How Much Research Per Prospect Is Actually Worth Doing

About two minutes of research per $1,000 of average deal value pays for itself. On a $12,000 average deal at a $60/hour fully loaded cost, roughly 26 minutes per prospect breaks even. Below about $2,500 average deal value the sum permits barely five minutes, which is under the human floor.

At a glance

• The metric: research minutes per prospect — total minutes spent gathering and writing prospect-specific detail, divided by prospects contacted in the same period.
• The break-even line: roughly 2 minutes per $1,000 of average deal value, at a $60/hour fully loaded researcher cost.
• The crossover: under ~$2,500 average deal value the sum permits about five minutes a prospect — less than it takes a human to open the record, read it and write a line, which is the practical floor. Over ~$50,000 the constraint stops being money and becomes coverage — one researcher covers about 17 prospects a week.
• Why it adds up: research touches three sequential links (reply rate, reply-to-meeting, meeting-to-close). Modest gains on each compound.
• The honest finding: personalisation does not scale by hand. The table below shows the exact deal value where it stops.

How do I measure research per prospect?

Almost nobody trying to increase the data they hold per prospect ever names the metric, which is why the argument between personalisation and volume never resolves. Measure two things over one ordinary week:

• Research minutes per prospect = total minutes spent finding and writing account- or person-specific detail ÷ prospects contacted in that week. Count only research that ends up inside a message or a call plan. Research that never reaches the prospect is pure cost.
• Contribution per contacted prospect = average deal value × gross margin × the contacted-to-closed rate. This is the number research is trying to move, and it is the only sane denominator for the time you spend.

The quotable version: research per prospect is not a virtue, it is a purchase — you are buying conversion rate with minutes, and every purchase has a price at which it stops being worth making. If your team cannot produce a research-minutes-per-prospect figure this week, that is the first thing to fix, not the research volume.

How much research per prospect pays for itself?

Break-even research minutes = (average deal value × gross margin × contacted-to-closed rate × the relative uplift research produces) ÷ fully loaded cost per minute. The table below runs that formula at 60% gross margin, a 0.495% contacted-to-closed rate and a 72% relative uplift (both derived in the next section, from 6% reply → 8%, 33% reply-to-meeting → 38%, 25% meeting-to-close → 28%), and a fully loaded researcher cost of $60/hour — salary, on-costs and tooling, so $1 a minute. The final column assumes 1,800 productive minutes in a researcher’s week. Those four assumptions are ours, not measurements of your business: the arithmetic is currency-neutral, so substitute your own inputs.

The two-minutes-per-thousand rule. At those inputs the break-even lands at 2.14 minutes of research per $1,000 of average deal value, which rounds to a rule worth remembering: two minutes per $1,000. It is a decision rule, not a law of nature: halve your gross margin and it becomes one minute per $1,000; double your contacted-to-closed rate and it becomes four. Recompute it with your own numbers once a quarter and write the answer on the wall, because the argument about how much research to do is really an argument about a number nobody has calculated.

Where does the 72% uplift number come from?

It is a worked example, not a measured result — the inputs below are illustrative and you should replace them with your own. Take 100 contacted prospects at a $12,000 average deal and 60% gross margin, so $7,200 of contribution per closed deal:

• Baseline: 6% reply, 33% of replies book a meeting, 25% of meetings close. 0.06 × 0.33 × 0.25 = 0.495% contacted-to-closed, worth $35.64 per contacted prospect.
• With research: reply 6% → 8% (×1.333), reply-to-meeting 33% → 38% (×1.152), meeting-to-close 25% → 28% (×1.12). Compounded: 1.333 × 1.152 × 1.12 = 1.72. Contacted-to-closed becomes 0.851%, worth $61.29.
• Gain: $25.65 per contacted prospect — 25.65 minutes at $1 a minute.

This is how a large multiple appears without any single heroic number: three unremarkable improvements on sequential links produced a 72% lift. It is also where the arithmetic misleads people, so say the wrinkle out loud: these three gains overlap and do not truly multiply. The same specific detail that earns the reply is part of why the meeting converts, so multiplying the links double-counts it. We use deliberately modest per-link figures for that reason. Our fuller treatment of why small conversion gains compound, and where the multiplication breaks down works through the overlap in detail, and the cost of each sales pipeline stage is where you find the per-stage inputs this calculation needs.

Which research actually moves the number, ranked by effect size?

Not all data per prospect is equal, and the ranking that gets published most often is upside down. The first three levers change the argument you make; the fourth only changes the opening line.

The amortisation column is the whole point. A B2B purchase is decided by a group, so account-level research divides its cost across every person you contact there, while person-level colour is paid for once per person and thrown away. That is why the cheapest way to raise data per prospect is usually to research fewer accounts more deeply rather than more people more shallowly.

It also matters that the buyer has already done their own homework. 6sense’s 2025 B2B Buyer Experience Report, a survey of more than 4,000 buyers published on 12 November 2025, found the journey has moved to roughly a 60/40 split between independent research and seller engagement, with 94% of buying groups ranking a preferred vendor before first contact and buying from that favourite 77% of the time. Research aimed at flattering the reader is aimed at the wrong target; research that demonstrates you already understand their operation is what gets you considered before the shortlist closes.

“Personalisation doesn’t scale” — correct, by hand

The objection is right, and the table above says exactly where it becomes right. A full-time researcher has roughly 1,800 productive minutes a week. At a $12,000 average deal the break-even is 26 minutes, so that researcher covers about 70 prospects a week. If your plan requires 500 contacts a week at that depth, you need eight researchers, and eight researchers is a payroll line that will not survive its first quiet quarter.

Two things break before the money does. Consistency goes first: research quality varies more between two humans on the same brief than between two weeks of the same human, and blended reply rates hide it. Storage goes second — research that is not written back onto the record is re-done by the next person who touches the account, which is one of the more expensive symptoms of poor CRM data hygiene. Both are solvable, but they are solved with systems, not with effort.

What should I change first?

1. Measure for one week. Research minutes per prospect, and the contacted-to-closed rate. Do not change anything yet.
2. Draw your break-even line with your own margin and close rate. Two minutes per $1,000 is the starting point, not the answer.
3. Compare. If you are researching well above the line, you are buying conversion rate at a loss and the fix is fewer, better-chosen accounts. If you are well below it, you have unspent budget and the fix is lever 1, not lever 4.
4. Move to account-level first. Every minute that amortises across the buying group is worth several that do not.
5. Re-measure after four weeks at the link level, not blended. A blended rate cannot tell you which of the three links moved.

This page sits inside our wider pipeline-stage cluster, which carries the full stage-by-stage conversion matrix these inputs come from.

What does running this yourself actually cost?

The method above is complete and you can run it. Here is the bill, honestly. You need a trigger source and a firmographic source, and someone who can read a company announcement and tell whether it implies a problem you solve — that skill is rarer than the tooling and it is the part that cannot be bought cheaply. You need the research written back onto the record so it is not repeated. You need weekly QA on a sample, because unmeasured research quality drifts downward. And you need to accept that this competes with selling time: Salesforce’s State of Sales research reports that sales reps spend 60% of their time on non-selling tasks, and adding research to a rep’s job without removing something else moves that number the wrong way.

At low deal values the honest conclusion from the table is that no human should be doing this at all, which is the case for AI-run outbound that gathers and uses prospect data at volume: not because it is better than a good researcher, but because the break-even line is below what a good researcher can physically charge for. Do that arithmetic with your own numbers before you decide either way.

Frequently asked questions

How much research should I do before a sales call?

Use the same break-even line: roughly two minutes per $1,000 of average deal value at a $60/hour fully loaded cost. A booked call is worth far more than a cold contact, so in practice most teams underspend here — on a $25,000 deal the sum permits close to an hour, and almost nobody spends it.

Does personalised outreach actually get more replies?

Directionally yes, but be careful which numbers you repeat. Instantly’s Cold Email Benchmark Report 2026, drawn from cold email activity across thousands of workspaces between 1 January and 18 December 2025, puts the overall average reply rate at 3.43% of emails sent, with top performers above 10% — but that report does not break reply rate down by personalisation depth. The widely circulated “personalisation doubles replies” figures we checked traced back to vendor restatements rather than a published dataset, so we have not used them here. Measure it on your own sends.

Is it better to research fewer prospects deeply or more prospects lightly?

Deep on accounts, light on people. Account-level research amortises across everyone you contact at that account; person-level colour does not. At a $12,000 average deal, 26 minutes spent once on an account of five contacts is about 5 minutes per prospect, while 26 minutes on one person is 26.

How do I know if my research time is being wasted?

Check what share of researched facts actually appear in a sent message or call plan. Anything below about 70% means you are paying for research that never reaches a prospect, and the fix is the handoff, not the research.

Who should own prospect research — the rep or a separate researcher?

Split it by amortisation. Account-level research belongs to a researcher or a system because it is reused; role-level research belongs to whoever is making the call, because they need it in their head, not in a field. Whichever you choose, remove an equivalent amount of work from the rep, or the research simply does not happen.