The competitor spreadsheet arrives, complete with company names, service descriptions and pricing notes. Yet the decision you were trying to make is no clearer. Should you introduce a fixed-fee offer, adjust your positioning or leave things as they are?
The problem may not be how carefully your VA gathered the facts. Your competitor research brief asked for information about a market, but not for evidence that could change a particular decision.
Instead of asking someone to research your competitors generally, give them a pending decision, a few answerable questions and clear rules for judging the return. This guide shows how to do that without passing your business judgment to the person collecting the evidence. If you are still choosing what kind of work to hand over, start with these research tasks suitable for a virtual assistant.
What a competitor research brief needs to do
Your brief should connect one pending business decision to the questions that could inform it. For each question, specify what evidence would count as an answer, how competing offers must be compared, which sources are allowed, how reliability and confidence should be reported, and what you will consider doing with each possible result.
That last instruction comes from you, not the VA. The assistant can collect and organize evidence. You decide what it means for your pricing, positioning or next move.
The principle applies beyond competitor research. An ISBA guide to working with market research agencies, endorsed by the Market Research Society, says research should begin with the business challenge it is meant to address. For a small delegated assignment, the practical version is a brief you can fit on one page, not an elaborate research programme.
Keep each brief focused on one decision. Continuous competitor monitoring is a different assignment because it needs an ongoing watchlist, update rules and a review rhythm.
Start with the decision, not the competitor list
“Research five competing agencies” gives your VA a topic but no test of relevance. The same five websites could yield hundreds of true facts. Only some would help you decide whether to change a service, price or offer.
Alan R. Andreasen’s approach to backward market research starts with how the results will be used and works back toward the information needed. Apply that logic by writing down the decision before listing websites to visit.
For example, replace “Look at other bookkeeping firms” with “Before we decide whether to test a fixed monthly bookkeeping package for small businesses, find out which comparable local firms publish monthly prices, what those prices include and which conditions could make the offers unlike ours.”
Three useful kinds of competitor questions
Competitive-intelligence researcher Jan P. Herring’s Key Intelligence Topics approach distinguishes strategic decisions, early warnings and key players. You can adapt those categories to a smaller brief:
- Decision question: Which comparable providers publish a fixed monthly package for the customer segment we are considering?
- Early-warning question: Have any named competitors publicly announced a new package, geographic expansion or pricing-model change that affects our planned offer?
- Key-player question: Which two providers serve the closest customer segment, and what verifiable features make them comparable to us?
These are question types, not three mandatory sections. Pick only the ones that could affect the decision in front of you.
Write your decision rules before anyone collects anything
Before the VA starts, complete two sentences: “If the evidence shows X, we will consider Y,” and “If it shows Z, we will hold, seek more evidence or take a different route.” Set any threshold yourself; do not ask the assistant to invent one after seeing the results.
For a fictional bookkeeping practice, the owner might write: “If at least three of four named comparable firms have verifiable public monthly packages with clearly described scope, we will consider testing our own package. If fewer than three can be verified, competitor websites alone will not justify the test.”
That is an illustrative business rule, not a recommended industry threshold. It triggers consideration, not an automatic launch. The owner still needs to weigh customer demand, margins and delivery capacity. If important answers are missing, “not verified” must not quietly become “no.”
This is related to keeping decisions with the owner, but the rule belongs at the start of the research, before the decision reaches an approval queue.
Turn each question into an evidence plan
Specify the evidence that could answer it
Good evidence is observable and tied to the wording of the question. To establish that a firm publishes a monthly package, look for a current public pricing or service page showing the package and its terms. A search snippet, someone’s opinion or an AI summary is not the same evidence.
Be precise about what the research can establish. Published starting prices do not tell you what every client actually pays. An advertised feature does not prove the competitor delivers it consistently. If you need customer experience or negotiated contract data, you are asking a different question that may require a different research method.
Compare like with like
Choose the dimensions before the VA collects examples. For a bookkeeping offer, they might include the location served, target customer size, pricing period, currency and VAT treatment, included bookkeeping tasks, exclusions, onboarding charges and the date the page was checked.
Without those boundaries, an annual package including payroll might appear cheaper or more expensive than a monthly bookkeeping-only offer for reasons unrelated to your decision. The same normalization principle matters when comparing vendor offers and their inclusions, although vendors you might hire are a different research subject from competitors you sell against.
Ask the VA to mark genuinely non-comparable offers as such instead of squeezing them into a ranking.
Name permitted sources and collection limits
Approved starting points can include official competitor websites, publicly accessible price lists, published annual reports, press releases and job advertisements. Public reviews may reveal customer perceptions, but individual reviews do not establish overall service quality. Job listings may suggest investment in a capability; they do not prove that capability is already delivered.
Tell the VA where to stop and ask. If the answer depends on an account login, restricted document, direct interview or paid source that has not been approved, it belongs in the unresolved-questions section, not in a workaround.
For the basic source URL, date checked and exception fields, use the existing research-and-data brief guidance rather than creating a second generic tracking sheet here.
Ask for confidence, not just answers
Two assessments should accompany an important finding: How dependable is the source? and How well does the available evidence support this particular claim? Those are different questions.
This distinction borrows a two-axis principle from intelligence-analysis practice. The US Intelligence Community’s ICD 203 analytic standards likewise require analysts to explain the evidence behind judgments and distinguish confidence from likelihood. The labels below are a simplified working vocabulary for a small-business brief, not an official intelligence grading standard.
Source reliability
- Established: An identifiable primary source, such as a current official competitor page, for the specific information it publishes.
- Mixed record: The source is identifiable, but its date, completeness or consistency raises a question.
- Unknown: The original source or its authenticity cannot be established.
Claim confidence
- Corroborated: Relevant, independently checked sources agree on the precise claim.
- Single-source: One directly checked source supports the claim; further confirmation is unavailable or unnecessary for the narrow question.
- Unconfirmed or conflicting: Essential details are missing, outdated or contradicted.
An official competitor page may be an established source for the statement “this price is advertised.” That same page cannot establish “most customers pay this price,” even if it looks professional. Confidence attaches to a precisely worded claim, not a source’s reputation alone.
Do not turn “single-source” into a made-up probability or treat “corroborated” as a guarantee. Where an estimate is genuinely necessary, keep the estimated likelihood separate from confidence in its supporting evidence.
AI-generated research deserves the same checks. A 2026 preprint examining citation links from commercial language models and deep-research agents found fabricated and non-resolving URLs among the systems it tested. The rates are specific to that study, not a universal error rate for business research. Require the VA to open the original source and check that it supports the claim before assigning confidence. For broader AI-use boundaries, see when not to use AI for business tasks.
One question, traced from decision to decision rule
Fictional example: A small UK bookkeeping practice is considering a fixed-fee monthly package for businesses with fewer than ten staff. The owner identifies four local practices believed to serve a similar audience. These practices and findings are invented solely to show how the brief works; they are not market research or Boost VA client results.
Decision: Should the practice explore a limited fixed-fee package test?
Question: Which of the four named practices publicly offer monthly bookkeeping packages relevant to that customer segment, with enough detail to compare what is included?
Evidence: Current public pricing and service pages, including linked terms where available. The VA must record the exact wording rather than assume that “from £X” means an all-inclusive service.
Comparison basis: Monthly price or starting price, VAT status where stated, target customer, bookkeeping scope, payroll and tax-work exclusions, set-up fees, and date checked.
Permitted collection: Public pages only. No false enquiries, staff approaches or private-access workarounds.
Possible return: Imagine Practice One and Practice Two each publish monthly packages with described scope; Practice Three asks visitors to request a quote; Practice Four has an outdated page that the VA cannot verify. The two published offers are recorded as single-source findings from established official pages. Practice Three’s unpublished pricing and Practice Four’s current offer remain unconfirmed. None of those gaps counts automatically as evidence that a package does not exist.
Owner’s pre-written rule: If at least three of the four have verifiable public packages with comparable scope, consider a limited test. Otherwise, do not use those public pages alone as the reason to proceed.
Decision at review: The hypothetical return has not met that evidence threshold. The owner can pause, ask for a narrower follow-up or use another approved method to understand customer demand. The VA reports what was and was not established; the owner decides whether to act.
Notice the difference: the research did not simply list four competitors. Every part of the answer has a job, and uncertainty remains visible rather than being hidden in an empty cell.
What your VA should not do to get an answer
Competitor research should use lawful methods, truthful identification where contact is authorized, and sources the business is entitled to access. Publicly visible does not automatically mean unrestricted collection or reuse.
The Strategic Consortium of Intelligence Professionals (SCIP) code of ethics emphasizes lawful conduct and accurate disclosure of identity and organization in interviews. Its ethics guidance also cautions against fake job applications, unauthorized competitor logins and misrepresentation.
Generally acceptable within an agreed scope:
- Read public competitor pages and record specific published statements.
- Review public filings, dated announcements and lawfully accessible information.
- Report conflicting, missing or outdated claims without filling the gaps by guesswork.
Off limits for this assignment:
- Obtain or use passwords, restricted accounts, leaked records or confidential documents without authorization.
- Pretend to be a customer, applicant or another person to extract non-public information.
- Ask current or former staff to reveal confidential information.
- Scrape or reuse information in ways that violate applicable law, site terms or agreed permissions.
- Collect personal details about named individuals without checking relevant privacy obligations.
Legal boundaries vary by jurisdiction. For example, US trade-secret law defines improper means to include conduct such as misrepresentation, while the UK Trade Secrets (Enforcement, etc.) Regulations 2018 address unlawful acquisition, use or disclosure involving breach of confidence. These examples illustrate why research access and methods matter; they are not a complete legal checklist or legal advice.
If a promising lead requires a questionable method, the VA should stop and ask, not improvise.
When competitor research is not worth commissioning
Sometimes the best competitor research brief is no brief at all. You can save effort by checking whether the answer could realistically alter your next action.
Consider postponing or changing the assignment when:
- No decision is pending. Curiosity alone may not justify a scoped one-off comparison.
- The answer changes nothing. If every plausible result leads to the same choice, the question is not decision-critical.
- The choice is low-stakes and easily reversible. Testing a small change may teach you more than researching competitors in depth, depending on the risks.
- The required evidence is not available through permitted public sources. Customer interviews, statistical market sizing, confidential records, specialist financial or legal analysis may call for different expertise and explicit authorization.
- You really need ongoing monitoring. Recurring alerts and change logs need their own scope and cadence, not a one-time decision brief stretched indefinitely.
Research is worthwhile when it can reduce an uncertainty that matters to the business decision. It should not become a standing request to collect facts nobody has agreed to use.
The decision-first competitor research brief: copy and adapt
Paste the following into a document or task description. Keep it short enough for the owner to approve before work begins, but specific enough for the VA to know when an answer is complete or uncertain.
1. Decision and owner
Decision to inform: [What are we considering?]
Decision-maker: [Name/role]
Decision needed by: [Date]
2. Why now
Trigger: [New offer, price review, market entry or other immediate reason]
3. Competitors in scope
Include: [Named businesses or clear selection criteria]
Exclude: [Geographies, segments or business types that would mislead the comparison]
4. Questions to answer
For each question, copy and complete this block:
- Question and type: [Decision / early warning / key player]
- Evidence that would answer it: [Specific observable public facts]
- Comparison basis: [Scope, unit, geography, period and other necessary dimensions]
- Permitted sources: [Official pages and any approved alternatives]
- Source reliability to report: [Established / mixed record / unknown]
- Claim confidence to report: [Corroborated / single-source / unconfirmed or conflicting]
- Owner’s rule, written now: If [finding X], we will consider [action Y]. If [finding Z or material uncertainty], we will [hold, seek more evidence or review another option].
5. Collection boundaries
Allowed: [Public, lawful source types and approved methods]
Off limits: [Misrepresentation, unauthorized access, confidential records, prohibited collection, unapproved personal-data use and unchecked AI summaries]
6. Return format
For each question: one direct answer or “not established”; supporting source links and dates checked; a like-for-like comparison; both ratings with a short reason; what could not be confirmed. Keep source facts distinct from assumptions.
7. Time box and stop rules
Effort limit or deadline: [Agreed scope]
Stop and ask when: [Essential sources conflict, paid or private access is needed, ethics are unclear, or the question cannot be answered as defined]
8. Owner review
Compare each answer with the decision rule set above. Open a sample of the supporting pages, inspect unresolved findings and record the decision yourself. For broader handoff details, use the guide to clear VA instructions.
The key is to ask fewer questions that have consequences, rather than more questions that generate rows. The VA can return careful, comparable evidence and explain what is still unknown. Your choice remains yours.
If you need help with a scoped competitor research assignment, send me your decision and draft questions. We can clarify what evidence is realistically available and what the final research handoff should contain before collection begins.


