Your clients asked for the shape everyone else was buying, and they got it. Nobody could have seen the long term. You are now well placed to raise it.
This may seem a difficult conversation to open with your clients - it could appear to criticise years of your combined work. It is worth having anyway: paid search could stop producing new enquiries for them - fading as costs rise, or falling away as buyers ask an AI instead - and neither the fading nor the falling is in their control, or yours. Many of them have no second source to switch to.
01
Every decision was the right one
Nobody decided to stop being able to reach people. It just happened, one sensible year at a time.
Your clients wanted quick, dependable results in the shape every one of their competitors was buying, and they got it. That was the right call on the information available. The by-product took years to show: most of them can no longer go out and find customers the way they once could. Which is what makes this a hard conversation to open. Anyone can raise something that went wrong. Nothing here went wrong.
Further reading
Getting better at a channel makes it perform, which justifies more of it - and every year that happens, the alternative gets a little less practice. That is the whole mechanism. Nobody in that chain did anything other than their job well.
The research puts a useful condition on that. The bias has nothing to do with the merits of either choice - it appears between two options of identical value, purely because one pays back at once and the other only improves with practice. Nobody had to misjudge anything for this to happen, and it turns up wherever an organisation gets good at something.
It also travels. Anything that starts slow and improves with use is subject to the same bias, which is worth recalling when a client is worried about early results on a slow-burn project.
The concept is called the competency trap and it has been in the organisational-learning literature since 1988.
It is worth having the name to hand in a client conversation: it moves the discussion from “you have been doing this wrong” to “this is a documented pattern that affects every organisation that specialises”, which is both truer and considerably easier to say.
The 2020 revisit is the more recent citation - Industrial and Corporate Change, lead author at Warwick Business School - and it is the one to reach for if a client asks whether this is just opinion.
Sources
Levitt, B. & March, J. G. (1988). “Organizational Learning.” Annual Review of Sociology, 14, 319-338. Definition at p. 322.
Denrell, J. & Le Mens, G. (2020). “Revisiting the competency trap.” Industrial and Corporate Change, 29(1), 183-205. Denrell is at Warwick Business School.
02
What new-business mojo actually is
It is four habits, not a mood. Every one can be taught, and none of them is choosing who to call.
When you tell a client their new-business capability has faded, they will hear a criticism of their people. It is not one. Their team can still work a lead that arrives; what has gone is the temperament for going out. Four observable behaviours, every one a practice rather than a trait:
tolerance of non-response
belief in a long horizon
composure through the many who are not in the market, and sharpness with the few who are
improving the message itself
Naming them turns an awkward judgement into a diagnosis, and a diagnosis has a remedy - not one the client can buy with budget, but one you can help them rebuild.
Further reading
Be precise with clients here, because they will assume you mean better prospect lists and smarter picking. You do not. The list is a one-off decision, made well, at the start; what their team has lost is the ability to work a whole target audience.
Be equally honest about where the work sits. Almost all of it is inside their organisation - the four habits belong to their people, and nobody outside can perform them on their behalf. What that leaves is a period rather than a problem: the months while their team is going out cold again and is not yet good at it. Paid search may well carry on performing throughout, which is what makes the period survivable - and also what makes it easy to abandon. The risk in front of you here is that the rebuilding stops before it is finished.
Your part is coordination rather than delivery. The team needs to practise genuinely cold outreach, because that is the habit that faded - but a long run of pure cold work with nothing to show for it is what makes people quietly stop. Coordinated alongside it, direct access to named decision-makers can put some warmer conversations in front of them while they get their hand back in: enough to sweeten the work, not so much that they stop practising the part that matters.
A conversation about the client's own capability is a consultative and strategic one. It is hard for a competitor to copy, and it opens work - campaign planning, data selection, fulfilment, creative, A/B testing across a whole audience - that sits well outside the auction channels everyone else is bidding for.
Nothing sabotaged this. It stopped because the other thing worked.
This is the safe way to raise the subject with a client. Nobody sabotaged their new-business capability and no supplier is to blame. Their inbound worked, so going out stopped being necessary, so it stopped being practised. The success did it, and that framing lets you open the conversation without implying anyone got anything wrong. A second source is buildable - but only with notice.
Further reading
The success framing is also what makes this proposable. The programme performed - that is not in dispute. What nobody controls is whether it goes on performing on the same terms.
So the case to put to a client is a resilience one: a second route to their market, built while the first is still working. It takes time because their team may no longer have the habits to work it - which is exactly why a client-results-focused agency raises it now, while there is room to respond rather than react.
Keep the scope tight in front of a client and the argument becomes close to impossible to object to.
It is not anti-digital. It is not a criticism of the campaigns. It is a statement that one route to market being reliable is precisely why a second route stopped being maintained - and that having only one working route is a commercial exposure regardless of how well that route performs.
If a client pushes on evidence, the position is a strong one. Capability lost through disuse is well established across several industries and five decades of research. It has not been measured in B2B sales specifically - but the mechanism is not industry-specific, and it would be a bold assumption that sales teams are the one exception.
Sources
Argote, L., Beckman, S. L. & Epple, D. (1990). “The Persistence and Transfer of Learning in Industrial Settings.” Management Science, 36(2), 140-154. Cited for the principle that past output overstates present capability, not for its figures, which are specific to the industries studied.
Darr, E. D., Argote, L. & Epple, D. (1995). Management Science, 41(11), 1750-1762.
Benkard, C. L. (2000). American Economic Review, 90(4), 1034-1054.
One channel collects. The other collects and builds.
It reaches the ones who are ready today. It also reaches the ones who will be ready later. Only one channel here does both.
Worth putting to a client plainly. Direct access is not a long-horizon alternative to their paid search. It reaches ready buyers too - including the ones who never searched, searched somewhere else, or asked an AI instead - and it adds the thing their search spend cannot buy at any price: a position with the buyers who are not looking yet, who are the majority of their market and much of the revenue that follows.
Further reading
Direct access pays in the current period, reaching ready buyers without waiting for them to search - which answers the payback question before a client raises it.
The difference is what happens to everyone else. In paid search they do not exist, and nothing is left behind with them. Here the same spend that produced a client's orders this quarter has also bought them a position for the next several.
By the moment of intent the shortlist is largely drawn - around four of the five suppliers a buyer seriously considers are on it from day one, and the one ranked first before any seller contact wins more than eight times in ten. Performance at that point measures how well a client converts a position decided months earlier, somewhere they were not present.
Both findings are citable in a client deck. Dawes at Ehrenberg-Bass for the 95:5 split; 6sense's 2025 Buyer Experience Study for the shortlist behaviour.
The timing half of this argument - that most of a market is not buying at any given moment - is the one we set out in Your Rented New-Business Pipeline.
Sources
Dawes, J. - Ehrenberg-Bass Institute, the 95:5 rule.
One campaign, judged by the wrong yardstick, closed the subject for a decade.
Start with what the objection is blocking. The one route a client fully controls is the one they will need to be good at, and skill there comes only from doing it often. What usually stops that is a single sentence: we tried that once. Do not argue with it - the campaign they remember may well have underperformed. Point out what it was up against: one attempt, at their least practised, judged by inbound norms, while the channel they kept was allowed to iterate monthly for years.
Further reading
Ask the question rather than assert the answer. In most cases what they measured against was an inbound conversion benchmark, which makes a normal direct response rate look like a poor result.
So the yardstick was wrong, not the campaign. Nobody has yet found out whether the approach works - which means the capability they will need has never actually been tested.
Three citations worth having if a client challenges this. The mechanism is Denrell & March (2001) in Organization Science.
The 2024 generalisation shows it survives when a client merely does less of something rather than stopping - useful when they insist they still do some direct. And Dittmar & Duchin (2016) in Review of Financial Studies found the same asymmetry across 9,000+ managers, so it is not a laboratory artefact.
Sources
Denrell, J. & March, J. G. (2001). “Adaptation as Information Restriction: The Hot Stove Effect.” Organization Science, 12(5), 523-538.
Denrell, J. (2024). “Adaptive Sampling Policies Imply Biased Beliefs.” arXiv:2404.02591.
Dittmar, A. & Duchin, R. (2016). Review of Financial Studies, 29(3), 565-602.
Le Mens, G., Kovács, B., Avrahami, J. & Kareev, Y. (2018). Psychological Science, 29, 1475-1490.
06
It leaves when people leave
"We used to do a lot of direct" is a statement about the past, not the present.
When a client says they have done this before, treat it kindly and carefully. What they mean is that the company once had people who could. Some of them may have moved on, and what they knew went with them - it was temperament and craft, not process, so none of it was written down. And the chances are nobody will have been keeping those skills up in the meantime.
Further reading
Where this matters commercially is expectation setting. A client who believes they are restarting something will expect early results and may become disheartened if they seem to underperform - recreating the exact hot-stove pattern the previous card describes.
Setting it up as a rebuild, with the early efforts framed as practice, is what stops a normal start being read as a bad one.
The 2023 review in The Learning Organization is the citation if a client wants one. Its conclusion is that tacit knowledge loss does more damage than explicit knowledge loss, because tacit knowledge cannot be codified and therefore cannot be handed over.
Useful when a client points at their CRM as evidence the capability is retained.
Sources
“Knowledge loss induced by organizational member turnover: a review of empirical literature” (Parts I and II), The Learning Organization, 30(2), 2023, pp. 117 and 137.
Argote, L., Beckman, S. L. & Epple, D. (1990). Management Science, 36(2), 140-154.
07
Challenge is the service
Supplying what was asked for is fulfilment. Asking the better question is the job.
If you have run campaigns for a dozen clients, you have a view none of them has. A client sees one company. The platforms report by period. Nobody inside the business is looking across ten years. You are. And across enough clients the same shape shows up: paid search has brought in their new business for years, and because it worked, going out to find customers directly stopped being necessary - so the ability to do it has quietly faded. Delivering what a client asks for is hygiene; it keeps the account. Raising what they have not framed yet is what makes an agency a first pick, and, put with diplomacy, a retainer conversation rather than a rate-card one.
Further reading
What this adds is a name for the pattern and the research behind it. None of it says anything negative about paid search: it has done exactly the job it was bought to do, every month, as contracted - and most likely still is. What it could not do, structurally, was leave behind a capability the client owns.
Test it against your own portfolio. The pattern should be clearest in the clients whose inbound has performed longest, since they are the ones least likely to have approached anyone directly in years. If it holds across them, you have found your own evidence.
There is also a straightforward first-mover argument. This question will be asked of your clients eventually, by a finance director, a new marketing hire, or a competing agency in a pitch. The agency that raised it first is holding the answer when it lands, with a recommendation already on the table.
Your clients' new business should probably arrive by more than one route.
Your clients have not simply become dependent on one channel. Every route that means approaching decision-makers directly runs on the same underlying capability, and that capability may no longer be there. Which could mean that until it is rebuilt, no other route is reliably open to them.
Further reading
The commercial effect is that it widens what can be proposed. A client with the capability back can approach decision-makers directly, which is something probably closed to them today.
So restoring it puts options back on the client's table, and a different kind of work on yours: designed rather than bid for.
It is one investment rather than several. However a client chooses to reach decision-makers directly, it runs on the same capability. So the question is not which route to pick, but whether any of them is open at all.
09
Restoring it: who does which half
The team supplies the persistence. The manager supplies the timeframe. Neither has done this in years.
Your place in this is the part the client cannot supply from inside. Direct work has a craft of its own - procuring the right list, keeping a team steady through a normal response rate, testing the approach itself - and a client who has not run it in years will likely have lost that too. Some of their marketers will never have had it, having come up entirely on inbound. Your role is to guide them as they grow the skills.
Further reading
Most of this is work an agency can do, and most of it is work nobody is doing for them now. Choose the data supplier and specify the targeting. Write the emails and the letters, and test them properly rather than sending one and hoping. Arrange print and mailing where the route is postal. Plan the cadence and volume, and coach the team through it.
The most valuable piece sits furthest from the media work: helping the client understand how this gets measured, to allow the time for the redevelopment of the skills. That means explaining how direct builds over time, and what it is worth to have a route that keeps working whatever else changes. That conversation is worth more to a client than the campaign it sits inside.
Negative early outcomes reduce how often something is retried, which means the correcting evidence never arrives. Applied here: a client measuring a long-horizon activity on a short window will generate a poor first read, stop, and conclude the approach does not work for them.
You are not asking them to relax their standards. You are asking them to measure over a window that can actually contain the result, and to value the part of it that is hard to measure.
Sources
Denrell, J. & March, J. G. (2001). “Adaptation as Information Restriction: The Hot Stove Effect.” Organization Science, 12(5), 523-538.
Denrell, J. (2024). arXiv:2404.02591 - the bias survives even where a negative belief merely reduces sample size.
10
The two kinds of hard
Some of this is just hard. The rest is avoidable.
There is no Plan B behind this. A client needing to rebuild a direct route is building the only alternative they have, so it deserves the best run they can give it.
Separate the two kinds of hard early. Nobody can promise outbound will be easy, and the real difficulty is the exercise itself - that part stays.
Everything else can be minimised: dead numbers, wrong contacts, bounced emails. All of it wastes time and squanders motivation the team cannot spare.
Further reading
The case for better data is not that it is more accurate. It is that a team's willingness to keep going is finite, and every wasted contact spends some of it on nothing at all.
When the programme is the client's only alternative route to market, giving it the cleanest possible start is not an optimisation. It is the whole point of doing it properly.
No data is perfect, and a supplier who implies otherwise is worth less than one who does not. What a team needs is not perfection but confidence: that what the client already holds has been cleaned as far as it can be, and that what they buy comes from someone with something at stake.
Skin in the game. A quality guarantee only means anything if it costs the supplier when it fails - money back, or replacement records at better than one for one. Anything softer than that is a statement of intent.
Provenance and documentation. When a client asks where the data came from, you want an answer you did not have to invent. Choosing a reputable supplier who provides data with genuine provenance, and the documentation to support it, means the question is answerable on the spot.
11
Your clients can only build it while they don't need it
Search can change inside a quarter. The alternative takes several to build.
The timing argument for your clients. Their exposure can appear in a quarter; the alternative takes several to build. So the right moment is while their current channel is still performing well - which is exactly when nobody feels any urgency. Raising it then is what a strategic partner is for.
Further reading
If a client asks why they should act while results are good, the honest answer is that the fix takes several quarters and the problem takes none. It is the same logic as any insurance decision, and it does not require predicting that anything will go wrong.
The planning consequence is worth writing into any proposal. Frame the first campaigns as calibration rather than performance, with success defined as a measurable improvement in the approach rather than a return.
A client who signs off on that framing will not abandon the channel in month three.
The question takes a minute. The answer takes years to change.
A question worth putting to your clients, and one they will not have been asked: what proportion of their market could they contact directly on Monday morning, without paying anyone for access? Most will not know, and the answer is rarely comfortable - not because the names are unavailable, but because nobody there has needed to go out and find customers for years.
Further reading
Whatever the proportion is, it will not improve on its own. The names can be sourced. The practice behind using them cannot be bought at any speed.
Which is why the timing of the question matters as much as the asking. Put while a client's search spend is still performing, it is a strategic conversation. Put after it falters, it is an explanation.
Pick a sector. We will run the analysis.
Nominate a sector and a size band and we will produce the analysis. We will not ask you to name the client - sector and size is enough. The output is built to be presentable in a client review under your own name.
