TL;DR: Implementation teams pay a hidden tax every day: the time spent reconstructing account context that already exists somewhere in their tools. The research puts numbers on it: 1.8 hours per day searching for information (McKinsey), 23 minutes of lost focus per interruption (UC Irvine), about 4 hours a week lost to app toggling alone (HBR). Our own interviews with forward deployed engineers add the delivery-specific bill: three hours of prep per customer meeting and handoffs described as "broken telephone." The customer pays too: 24% cite repeating themselves as a top onboarding complaint.

Every implementation team knows the feeling, even if nobody has priced it. The fifteen minutes before a customer call spent scrolling the Slack channel. The status meeting that exists so the manager can extract what the tracker doesn't hold. The new team member asking "where do I find out what we promised them" and receiving, in order: a shrug, three links, and "ask Sarah."

None of this appears in any budget. All of it comes out of delivery capacity. Call it the context tax: the recurring cost of reassembling account knowledge that your tools captured but scattered. This article puts numbers on it, from public research and from our own interviews with the people who pay it most.

The baseline: what knowledge work already loses

Start with what's documented across knowledge work generally, before adding implementation's specific multipliers.

McKinsey's analysis found employees spend 1.8 hours per day, about 9.3 hours a week, searching for and gathering information. Not doing the work: finding the inputs to the work.

A Harvard Business Review study instrumented workers at three Fortune 500 companies and counted roughly 1,200 toggles between applications per day, costing about four hours per week purely in reorientation, before any actual searching starts.

And Gloria Mark's research at UC Irvine measured what each significant interruption costs: an average of 23 minutes and 15 seconds to fully regain focus. Asana's Anatomy of Work research rounds out the picture: 58% of knowledge workers' time goes to "work about work," coordination, status, and searching, rather than the skilled work they were hired for.

The implementation multiplier

Those numbers describe everyone. Implementation, onboarding, and forward deployed teams sit at the worst intersection of the causes:

They run many accounts in parallel. An implementation manager or FDE carries anywhere from 3 to 20 concurrent customers. Each account switch is a full context reload of different promises, stakeholders, and blockers. One FDE we interviewed described talking to four clients a day. At UC Irvine's 23-minute refocus cost, the switching alone is a working hour, and that assumes each reload succeeds on the first try.

Their truth lives in conversations. Delivery reality is made on calls and in shared Slack channels, then partially and inconsistently written down. One interviewee described the daily experience as playing broken telephone with outdated documentation: the wiki says one thing, the customer remembers another, and the actual answer is in a call recording nobody has time to rewatch. Another put meeting prep at roughly three hours per customer meeting, most of it reconstructing what was already said.

They inherit context loss from upstream. The sales-to-delivery handoff is where the tax compounds. OnRamp's research with 182 customer engagement leaders found only 47% of teams rate their sales handoff as excellent, even though 66% say a structured handoff makes customers far more likely to reach first value on time. Whatever context sales doesn't transfer, delivery pays to rebuild, usually by asking the customer to repeat themselves.

The customer pays it with you

The context tax isn't internal. In ABBYY's State of Intelligent Automation survey, customers named their top onboarding complaints: onboarding too slow (25%), lack of communication (25%), and having to repeat themselves (24%). All three are context failures. The customer told your company everything during the sales cycle. When implementation asks again, they learn your organization doesn't retain what they say, three weeks after signing.

That lesson shows up in revenue. The same OnRamp research ties first value inside 30 days to renewal and expansion likelihood, and every context reconstruction cycle pushes first value out.

A back-of-envelope for your own team

Take a conservative version of the numbers for a 10-person implementation team:

  • Information hunting at even half the McKinsey figure: 0.9 hours per person per day, so 45 hours a week across the team.
  • Meeting prep reconstruction: 10 customer meetings per person per week at even 30 minutes of context rebuild each, another 50 hours.
  • One status ritual (the meeting that exists to extract status): 10 people, 1 hour, weekly.

That's over 100 hours a week, more than two full-time delivery people, spent producing zero customer value. Capacity per implementation manager is usually the binding constraint on how many customers a team can onboard, and the tax is the largest reclaimable line item in it.

Why the usual fixes don't hold

Teams have tried three fixes for decades, and each fails for a structural reason:

  • Write more down. Documentation decays the moment the project moves, and the writing itself is more tax. The broken-telephone problem our interviewees described happens with documentation, not despite it.
  • Add a status ritual. Weekly syncs extract context from heads into a meeting, where it evaporates again. The ritual is the tax with a calendar invite.
  • Buy a better tracker. Trackers hold what humans remember to type. The context that matters, what was said, promised, and left hanging, never enters them.

The structural answer is different: stop asking humans to move context, and derive it from where it already lives. That's the premise of the project brain, a system that continuously assembles each account's state from calls, Slack, email, CRM, and tickets, with every claim linked to its evidence. It's the approach Tandem takes with its account brain, and the reason teams like Aircall and Qonto run delivery on it: the tax doesn't shrink by working harder, it shrinks by removing the reconstruction step.

To be fair about limits: no system eliminates context work entirely. Judgment calls, customer politics, and genuinely ambiguous situations still require a human holding the thread. What's removable is the mechanical reconstruction, and that's most of the bill.

If you want to see your own team's number, count one week of prep time, status rituals, and "quick question" pings, then look at what the accounts would answer for themselves.

FDE teams working in Claude Code pay a version of this tax with its own mechanics: the shared context problem for FDE teams.

Part of the tax is paid one unrecorded yes at a time. How Slack promises become undocumented scope traces that cost through four steps.

FAQs

What is the context tax?

The recurring, unbudgeted cost of reconstructing account knowledge that already exists across your tools: prep before calls, status extraction meetings, handoff rebuilds, and information hunting. Public research puts the general baseline at 1.8 hours per person per day, and implementation work multiplies it.

Why are implementation teams hit harder than other teams?

Three multipliers: many concurrent accounts (each switch is a 23-minute refocus), truth that lives in conversations rather than documents, and inherited context loss from sales handoffs. Forward deployed engineers report the extreme case, with hours of prep per customer meeting.

Does better documentation solve it?

It helps and it isn't sufficient. Documentation decays as projects move, writing it is itself a cost, and the highest-value context (what was said and promised on calls) rarely gets written at all. Our interviewees' phrase for the result: broken telephone with outdated documentation.

What actually reduces the context tax?

Removing the human from context transport: deriving account state automatically from calls, Slack, email, CRM, and tickets, with evidence links so it can be trusted. See what is a project brain for how that works.

How do I measure it on my team?

Track one week of three items per person: pre-meeting prep time, time in status-extraction rituals, and interruptions asking for account context. Multiply by loaded cost. Most 10-person teams find the equivalent of two or more full-time roles.

Key terms

  • Context tax: The recurring cost of reassembling scattered account knowledge before productive work can start.
  • Work about work: Coordination, status reporting, and information searching, measured by Asana at 58% of knowledge workers' time.
  • Refocus cost: The time to regain full focus after an interruption, about 23 minutes per UC Irvine research.
  • Handoff: Transfer of a customer between teams, the largest single point of context loss.
  • Status ritual: A recurring meeting whose main function is extracting project state from people's heads.
  • Account state: The live truth of a project: commitments, decisions, blockers, risk. What a project brain maintains automatically.