Pipeline, not just traffic

Benchmarks put early-stage B2B SaaS around 25-40% of revenue and mature companies nearer 10-15%, but the constraint that actually binds is CAC payback. If payback sits under about 12 months and you are growing at or above your segment, the spend is defensible regardless of the percentage. We plan budget from payback and pipeline coverage rather than from a revenue ratio.

Lead generation counts form fills. Demand generation creates the intent that makes a form fill happen at all, then captures it when it surfaces. A gated ebook converts someone who was already looking; a category argument, a comparison page or a product-led trial creates the looking. Teams that only run capture see MQLs rise while pipeline stays flat, which is the single most common pattern we are called in to fix.

Contract value decides it more than preference. Under roughly $5K ACV, self-serve and product-led motions carry the economics because a sales conversation costs more than the deal returns. Above that, a sales-assisted motion usually pays for itself. Most companies we work with run both: product-led for the long tail, sales-assisted for the accounts worth pursuing, with marketing feeding different signals to each.

By writing for each seat rather than an average. The champion needs proof it works, the economic buyer needs a business case, and security or procurement needs the questions answered before they ask. One page pitched at the average persuades nobody. We build the evidence each role needs and put it where that role looks, which is usually your docs, your comparison pages and your search results, not a nurture sequence.

CAC payback, pipeline coverage, win rate by source, and net revenue retention. Those four tell you whether acquisition is affordable, whether the quarter is covered, whether the leads convert, and whether the product keeps what it wins. MQL counts and traffic belong under them as diagnostics, never as the headline.

Paid capture produces pipeline in weeks. Category and content-led demand takes two to three quarters to show in win rate and inbound quality, because you are changing what buyers believe before they search. We run both concurrently so there is something to report while the slower half compounds, and we separate the two in reporting rather than blending them into one attribution number.