How I think about demand generation

Most demand gen plans start with channels. Mine starts with a number and works backward.

Give me a revenue target and I can derive the pipeline it needs, then the opportunities, then the qualified meetings, then the accounts I have to reach, then the budget it takes to reach them at a depth that actually registers. Do it in that order and the channel decision falls out of the math. Do it the other way and you're defending a media plan you can't connect to a revenue number, which is exactly where most demand gen leaders are standing when the CFO starts asking questions.

The arithmetic isn't sophisticated. Revenue goal, closed won, opportunities, qualified meetings, leads, clicks, budget. Every stage is a volume and a cost per unit, and volume times cost per unit is your budget. What you get from writing it out is that every assumption becomes something someone can argue with. When somebody wants to cut 20%, you can point at the stage that thins out and roughly when it shows up in bookings.

Three motions, and they're not interchangeable

I run demand gen as three motions in parallel, each with its own job. A lot of the mess I see in other programs comes from collapsing them into one line item.

Always-on is brand and demand creation across the full ICP. It's the baseline, it runs continuously, and it builds the awareness layer the other two sit on top of. Cut it and the other two get more expensive within a quarter or two, though good luck proving that in the dashboard while it's happening.

ABM is the account-specific motion against a target list, worked on a continuous six to twelve month cycle.

Signal-based is a prioritization layer on top of ABM. When fit, relevance, and engagement land on the same account inside a tight window, that account gets elevated and worked right away.

Naming them separately is a budget defense more than a taxonomy. Each one earns a different metric, and the fastest way to get good work killed is holding an awareness motion to a direct response bar. Brand earns its keep on reach, frequency, video completions, dwell time, and capture on cost per demo. Grade the first with the second's metric and you'll cut it every single time.

The split almost everyone gets wrong

Two jobs live inside demand gen and they don't substitute for each other.

Demand capture goes after people looking for a solution right now. Branded search, high-intent keywords, retargeting, demo campaigns. It converts demand that already exists, it attributes cleanly, and it shows up fast.

Demand creation is the longer investment and where the math gets uncomfortable. You're talking to the roughly 95% of your ICP who aren't solution-aware today, building the memory that makes a buyer think of you when the need eventually shows up. Attribution is bad, the lag can be two or three quarters, and it's always the first budget on the chopping block because of that.

The IPA and LinkedIn B2B Institute research puts the healthy split near 60% creation and 40% capture. Most B2B teams run close to the inverse, something like 80/20 toward capture, and it's incentives doing that rather than stupidity. A team measured on attributed pipeline will always over-invest in the channels that attribute.

You pay for it as a ceiling. You exhaust the people actively looking, you fight for that same small pool against every other vendor, CPA climbs, and nothing's forming behind it. Only about 5% of your market is in-market in a given quarter, so if you only ever talk to that 5%, you've capped yourself at a fifth of the market by construction.

One caveat I hold on myself here, because I sell paid media thinking and it would be convenient to skip it. Paid is an influence channel. It's very rarely the sole engine of anything, it's maybe 30% of revenue at the top end, and it's usually closer to 10 or 15%. Any plan built on the assumption that paid carries the whole number is going to disappoint somebody two quarters out.

The reach floor, and why most LinkedIn programs fail the math

Here's the operational standard I hold a paid social program to. Ten frequency against 80% of the target audience over 90 days, measured on a unique CPM basis instead of an impression CPM.

Under that floor, your money buys impressions without recall. The program is structurally incapable of building the memory it exists to build, and you end up at the familiar conclusion that "we tried LinkedIn and it didn't work." It never had the budget to do the job.

Shorter windows have their own benchmarks, which matter because 90 days is a long time to wait to find out you were underfunded. Over 30 days, roughly 35% penetration is average, 47% is strong, 67% is top quartile. Cold audiences want 30 to 50% penetration at six to ten frequency. Warm audiences can absorb 70 to 90% at eight to ten plus.

The useful thing about the math is that it drags the tradeoff into the open. If your budget can't clear the floor against the full target list, you've got two honest options. Raise the budget, or shrink the audience by cutting from the bottom tier up until what's left clears the floor. Both are defensible. Spreading the same money thin across everybody is the default when nobody runs the calculation, and it's the one option that can't work.

A few mechanical things sit underneath this and they leak budget without telling you:

• Track unique CPM, not CPM. CPM will happily reward you for hitting the same 300 people 40 times.
• Kill Audience Network and Audience Expansion. Both exist to spend your budget, not to reach your list.
• Bid manually. It tends to run about 30% cheaper per lead than the automated options.
• Audit your audiences before you trust them. Super titles balloon a 20K audience into 100K+, the platform's seniority buckets file twelve-year experts under entry level, and exclusions override inclusions in ways that strip out exactly the people you meant to reach. That's where the drift comes from.

One honest limit on all of that. It applies to awareness. For direct response, frequency starts giving you diminishing returns somewhere past six to eight impressions a month, and running both objectives against the same benchmark will mislead you in both directions.

Match the stage or the campaign doesn't work

The most common campaign planning mistake I see has nothing to do with budget. It's serving content built for one awareness stage to an account sitting in a different one.

The stages I use are unaware, problem aware, solution aware, brand aware, most aware. An unaware account doesn't want a demo CTA, and serving a definitional blog post to your most-aware accounts is worse, because you're wasting the impression on the people closest to buying.

So every campaign goes through a four-question gate. If it can't answer all four, it isn't ready to launch.

• Data. Who exactly am I reaching, and what fit and signal evidence puts them on this list?
• Distribution. What channel and format reaches those people at that stage?
• Destination. Does the page they land on match the stage they're in?
• Direction. Is the CTA proportional to where they actually are?

Most briefs answer Data and Distribution and skip the other two. That's how you end up with a beautifully targeted awareness campaign pointing at a demo form.

Measure the cycle, not the quarter

The most expensive measurement mistake in demand gen is judging it inside one quarter.

At Vector, first touch to booked demo runs about 136 days, and 107 when the first touch is a paid impression. In our Q1, 68% of closed-won came from pipeline created in earlier quarters. Most of what got booked in a quarter was seeded one or two quarters before it, and most of what I ship this quarter lands in the next two. I wrote the long version of that argument in 136 Days, and What ABM Is Actually For.

Measure by pipeline created and attributed inside the same quarter and you'll systematically undervalue everything with a lag. Then you'll cut it right before it converts, and blame the channel ninety days later when pipeline thins out.

I want to be careful here, because there's a real argument on the other side. Teams that rebuilt rigorous multi-touch attribution, with origin UTMs and journey stitching, genuinely do allocate spend better than teams working off instinct. I'm not saying don't do that work. I am saying don't wait for it. We get rated on a timeline that rigorous attribution can't meet, so the move is to name a window you can define, pick an end point you can count, measure the same window every quarter, and watch whether the number is shrinking. A real team can clear that bar this month. It beats a sophisticated report nobody trusts.

Count account progress, not MQLs

MQLs measure marketing activity. They don't tell you whether buyers are actually progressing, and they set up the attribution fight where marketing takes credit for volume and sales takes the blame for conversion.

What I run instead is a set of account lifecycle stages. Identified, aware, interested, considering, selecting, customer, expanded. The question stops being how many leads we produced and becomes how many target accounts moved a stage, and what moved them.

That reframe earns its keep on the downside too, which is the part people miss. It catches regression. An account going quiet is information, and a lead-volume model can't see it at all.

ICP accounts are worth the specificity. They win at meaningfully higher rates, carry bigger deals, close faster, retain better. Once you're measuring the progression of named accounts instead of the volume of anonymous forms, every budget and content decision ties back to the accounts that actually pay you.

Concentrate, then compound

Least popular principle on the list. Fewer channels, run properly.

The instinct when you're new or under pressure is to hedge and launch everywhere. You'll learn less from five channels at half effort than from one channel you actually understand, and you don't get the budget back that you spent proving it. Pick where your ICP genuinely is, clear the reach floor there, and layer after that.

The layering is where compounding starts. One core audience, awareness at the top, retargeting and thought leadership in the middle, a search layer catching the demand the awareness created. Each piece makes the others cheaper. Five disconnected campaigns fighting over the same budget is a spreadsheet with a strategy label on it.

I'll concede the tension, because I haven't fully resolved it in my own program. I run more than one channel. The discipline I hold is that a new channel has to clear the reach floor on its own budget before it opens, so it never gets funded by thinning out the one that's already working.

The first five things I'd check

Hand me a demand gen program tomorrow and I'm going here, in this order:

• What the creation-to-capture split actually is.
• Whether the paid audience is built from your own account list or from platform defaults.
• Whether the program clears the reach floor against that list.
• Whether offline conversions are flowing back to the ad platforms, so they optimize toward real buyers instead of form-fillers.
• What the full first-touch-to-meeting window measures.

Most programs have a problem somewhere in that list well before they have a creative problem.