A B2B demand generation agency builds and runs the programmes that create and capture buying demand for a company's product: content and paid media that reach an in-market audience, the capture and qualification layer behind them, and the reporting that ties spend to pipeline. The deliverable is qualified pipeline and the system that produces it, not a lead count.
This guide is written for the buyer, not the seller. DevCommX does not sell demand generation as a packaged service, so nothing here is steering you toward a retainer with us. What follows is what a competent engagement contains, how the work is priced, how it should be measured, and the conditions under which building the function in house beats hiring anyone. The measurement question decides most of the others, and we set out the underlying argument in our piece on lead quality versus lead quantity.
The short answer: what a B2B demand generation agency does
Strip the category language and the work reduces to six functions. Audience definition, which means agreeing who counts as a buyer and which accounts are worth spending on. Demand creation, the content, media and events that make a category problem legible to people who have not yet decided to solve it. Demand capture, the search, retargeting and offer surfaces that catch the ones who have. Qualification, the scoring and routing that decides what sales sees. Measurement, the attribution and pipeline reporting. And the operating cadence, the weekly and monthly ritual that keeps the other five honest.
The reason the job looks like this is that most of a modern B2B purchase happens where no seller is present. Gartner's research on the B2B buying journey reports that buyers spend roughly 17 percent of their total purchase time meeting with potential suppliers, and describes the journey as a nonlinear loop across a set of buying jobs rather than a tidy funnel. A programme designed around that 17 percent is a sales programme. Demand generation is designed for the remainder, where the buyer is reading, comparing and arguing internally without you in the room.
That share keeps moving in the same direction. A Gartner survey of 646 B2B buyers, fielded from August to September 2025, found that 67 percent prefer a rep-free buying experience and 45 percent used AI during a recent purchase. The practical consequence for an agency brief is that the assets have to survive being read without a human present, and the capture layer has to work for someone who wants to self-serve until the last possible moment.
Demand generation vs lead generation, and why the distinction decides the budget
Lead generation captures demand that already exists. Paid search, review site listings, bought lists and gated assets all work on a buyer who has already decided they have a problem. Demand generation includes that capture layer but also spends against people who have not decided anything yet, on the theory that being remembered when they do decide is worth more than being found once they are already shortlisting.
The arithmetic behind that theory is worth stating plainly. The LinkedIn B2B Institute's 95-5 rule holds that roughly 95 percent of potential buyers in a category are out of market at any given moment, with only the remaining sliver actively shopping. If your entire budget sits in capture, you are bidding against every competitor for that sliver, and the price of it only goes up. If your entire budget sits in creation, you have no way to convert the attention you built. B2B demand gen budgets are an allocation argument between those two halves, and an agency that cannot articulate its split is not running a strategy.
This is also where most reporting disputes start. Capture channels look efficient because they claim the conversion that creation work made possible. Creation channels look wasteful because their effect lands a quarter or three later. We have written about the downstream version of this problem in the case against the MQL as a shared unit of account, and the same logic applies upstream: if the only thing you count is the last click, you will defund the thing that made the click happen.
What the engagement actually includes
A real scope of work names deliverables, not activities. The ones worth insisting on: a written ICP and segment definition with the fields it depends on; a messaging framework and at least one tested offer per segment; a content production calendar with named formats and volumes; media management with a stated budget split between creation and capture; the capture infrastructure itself, meaning forms, tracking, routing rules and lifecycle definitions; a qualification model; a reporting layer; and a documented handover so the whole thing keeps running if the engagement ends.
The item most often missing is the fifth one. Plenty of demand generation services proposals cover creative and media in detail and then assume your CRM and marketing automation are ready to receive what they produce. They usually are not. Fields are inconsistent, routing is manual, and the lifecycle stages mean different things to marketing and sales. Ask who is doing that work, and if the answer is you, price your own effort into the comparison.
The scoring model deserves the same scrutiny. A model that scores on job title and email domain will flatter the pipeline and disappoint the sales team. One calibrated against accounts you actually won behaves differently, which is the argument in our write up on win rate calibrated ICP scoring. Whoever builds it, insist that the inputs are auditable and that you can see the rule, not just the number it produces.
The channel mix, and who owns what
Most disappointment in agency work traces back to an unstated ownership boundary rather than to poor execution. The table below sets out the split that tends to hold. The left column is what a competent partner can genuinely take off your plate. The middle column is what will fail if you hand it over, usually because it depends on institutional knowledge, legal exposure or a relationship that cannot be outsourced. The right column is the strategic label: whether the channel is mainly building future demand or harvesting present demand.
Two rows deserve a note. Outbound and SDR sequencing is listed as capture because it works best when it is triggered by something, not sprayed at a list. The signal layer that makes that possible is a separate build, and we cover the inputs in our guide to intent data and buying signals. Lifecycle email and nurture is the row where handing over too much creates real risk: consent records and sending reputation are yours, and no agency should be the only party who understands them.
Account based programmes cut across the whole table rather than sitting in one row, which is why they are hard to brief. If the plan is account based, the target list and its tiering have to be agreed before any channel work starts, along the lines set out in our ABM campaign strategy guide. An account based programme without an agreed list is just advertising with a nicer name.
How a B2B demand generation agency is measured: pipeline, not MQLs
The primary measure is qualified pipeline created, expressed in currency, with a stated definition of qualified that sales has signed. Around it sit four secondary measures: cost per qualified opportunity, win rate on sourced pipeline compared with the rest, time from first touch to first meeting, and coverage against the number the company has to hit. Everything else, impressions, clicks, raw form fills, session counts, is diagnostic. Diagnostics belong in an appendix, not on the first slide.
The definition of qualified is where contracts quietly fail. If marketing and sales are working from different thresholds, the agency will optimise toward whichever one is cheapest to satisfy, and it will be doing exactly what you asked. Fix the handoff definition before the first invoice, using something closer to the framing in our breakdown of MQL versus SQL qualification rather than a score threshold nobody can explain.
Attribution is the other thing to settle up front. Google's Analytics documentation explains that attribution models distribute credit for a conversion across the touchpoints on the path, and that the data-driven model assigns credit based on how each interaction changes the estimated conversion probability. Whichever model you choose, choose it before the work starts and keep it fixed for the length of the engagement. Changing the model mid-contract makes every prior month unreadable, and it is the single easiest way for a mediocre programme to look like a good one.
One honest caveat: no attribution model settles the creation versus capture argument on its own, because the creation effect lands outside most reporting windows. Pair it with a holdout or a geography split if the budget justifies one, and with branded search volume and self-reported source on forms if it does not.
Pricing models, and what drives the number
Four shapes cover almost everything on the market. A flat monthly retainer buys a defined team and output volume, and is the easiest to compare across vendors. A retainer plus managed media adds a fee on the advertising spend, which is worth watching because it rewards spending more rather than spending better. A fixed scope project suits a single deliverable such as a messaging rebuild or a capture infrastructure rebuild. A hybrid puts part of the fee at risk against an agreed outcome, which only works if the outcome is something the agency genuinely controls.
We are deliberately not publishing a price range here. A number without its scope attached tells you nothing. What you can do instead is reconstruct it. Ask for the channel count in scope, the named roles and their allocation in days per month, the content volume by format, the number of segments and languages, and whether data and operations work is included or assumed. Two proposals at the same price routinely differ by a factor of two in actual delivered effort, and this is the only way to see it.
Three cost drivers are consistently underestimated by buyers. The state of your data, because a capture layer built on inconsistent fields costs more to make reliable than to design. The number of segments, since each one multiplies creative and nurture work rather than adding to it. And your own review capacity, which is free to the agency and expensive to you. If nobody internally has time to approve content weekly, the retainer will underdeliver and the cause will be on your side of the table.
When to hire one, and when to build in house
The case to hire a demand gen agency is strongest when you need several channels live at once, have no internal expertise in most of them, and are willing to name an internal owner who will absorb the knowledge. It is also strong when the work is genuinely seasonal or campaign shaped, where hiring permanently would leave you overstaffed in three quarters out of four.
The case to build is strongest when one channel carries most of your pipeline and is already working, when your category is technical enough that outsiders cannot write credibly without heavy supervision, or when the motion itself is the intellectual property you are trying to accumulate. There is a middle path that works more often than either extreme: an internal owner who holds strategy, data and the definition of qualified, with agency capacity underneath for execution breadth.
Whichever way you go, the operations layer stays yours. Routing, lifecycle definitions, field governance and reporting are not things to rent, because every future vendor will depend on them and every handover will cost more if they are undocumented. That is the argument we make at greater length in our revenue operations consulting guide for mid-market teams, and it is the part of this decision that outlives any particular contract.
Plan Your Demand Generation With DevCommX
DevCommX does not sell demand generation as a packaged service, and this guide is not a disguised pitch for one. What we build is the engineering layer underneath a demand programme: signal capture, win rate calibrated scoring, routing, and outbound systems the client owns outright when we leave. Our benchmark is 40+ qualified demos in ~6 weeks, from our AI SDR work on a fully scoped programme with a defined ICP. If you are scoping a b2b demand generation agency and want a second opinion on the measurement model and the ownership boundary before you sign anything, start a conversation with us. You will get an argument about your funnel definitions, not a proposal.
References
- Gartner, The B2B Buying Journey, source for buyers spending roughly 17 percent of total purchase time with potential suppliers and for the nonlinear buying jobs model
- Gartner newsroom, Sales Survey Finds 67% of B2B Buyers Prefer a Rep-Free Experience, source for the 646 buyer survey, the 67 percent preference and the 45 percent AI usage figure
- LinkedIn B2B Institute, The 95-5 Rule, source for roughly 95 percent of category buyers being out of market at any given moment
- Google Analytics Help, Get started with attribution, source for how attribution models distribute conversion credit and how the data-driven model assigns it
FAQ
What does a B2B demand generation agency do?
A B2B demand generation agency builds and runs the programmes that create and capture buying demand: audience and segment definition, messaging, content and paid media, the capture and qualification layer behind them, and reporting that ties spend to pipeline. A good engagement also hands over the systems it builds, so the capability stays with you when the contract ends.
How much does a demand generation agency cost?
Pricing follows one of four shapes: a flat monthly retainer, a retainer plus managed media spend, a fixed scope project, or a hybrid with a performance component. What drives the number is the count of channels, the volume of original content, the number of segments, and the state of your data. Ask for the per channel breakdown rather than a single figure.
Is demand generation the same as lead generation?
No. Lead generation captures demand that already exists, usually through forms, paid search and list buying. Demand generation also creates demand among buyers who are not in market yet, then captures it when they are. The LinkedIn B2B Institute puts roughly 95 percent of category buyers out of market at any moment, which is the budget argument in one line.
How long does demand generation take to show results?
Capture channels can move within weeks because they harvest demand that already exists. Creation work runs on the buying cycle, so a category with a twelve month replacement cycle will not show its effect inside a quarter. Agree the leading indicators you will judge early, such as qualified meetings and account engagement depth, and the lagging ones you will judge at the end.
What should a demand generation agency report on?
Qualified pipeline created, cost per qualified opportunity, win rate on sourced pipeline, and time from first touch to first meeting. Volume metrics such as impressions, clicks and raw form fills belong in an appendix as diagnostics. Agree the attribution model and the definition of a qualified opportunity in writing before the first invoice, because both are arguable after the fact.
Should we hire a demand gen agency or build the function in house?
Hire when you need several channels running at once, have no in house expertise in them, and can name the owner who will take the work over. Build when one channel carries most of your pipeline, the motion is already working, and the knowledge is the asset. Many teams do both: an agency for execution breadth, an internal owner for the system and the data.
.webp)
