How do you stop starting desk research from scratch on every engagement?

Desk research template with a question frame, source map and figure register for consulting engagements

Build a desk research template from three assets that don't change between sectors: a question frame per recurring engagement type, a source map ranked by source type, and a figure register recording each number's basis and date. Then only the sector facts are new, and a first pass of those can be generated in minutes.

Why does desk research restart from zero on every engagement?

Because most teams keep the output and discard the scaffolding. The deck is filed; the questions asked, the sources tried and rejected, the definition chosen and the reason a number was dropped all live in browser tabs that close when the engagement does. The next project, in a different sector, inherits a deck it cannot use and none of the reasoning it could.

The search is the expensive part. Atlassian's State of Teams 2025 survey of 12,000 knowledge workers and 200 executives found that leaders and teams spend 25% of their time searching for answers; half the workers said their teams had unknowingly worked on the same things. If a US management analyst at the May 2025 median wage of $101,860 spent that share of the year searching, it would cost about $25,500 per person in salary alone. This means starting from scratch is a payroll line, not an inconvenience.

25%

of their time leaders and teams spend searching for answers, in a survey of 12,000 knowledge workers and 200 executivesAtlassian, State of Teams 2025

The usual remedy, a library of past decks, fails because a deck is the most sector-specific and fastest-aging thing an engagement produces. Last year's logistics market slide is no use on this year's healthcare project; the method behind it is. Reuse has to happen one level below the deck.

When four hours of searching yield three usable numbers, the problem is not search skill. Nothing from the last engagement was stored in a form the next one could use.

What carries over from one engagement to the next?

Sort what an engagement produces by two tests: does it survive a change of sector, and how fast does it go out of date. Most of the output fails at least one. Three things pass both.

What an engagement produces, and how much of it the next one can use
AssetSurvives a new sector?Shelf lifeStore it as
Sector facts: size, shares, growthNoMonthsRegister rows with basis and date
Question frame per engagement typeYesYearsA one-page issue tree with stop rules
Source map, ranked by typeTypes yes; titles noYearsTypes in order of trust, titles beneath
Definitions and basesThe discipline, not the definitionsUntil the classification changesScope, unit, period and geography per figure
Reconciliations already doneWithin the sectorUntil a source updatesBoth figures, the verdict, the reason
Slide formatsYesYearsTemplates
Client data and conclusionsNo: they are the client'sNot applicableThe engagement archive only

This means three of the seven are worth building once and maintaining: the question frame, the source map and the figure register. Together they are the desk research template. Sector facts are still gathered each time, but they land in a structure that already exists. A slide master is not the template: it saves the hour of formatting, not the days of searching before it.

How do you build a question frame that works in any sector?

Most advice on how to do desk research stops at where to search. The reusable part is what to ask, written per engagement shape rather than per sector. Engagements recur in a handful of shapes, and each shape asks the same questions in every industry. The sector overview, behind most research-heavy slides in a strategy deck, runs in six steps, and the order matters.

  1. Define the market before sizing it: scope, unit, period and geography, set by the decision in the brief rather than by the first report found.
  2. Size it from two or three published estimates, each with its basis, and report a range. Stop when you can say why the estimates differ.
  3. Map the structure: how much the top tier holds, how long the tail is, where the profit sits. Stop when you can name who has pricing power.
  4. Name the demand that is not discretionary: regulation, standards, contracts. Stop when each driver has a rule or a date attached.
  5. Name who is buying, building or leaving: active acquirers, entrants and their pace. Stop at the handful that set prices.
  6. Write the so-what for this client: what the market means for the decision in the brief, in one sentence a partner can put on a slide.

The stop rules are where the time comes back. Without them, each branch expands to fill the days available; with them, the frame says when a question is answered well enough for the decision at hand.

One frame, five engagement shapes: what each reuses and what is new
Engagement shapeSteps it reusesNew every time
Market sizingDefinition, sizing, register rowsThe client's own segment cut
Competitive landscapeStructure, players, deal recordsThe client's position against the players
Sector overview for a strategy deckAll sixThe so-what for the client's decision
Market entryDefinition, sizing, demand drivers, playersEntry mode and the client's right to win
Acquisition or roll-up screenStructure, demand drivers, players, deal recordsTarget criteria and what the client can pay

The players module is the one most often run on its own, and the method to build a competitive landscape quickly slots into step five unchanged.

The same logic answers the agency problem of pitches in unrelated industries. A beverage pitch and a banking pitch share no facts but the same frame: category size and growth, who holds share, what the regulator or the retailer forces. What carries forward is the frame, never the facts.

Which sources belong on a reusable source map?

Source titles change with the sector; source types, and the order of trust between them, do not. Caspr's UK facilities management analysis ranked three tiers: government and regulator publications, then transaction trackers and company disclosures, then commercial commentary. Where figures conflicted, it favored the most recent source with the clearest method.

A source map by type, with the US entry point for each
TierSource typeUS entry pointAnswers firstWatch for
1Official statisticsCensus Bureau data by NAICS code; Bureau of Labor StatisticsEstablishments, employment, payrollWhere the code draws its boundary
1RegulatorsFederal Register; agency rule pagesSpend customers cannot deferProposed rules read as final
2Company filingsSEC EDGAR: 10-K, 10-Q, 8-KSegment revenue, margins, risksAdjusted figures against reported
2Deal recordsAcquirers' announcements; advisory firms' deal trackersBuyers, pace, multiplesEach tracker's own scope
3Industry bodiesTrade associations' member surveysPractice, sentiment, playersMember samples; advocacy
3Syndicated reportsPublishers' summariesHeadline size and growthAn unstated basis: directional only

The boundary problem transfers as reliably as the ranking. The UK analysis treated the official facilities management category, SIC 81.00, as one visible slice of the market. A US analyst meets the same trap in NAICS 561210, Facilities Support Services: it covers establishments providing a combination of support services inside a client's facilities, while one providing a single service, such as janitorial work, is classified in that service's own industry. A count built on 561210 alone leaves out much of what a client calls facilities management.

What changes from sector to sector is the list of titles under each tier. Finding market data you can cite is then done once per sector, and the second engagement in that sector starts from a list rather than a search bar.

What goes in a figure register, and why does it outlive the deck?

A number stored without its basis cannot be reused; it can only be found again. The register keeps each figure with what it counts: value, unit, period, scope, geography, publisher, publication date, the date it was last checked, and a status of current, superseded or disputed. Five rows from the UK facilities management analysis show the form.

A figure register: five rows from Caspr's UK facilities management analysis (data as of August 25, 2026)
FigurePeriodWhat it countsPublisherRegister status
$83.29 billion2026Total UK FM market, in-house and outsourcedMordor IntelligenceCurrent; total-market basis
Over £35 billion2024Outsourced FM onlyFrost & Sullivan, cited by CBRECurrent; the roll-up basis
37.3% and 68.0%Year not statedTurnover share of the top 5 and top 20, among 151 companiesThe AccountsDirectional: groups file through several entities
181 against 1182025 against 2024Completed UK FM dealsMoore Kingston Smith, June 1, 2026Current; one tracker's scope
7.5x to 12.5x EBITDADisclosed dealsUK testing, inspection and certification multiplesCrowe UKCurrent; disclosed deals only

The first two rows look like a disagreement and are not one. Mordor Intelligence's $83.29 billion counts the whole market, in-house and outsourced; Frost & Sullivan's figure of over £35 billion counts outsourced contracts only, and a roll-up buys service providers, not in-house estates teams. This means the next UK facilities management engagement inherits the decision about which figure answers which question, instead of arguing it again.

Recording that verdict, definitional or genuine, is what turns a one-off effort to reconcile two market size estimates into an asset. The reconciliation is done once per pair of sources and redone only when one of them publishes a new edition.

Status is the column most teams skip and most need. Without it, a register slowly fills with numbers nobody remembers retiring.

Where should the hours you save go?

Into the part no template covers: what the market means for this client's decision. The frame says what to ask and the register says what is already known. The so-what, which option the client should take and what would change that answer, is still the analyst's work, and it is the part the client pays for.

The first pass of sector facts is now the cheapest part of the job. Caspr, an analytical AI, turns a single-line prompt into a structured, cited report in under 15 minutes. Put the frame's questions in the prompt and the first pass answers them, every figure cited and ready to enter in the register. The analyst's hours move from compiling to judging.

Research agencies carry the same overhead earlier in the project: the desk research that must precede a screener or a discussion guide. When that work is absorbed into project cost rather than billed, every hour saved is margin. Teams running several projects at once gain most from one shared register, because two projects touching the same market should never pay for the same number twice.

The first step is small. Pick the two engagement shapes the team runs most often, write their frames with stop rules, and start the register on the current engagement. The second engagement is where it pays.

Worked example

UK facilities management: one frame, seven verticals

Caspr's analysis of roll-up opportunities in UK facilities management (data as of August 25, 2026) is a desk research template run seven times in one document. It fixed the bases first: a total UK FM market of $83.29 billion in 2026 (Mordor Intelligence), an outsourced segment of over £35 billion in 2024 (Frost & Sullivan, cited by CBRE), and the official SIC 81.00 category as one visible slice. It ranked its sources in three tiers, official and regulatory first. Then it applied one investor lens of seven criteria, from fragmentation and recurring revenue to labor scarcity and integration feasibility, to seven verticals, from fire and security to soft FM. The same questions returned different answers. Fire and security led: 56 deals in 2025 against 42 in 2024, 70% of them by PE-backed buyers (Grant Thornton), with post-Grenfell fire safety law supporting inspection and maintenance demand. Testing, inspection and certification had the deepest runway, with more than 280 UK transactions in 2025 and disclosed multiples of 7.5x to 12.5x EBITDA (Crowe). Volume cleaning ranked lower: a 4.1% rise in the National Living Wage in April 2026 lands on a cost line dominated by wages. The verdict: the best roll-up conditions sit in regional hard FM and specialist compliance businesses, not in broad soft FM. Reusing the frame took no knowledge of the sector; answering it took all of it.

Frequently asked

What should a desk research template include?

A desk research template has three parts: an issue-tree question frame for each recurring engagement shape, with a stop rule on every branch; a source map ranking source types, with each sector's titles beneath; and a figure register logging every number with its unit, period, scope, publisher and date. Slide formats help but save the least time.

Can you reuse research across clients in different industries?

You can reuse the method across clients in different industries, but never the findings. Question frames, the source ranking and the register's discipline move across industries unchanged; public-source figures move within a sector, dates attached. What one client paid for, its data, interviews and conclusions, stays with that engagement under whatever confidentiality terms the engagement letter sets.

How do you speed up the desk research phase of a research project?

You speed up desk research by starting from the question frame for the project's shape, not a blank page. Pull the register's current rows for the sector, refresh only rows whose sources have published since, and generate a first pass for the gaps. The saving comes from not deciding scope, sources and definitions again on every project.

How do you manage desk research across several projects at once?

Use one register for the whole team's desk research, keyed by sector and figure, so two projects touching the same market draw on the same rows. Keep one frame per project shape and one named owner for each sector's source list. Retire superseded rows as new editions appear, so no project quotes a number another has replaced.

Keep reading