ABM without the six-week research sprint

A target-account list can look promising on Monday and be untouched six weeks later. The team is still collecting annual reports, checking hiring pages and assembling slides for its top accounts. By the time the briefs are ready, there is barely enough capacity left to act on them.

I would use AI to shorten that research step, then spend the saved time on more accounts. That is the opportunity for ABM: not more polished dossiers for the same handful of companies, but better coverage of accounts that fit your market and currently receive little relevant attention. Research can get faster without qualified pipeline improving, so the set-up has to start with the pipeline gap.

Start ABM with the accounts you cannot currently cover

First, I would take the existing target-account list and ask a fairly uncomfortable question: how many accounts meet our ideal customer profile but have no active, account-specific plan? Exclude customers, open opportunities and accounts a seller is already working properly. What remains is the coverage gap.

Then I would narrow the pilot to one segment where the commercial case is clear. It might be accounts above a defined deal-size threshold in a territory with spare sales capacity, or an expansion segment where the team knows the product use case well. Sales and marketing should agree the entry criteria before generating a single brief. Otherwise AI will make it wonderfully quick to research companies nobody intends to pursue.

I would keep bespoke, one-to-one work for the few accounts that warrant it. The pilot should test whether account-specific research can improve a one-to-few programme: a shared play for companies with a similar problem, informed by evidence about each company. In 6sense’s April 2026 survey of 427 B2B marketing and GTM professionals, 58% of respondents using ABM reported a one-to-few approach. That describes how those respondents work, not proof that one-to-few performs best, but it is a useful reminder that personal treatment need not mean a separate campaign for every logo.

Build a brief a seller can challenge in five minutes

Next, I would give the research process a strict output: a short account brief, not a presentation. It should contain what the company does, a recent and verifiable business change if one exists, the likely relevance of our offer, what we already know from the CRM, and the one thing a seller needs to check before outreach. Every factual claim needs a dated source. The proposed angle should be labelled as an inference, not passed off as something the buyer said.

AI is suited to assembling a first draft from scattered material. OpenAI’s guidance on deep research explicitly lists account research among sales use cases and advises users to inspect citations and validate key claims. That is a capability and a caution, not evidence that a finished brief will always take minutes. I would time the whole process, including review, rather than just the generation step.

The inputs should be limited to approved public sources and internal records the team is authorised to use. Start with company news, product pages, published reports and the account’s CRM history. Skip speculative personality profiles and supposed buying intent inferred from one job advert. If a posting mentions a new data role, that may justify a question about data ownership. It does not establish a funded project, a buying committee or a deadline.

Give one marketer responsibility for checking the evidence and one seller responsibility for testing commercial relevance. A brief can be factually sound and still useless if its central observation has no connection to a problem you solve. In Forrester’s September 2026 account of its B2B marketing research, 50% of marketing decision-makers said AI often presents incorrect information with unwarranted confidence. Verification time belongs in the economics of this programme, not in a footnote.

Turn account research into a reason to make contact

Once the brief survives review, I would choose one existing play that fits the account’s situation. A relevant customer story, an invitation to a small discussion or a useful comparison may be enough. Only adapt the part that genuinely depends on the account research: the opening observation, the question for the buyer or the choice of proof. There is no need to commission a custom report for every account.

For example, suppose a public product announcement raises questions about how a company will manage a larger customer base. If your product genuinely addresses that operational issue, a seller could use the announcement to ask how the team plans to handle the change, and offer a relevant example. That is a proposed approach, not a claim that the company has the problem. If the announcement has no real bearing on what you sell, leave it out.

I would keep the first run deliberately small: one segment, one play, a named sales owner for every account and a fixed weekly review. Each account gets one of three outcomes: pursue now, nurture with a reason, or remove from the list. No owner means no activation. Research without a route into a real conversation is just a cheaper pile of PDFs.

What usually goes wrong is scaling the output before checking the response. Teams generate hundreds of briefs, turn them into superficially personalised emails and report that they have ‘covered’ the market. But mentioning a funding round or a new office is not personal treatment if the message would make equal sense to fifty other companies.

Measure whether wider coverage produces qualified pipeline

Before the pilot starts, record the number of eligible, previously uncovered accounts, the cost of researching and reviewing each one, and the baseline outcomes for a comparable group. The first measure is coverage: how many of those accounts now have a reviewed, relevant play that sales actually uses? The next measures are conversations with the right people, qualified opportunities accepted by sales and cost per qualified opportunity. Follow opportunity progression and win rates as they mature; do not declare a revenue win because someone replied.

For a cleaner comparison, I would hold back a similar set of accounts under the existing process, or compare cohorts with the differences stated plainly if a holdout is impractical. Count the research time, seller review and channel spend. If the new approach creates more activity but the same number of qualified opportunities at higher total cost, it has not improved pipeline coverage in a commercially useful way.

My bet is that the constraint will move. Once a brief is quick to produce, the scarce resource becomes the team’s ability to spot a real reason to engage and follow through. If the six-week research sprint disappears but ABM still reaches the same twenty accounts, all you have bought is a faster way to maintain the old ceiling.

This blog was created by Jan’s AI content agent. Get in touch if you want to know how it works.