Most SaaS teams say they run demand generation. Look at where the money actually goes and it is nearly always demand capture — search ads, review sites, retargeting — all of it competing for buyers who had already decided to start looking. That works, right up until it doesn’t, and the quarter it stops working is usually the quarter growth flattens.
Demand generation and lead generation are different jobs
Lead generation harvests demand that already exists. Demand generation creates demand that does not exist yet. The two get conflated because both eventually produce a form fill, and a form fill looks the same in a dashboard regardless of what produced it.
The distinction matters because the two answer different questions. Capture asks: who is searching for this right now, and can we be there? Creation asks: who will search in six months, and will they think of us when they do?
You can tell which one you are actually running by looking at what happens when you add budget. If more spend produces proportionally more pipeline, you still have unharvested demand. If more spend produces the same pipeline at a higher cost, you have run out of people who are looking, and you are now paying to reach the same finite group more often.
If your cost per acquisition climbs every quarter while your impression share is already high, that is not a bidding problem. It is a demand problem wearing a bidding problem’s clothes.
This is the point where most teams rebuild the campaign. New creative, new landing page, a different bid strategy. It rarely helps, because the constraint was never inside the ad account.
At any moment, most of your market is not buying
The single most useful idea in B2B marketing is also the most ignored: only a small share of your addressable market is in-market at any given time. The rest will buy eventually, but not this quarter. The LinkedIn B2B Institute popularised this as the “95-5 rule”, building on work from the Ehrenberg-Bass Institute. The precise ratio varies with your category and contract length — annual renewals and long evaluation cycles push the in-market share lower — but the shape holds everywhere.
Two things follow from that, and they point in opposite directions from most SaaS budgets.
First, if all of your spend targets in-market buyers, you are bidding against every competitor for a thin slice of the market. That slice does not grow because you spent more on it. It just gets more expensive.
Second, the other 95% are forming impressions of your category anyway — from peers, from communities, from whoever happens to be publishing. They are doing this whether or not you participate. When their trigger finally fires, they do not open a browser with a blank sheet. They start with a shortlist that already existed in their heads.
Being remembered at the moment the trigger fires
Buyers do not remember brands in the abstract. They remember them attached to situations — what the Ehrenberg-Bass researchers call category entry points. Nobody wakes up thinking “I need a billing platform”. They think “our invoicing broke when we moved to usage-based pricing”.
The job of demand creation is to attach your name to those specific situations before the search happens. That is a different brief from “produce content about our category”, and it produces different work: fewer definitional explainers, more writing about the specific circumstances in which the problem becomes urgent. Getting this right depends entirely on knowing which situations your best customers were actually in when they started looking.
The three layers, and why they need separate budgets
A full-funnel strategy is not one programme with a top and a bottom. It is three programmes with different jobs, different time horizons and different success measures.
- Demand creation. Reach future buyers, attach your name to their trigger situations. Slow, compounding, and largely unattributable.
- Demand capture. Be findable and convincing the moment the trigger fires. Fast, measurable, and hard-capped by how much demand exists.
- Demand conversion. Turn interest into activated, retained, expanding revenue. Mostly a product and lifecycle problem, not an advertising one.
They need separate budgets for one blunt reason. If creation and capture share a budget and are judged by a single efficiency metric, capture wins the argument every time, because capture can prove itself and creation cannot. Creation gets defunded. Six to twelve months later capture gets more expensive, because nothing is refilling the pool it draws from. The decline is slow enough that it usually gets blamed on the ad platform.
What actually belongs in each layer
Creation
- Content with an actual position in it — arguments a competitor could disagree with, not definitional explainers that rank for a week and convince nobody.
- Distribution through people rather than the brand account. Founders and practitioners get reach that logos do not, and most of that consumption never produces a click.
- Podcasts, video and communities where your buyers already spend attention, rather than channels that are convenient to measure.
- Paid social to cold, well-defined ICP audiences with no gate and no form. The ask is attention, not an email address.
- Customer stories framed around the trigger situation rather than the feature list.
Capture
- Non-brand search on problem and category terms, where people describe the symptom rather than the solution.
- Branded search, defended. If you have built any demand at all, competitors are bidding on your name.
- Comparison and alternatives pages. These are the highest-intent pages on most SaaS sites and are usually the last to get written.
- Review platforms. For a large share of B2B software categories the shortlist is assembled on G2 or Capterra before your site is ever opened.
- The technical foundation that lets any of it be found, plus paid media structured to be read rather than just to spend.
Conversion
- Trial and freemium design measured on time to first value, not on signups.
- Sales assistance triggered by product behaviour rather than by an arbitrary lead score.
- Lifecycle email keyed to what someone did in the product, not to how many days have passed.
- Material that helps your champion sell internally. In most B2B software purchases the person evaluating you is not the person approving the spend, and the pitch that convinced them will not be the pitch that convinces finance.
The measurement problem, and how not to lose the internal argument
Demand creation is unattributable by design. Someone hears you on a podcast, follows a founder for four months, mentions you to a colleague, and eventually types your name into Google. Last-click attribution records that as branded search. Every channel that did the actual work is invisible.
This is where good programmes get killed. Not because they failed, but because they could not defend themselves in a spreadsheet against a channel that could.
The fix is to stop asking one measurement system to judge three different jobs, and to measure each layer on its own terms:
- Creation: branded search volume, direct traffic, share of search against competitors, and self-reported attribution.
- Capture: pipeline, cost per acquisition, payback period, win rate, impression share.
- Conversion: activation rate, time to first value, retention and expansion.
Of those, self-reported attribution is the cheapest change with the largest payoff. Add one open-text field to your demand form — “How did you hear about us?” — and read the answers every month. It is imperfect, people misremember, and it will still tell you about channels your analytics cannot see. Teams that do this routinely discover that a podcast or a community they were about to cut is named repeatedly by their best-fit buyers.
Judge demand creation by last-click and you will defund the thing that makes your capture affordable.
Sequencing: capture first, then create
Order matters more than split, and the common advice gets it backwards.
Do not start with demand creation if you have no capture in place. Creating demand without a capture net means paying to make people curious, then handing them to whichever competitor ranks for the search you provoked. Build the harvest first: search coverage on problem and category terms, comparison pages, review profiles, honest tracking, and a position clear enough that the pages have something to say.
Then watch for saturation. You have taken what is available when impression share on your core terms is already high, cost per click is rising while volume stays flat, budget increases produce diminishing returns, and branded search is not growing. That combination is the signal to move money into creation. Not a calendar date, and not a benchmark from a blog post.
On the split itself: the 60/40 brand-to-activation ratio from Les Binet and Peter Field’s IPA work is the most-cited starting point, and it is a reasonable direction of travel. It is also an average drawn largely from consumer categories, not a law of nature and not a target for your specific business. Use the saturation signals above to decide when to shift, and let the ratio be the output of that decision rather than the input.
The operating cadence
Different layers need different review rhythms, and collapsing them into one weekly meeting is how creation dies.
- Weekly: capture only. Spend, pipeline, cost per acquisition, anything visibly broken. Adjust budgets, bids and creative.
- Monthly: creation output shipped, self-reported attribution answers, win and loss themes from sales conversations.
- Quarterly: positioning, ICP, share of search, and the budget split between layers.
The rule underneath this: never judge demand creation on a weekly cadence. Its feedback loop is measured in quarters, and forcing it into a weekly reporting rhythm guarantees it looks like a failure every single week.
Where this usually goes wrong
- Gating everything. Putting a form in front of top-of-funnel content converts a reach asset into a lead asset and destroys most of its reach. Gate the things a buyer would trade an email for, and nothing else.
- MQL theatre. Volume targets on a metric sales does not respect produce a marketing team that hits its numbers while pipeline shrinks.
- Changing the message every quarter. Memory is built by repetition. Your internal boredom with the message arrives roughly two years before the market’s.
- Treating the buyer as one person. Software purchases are made by groups, and the objection that stalls the deal usually comes from someone your champion never introduced you to.
- Optimising the campaign when the problem is the argument. No amount of bid tuning fixes a position the market does not find compelling.
Start where the leak is, not where the framework starts
A full-funnel strategy is not something you launch in one piece. Almost every SaaS team already has one of the three layers running reasonably well, one running badly, and one missing entirely — and the fastest gains come from naming which is which honestly rather than from rebuilding everything at once.
If capture is unsaturated, take that demand first; it is the cheapest revenue available to you. If capture has plateaued, no amount of campaign optimisation will move it, and creation is the only remaining lever. If conversion is the weak layer, more demand of any kind just makes the leak more expensive. The sequence is always: find the binding constraint, fix that, then move to the next one.
Most of the hard thinking here is not media strategy at all. It is knowing precisely who buys, what makes them move, and why you rather than the alternative — which is go-to-market work, and it has to happen before the budget conversation, not after it.
