AI for B2B Marketing ROI: Valuing Pipeline Before Deals Close

Why B2B ROI looks wrong for months

In a business with a 90 to 180 day sales cycle, the money you spend on marketing in January shows up as revenue in April or later. A monthly ROI report built on closed revenue will therefore look terrible when you increase investment and brilliant when you cut it, which is exactly backwards.

The fix is to measure two things that are available now: the full cost of each stage of the funnel, and the expected value of the pipeline that spending has created. Then, as deals close, replace the expected value with real bookings and see how good the estimate was.

Step 1: Cost per stage, not just cost per lead

Cost per lead is the most quoted B2B metric and the least useful on its own, because leads vary enormously in quality. Carry the cost down the funnel.

• • Cost per lead (CPL) = total cost ÷ leads
• • Cost per qualified lead = total cost ÷ leads that meet your sales team's definition
• • Cost per opportunity = total cost ÷ qualified leads that became real sales opportunities

"Total cost" should include people and tools, not only media. AI-assisted marketing often shifts cost between those lines (less agency or staff time, more software), so a media-only CPL can improve while the true cost does not.

Calculator

Cost per lead calculator

Work out what a lead, a qualified lead and an opportunity really cost once people and tools are included.

Monthly media spend (£)Monthly people cost (£)Monthly tools cost (£)Leads per monthShare that are qualified (%)Share of qualified that become opportunities (%)

The starting figures are illustrative round numbers, not AfM prices or client results. Replace them with yours.

Total monthly cost

£5,900

Media plus people plus tools

Cost per lead

£98.33

Total cost divided by all leads

Cost per qualified lead

£245.83

Total cost divided by 24 qualified leads

Cost per opportunity

£819.44

Total cost divided by 7.2 opportunities

Assumptions

• Media-only cost per lead understates the true cost; people and tools are included here.
• Use the same qualification definition every month, or the trend means nothing.

Worked example

Illustrative example with hypothetical round figures. A B2B software company reviews one month of AI-assisted demand generation.

Inputs:

• • Media spend: £6,000
• • People cost (internal time plus managed service fee): £4,000
• • Tools: £500
• • Leads: 150
• • Qualified: 40%
• • Qualified leads that become opportunities: 25%

Total cost: £6,000 + £4,000 + £500 = £10,500.

Cost per lead: £10,500 ÷ 150 = £70.

Qualified leads: 150 × 40% = 60. Cost per qualified lead: £10,500 ÷ 60 = £175.

Opportunities: 60 × 25% = 15. Cost per opportunity: £10,500 ÷ 15 = £700.

Enter the same six inputs in the calculator above and it shows £10,500, £70.00, £175.00 and £700.00.

Step 2: Value the pipeline you cannot bank yet

Now estimate what those 15 opportunities are worth, using your own history, not an industry benchmark.

Assumptions from the company's CRM over the last two quarters:

• • Opportunity to closed-won rate: 20%
• • Average first-year contract value: £18,000
• • Gross margin on that contract: 70%

Expected gross profit per opportunity: 20% × £18,000 × 70% = £2,520.

Expected gross profit from the month's opportunities: 15 × £2,520 = £37,800.

Against the month's cost: £37,800 ÷ £10,500 = 3.6 times the cost, or an expected net of £37,800 minus £10,500 = £27,300.

The company can also state its ceiling: at £2,520 of expected gross profit per opportunity, it can afford up to £2,520 per opportunity before the activity loses money. At £700 it has room, which is a far more useful statement to a board than "CPL was £70".

Step 3: Track the lag and correct the estimate

Expected value is a forecast. Keep it honest:

1. • Tag every opportunity with the month and source that created it.
2. • Each month, record how many from earlier cohorts closed, lost or stalled.
3. • When a cohort is fully resolved, compare actual gross profit with the estimate from Step 2. If the estimate is consistently high, lower the win rate you use.

This is also where the ad platforms can help. Google Ads lets advertisers import offline conversions from a CRM, and recommends enhanced conversions for leads as the upgraded route; LinkedIn's Conversions API can attribute CRM outcomes back to LinkedIn campaigns. Feeding qualified and won stages back means the bidding systems optimise towards opportunities rather than raw form fills.

What a falling CPL does and does not tell you

AfM's case record for an anonymised global technology platform, 97% Lower Cost Per Lead Over About 11 Months, paid media Jakub ran from Q4 2024 to Q3 2025, mostly before founding AfM, shows why a CPL headline needs reading carefully. Cost per lead was "$270+" in the engagement's own first quarter, on LinkedIn only, and "$8.86" blended across Meta and LinkedIn in the last. Most of that fall came from moving budget to Meta: LinkedIn on its own still cost $201 a lead in the final quarter. The record also states that it holds no lead-quality or enrolment figures.

That is the right way to read any CPL improvement, including your own: it is an input-cost result, and part of it may be a change of channel mix rather than better performance within a channel. Whether it produced return depends on the qualified rate and the opportunity rate beneath it, which is why this article carries the cost all the way to the opportunity.

Where AI changes the B2B equation

In B2B, AI assistance tends to affect three parts of this funnel:

• • Cost per lead, through faster creative testing and audience work.
• • Qualified rate, through better targeting and clearer qualification at the form or first reply.
• • Speed from lead to first conversation, through automated enrichment and routing.

Measure each separately. A change that lowers CPL while lowering the qualified rate can raise cost per opportunity, and the calculator will show it immediately. For the profit side (CAC payback and margin), see AI Marketing Profitability B2B.

FAQ

Which win rate should I use if I have little CRM history?

Use the lowest rate you can defend from the deals you do have, state it as an assumption, and replace it as cohorts close. Avoid borrowing a benchmark from another company's funnel.

Should people cost include my own sales team's time?

For cost per opportunity, include the marketing and qualification effort needed to create the opportunity. Selling time after that point belongs in customer acquisition cost, covered in the profitability article.

Why use first-year contract value rather than lifetime value?

It keeps the estimate close to what you can observe within a year. Lifetime value depends on retention assumptions that take years to validate.

How long before the pipeline estimate can be checked?

Roughly one full sales cycle after the month in question. For a 120 day cycle, January's cohort can be judged properly from around May.