

What pushes price up specifically in this category: revenue recognition under ASC 606 or IFRS 15 with percentage-of-completion, which is a real engineering problem and not a checkbox, and typically adds $30k to $60k. Multi-currency and multi-entity, where a UK subsidiary bills a US client and intercompany transfer pricing has to net out. Deep two-way integration with a legacy ERP (Enterprise Resource Planning), Deltek Vantagepoint and older NetSuite instances are the expensive ones. Client-specific portals where the client sees approved time before invoicing, common in staff-augmentation-heavy firms. And migration of historical engagement data if partners want trend analysis on the first day, which is usually worth doing and usually underestimated.

Build versus buy: take the off-the-shelf tool seriously first

Stay on Harvest plus Float if you are under about 30 to 40 billable people, run mostly time-and-materials on a simple rate card, and have one legal entity in one currency. The tools cost a few hundred dollars a month, and the spreadsheet reconciliation genuinely does not hurt enough yet. Same answer if your firm is one practice doing one shape of work, because the complexity a custom build absorbs does not exist in your business.

Kantata and Deltek Vantagepoint are real products and they solve real versions of this. Buy one if your process can bend to theirs and you have the appetite for a six-figure implementation plus annual licensing that scales with headcount. Many firms should.

Build when three of these are true: your rate cards have exception logic that no tool can express, so a person is the source of truth; more than 15% of delivery runs through subcontractors; you operate more than one entity or currency; your resourcing manager is the single point of failure and everyone knows it; or you have already paid for a suite implementation and the firm quietly went back to the spreadsheet. That last signal is the loudest one. It means the mismatch is between your operating model and the tool's assumptions, and another vendor will not fix it.

My position: firms over 50 consultants with real rate complexity should build the margin and rate engine and integrate around it. Do not rebuild general ledger, do not rebuild CRM, do not rebuild HR (Human Resources). Build the thing that is specific to how you make money, which is the contract-to-time-to-margin chain, and let it push clean journal entries into whatever accounting system you already own.

How to choose a developer for consulting firm software

Ask them to whiteboard the data model for a rate card in ten minutes. If they draw a rate as a column on a person or a project, walk away. The right answer involves effective-dated rate records, role level mappings, contract-scoped overrides, and a resolution function that takes a time entry and returns a billable value. A developer who has not built this before will not know to ask about escalators or volume tiers, and you will find out in month four.

Ask what happens to a timesheet entry after it is approved but before it is invoiced, and whether it can change. The answer should involve immutable entries with adjustment records, not editing history in place. Firms get audited, clients dispute invoices, and partners re-open closed periods. A developer who has not lived through a client demanding backup for a $40,000 line item will build a system that quietly overwrites the evidence.

Push hard on the accounting integration. Ask specifically how they handle WIP, unbilled revenue, and the journal entries at period close for your system of record, whether that is NetSuite, QuickBooks, Xero, or Sage Intacct. The right answer names the actual objects and the reconciliation strategy when the two systems disagree, because they will. A vague "we'll use their API" means they have not done it.

Ask about data residency and client confidentiality constraints. If you serve regulated clients, some MSAs require that engagement data stays in a specific jurisdiction or is segregated. A developer who has built for consulting firms will bring this up before you do, along with how they handle SOC 2 evidence collection, because your clients' procurement teams will ask you for it and you will need to answer.