From Discovery Call to Defensible Project Estimate

Pass one: capture, don't interpret

During the conversation, your only job is capture. Recordings, transcripts, notes, and shared documents all belong to the same engagement record. Interpretation during the call costs you the detail you will want later.

The failure mode here is a single person holding context in memory. If the estimator and the person who ran discovery are different people, undocumented context becomes invisible.

Pass two: split the transcript into four buckets

Everything said in a discovery session sorts into four categories, and each one has a different downstream use.

• Requirements — things that must exist or happen for the engagement to succeed.
• Open questions — things you cannot estimate confidently until answered.
• Assumptions — beliefs you will proceed on unless corrected.
• Risks — plausible conditions that would change effort or approach.

Pass three: resolve or price the open questions

Open questions are the real source of estimate volatility. Each one has three legitimate endings: answer it before estimating, convert it into a stated assumption, or exclude the work it touches.

What is not legitimate is quietly absorbing it into a padded line item. Padding hides the question, and hidden questions reappear as change orders that feel like surprises to the client.

Pass four: build the estimate from named services

Estimating from a defined catalogue of services rather than a blank spreadsheet does two things: it keeps your language consistent across engagements, and it makes each line explainable in terms the client already heard in discovery.

Structure the estimate in phases that match how you will deliver. A phase structure that mirrors delivery makes the eventual Statement of Work, and the eventual project plan, fall out almost for free.

• Each line names a service, a quantity, and the requirement it satisfies.
• Effort for management, coordination, and review is visible, not buried.
• Phases match delivery reality so scope can be deferred cleanly.

Pass five: test the estimate against three questions

Before the estimate leaves your team, walk it through the questions a sceptical client or partner will ask.

• Which requirement does this line exist to satisfy?
• What would have to be true for this number to be wrong by half?
• If the budget were 20% smaller, what would we defer — and does the phase structure allow it?

Review it with the client while it can still change

The most expensive estimating habit is sending a finished number and waiting. A live review, where scope can be trimmed, deferred, or reshaped in the moment, converts a negotiation into a joint decision.

It also produces something valuable for the document stage: a record of what the client chose, not just what you proposed.

Why the pre-sales cost of this is worth measuring

Discovery and estimating are largely non-billable, and they are not cheap. Service Performance Insight's 2024 Professional Services Maturity Benchmark reported an average of 109 non-billable hours per consultant per year on business development and sales support, against average billable utilisation of 71.2%. The same benchmark put the average sales cycle from qualified lead to signed contract at 81 days and the win rate at 48.2% of bids.

Read those together and the arithmetic is uncomfortable: roughly half of the scoping effort your senior people spend produces no revenue at all. That is not an argument for scoping less carefully. It is an argument for scoping the same way every time, so the effort is reusable and the parts that repeat stop being reinvented.

The four artifacts a discovery pass should leave behind

A discovery session is finished when it has produced documents, not impressions. If a colleague could not pick up the engagement from what you wrote down, the session is not complete.

• A requirements list, each item traceable to something someone actually said.
• An open-questions register with an owner and a due date per item.
• A stated assumptions list, written in language a client can challenge.
• A risk list with the effect on effort or approach, not just a label.

Estimate ranges early, single numbers late

Early in discovery you know the shape of the work and not much else. A single number at that stage is a commitment disguised as an opinion, and clients anchor to it permanently.

Give a range with the reason for its width — “the spread is driven by how many of your six source systems need custom mapping” — and the conversation moves to the driver rather than the price. As open questions close, the range narrows for a stated reason. By the time you present a single number, the client has watched it earn its precision.

Handle the estimate the client cannot afford

Sooner or later your defensible number exceeds the available budget. The instinct is to shave hours, which quietly converts a scoping problem into a delivery problem and a margin problem.

The alternative is structural. Because the estimate is built from named services in delivery-shaped phases, you can defer a phase, reduce a driver, narrow the audience, or move a deliverable to the client's own team — each with a visible price. The client chooses what to give up, which is a materially different conversation than being told the price came down.

• Defer a phase intact rather than thinning every phase.
• Reduce a driver: fewer integrations, fewer reports, fewer entities in phase one.
• Shift effort to the client where they have capacity, and state the dependency.
• Trade scope for terms: shorter review windows, fewer revision rounds.

Close the loop after delivery

The estimate stops being a guess only when you find out how it landed. Compare actual effort by phase and role against what you estimated, and attribute each variance to a cause: missing scope, wrong driver, client delay, or execution.

SPI's 2024 benchmark reported an average project overrun of 9.9% and average project margin of 35.5%, with surveyed firms achieving 89.7% of their annual margin target. Those gaps are calibration data. Firms that review them get better at estimating; firms that do not get more confident.