Concentration risk in paid search: the case for a second engine

Most performance teams would never let a single supplier, channel partner or revenue line account for 90% of the business without flagging it. Yet that is exactly the shape of a typical paid search budget, and almost nobody calls it what it is. Concentration risk is a familiar idea in finance and procurement. It deserves the same standing in the media plan.

The single-platform trap

The pull towards one platform is rational at every individual step. Google carries the most query volume, so the marginal pound usually clears there first. The reporting is mature, the team is trained on it, and the auction is well understood. Every quarter the easiest decision is to keep doing what already works, and the easiest budget to grow is the one already proving out. Compounding over a few years, that produces a search programme that is effectively a single-vendor operation.

The trap is that none of the reasons for concentration are reasons it is safe. They are reasons it is convenient. A portfolio that is 90%+ weighted to one position is exposed to anything that moves that position, and in paid search the position can move for reasons entirely outside your control: an auction dynamics change, a policy shift, a product change that reshapes how your inventory is bought, or simply rising competition in your core terms. When the whole programme rides on one engine, you have no second reading on whether a downturn is the market or the platform.

There is also a quieter cost. A single-engine programme has no benchmark. When CPCs rise or conversion rates soften, you cannot tell whether the change is structural or specific to the channel, because there is nothing to compare it against. A second engine is not only a hedge; it is an instrument that tells you something about the first.

This is not an argument for search diversification as an article of faith. It is an argument for treating allocation as a decision rather than an accident.

What changed (PMax, CPCs)

For years, the case for staying single-platform rested on the idea that the dominant engine gave you fine-grained control in exchange for your spend. That bargain has shifted. Performance Max consolidates placements, creative and targeting into a single automated campaign type and hands much of the steering to the platform. Plenty of teams run it and get results, but the trade is real: you concede placement-level visibility and a good deal of the levers you used to pull by hand.

When control moves to the platform, the case for keeping all your spend on that platform weakens, because the thing concentration used to buy you, granular command, is partly gone. You are increasingly buying outcomes you can influence but not fully inspect. Diversifying the engine is one of the few ways left to retain genuine optionality over how search demand is captured.

Cost is moving too. Competitive intensity in the dominant auction has pushed CPCs up in many categories, which raises the cost of capturing the same demand and squeezes the efficiency that justified the concentration in the first place. By contrast, Microsoft Ads typically runs at materially lower CPCs than Google, commonly cited at around 33% lower on average. A second engine with a structurally different cost base is not a like-for-like duplicate of the first; it is a different point on the efficiency curve.

Risk language for the boardroom

Media teams tend to argue for a second engine in media terms: reach, incremental conversions, audience quality. Those arguments are true, and they routinely lose, because they sound like a request for more budget rather than a reduction in exposure. The more persuasive frame is the one the board already uses for everything else.

Concentration risk is the language. A single point of failure that controls 90% of a critical function would be escalated anywhere else in the business. Framed that way, a second engine is not incremental spend; it is risk mitigation with a positive expected return, which is an unusually good deal. You are buying down exposure and, because the alternative engine is often cheaper, doing so while improving blended efficiency rather than degrading it.

The supporting points are concrete. Microsoft holds roughly 14–16% of UK desktop search share, so a second engine is not a fringe experiment but access to a meaningful and distinct slice of the market. That audience also skews valuable: a meaningful share of Microsoft’s users are higher-income and desktop-dominant, and in the US around 41% of Bing users earn over $100k. Diversification that also reaches a higher-value cohort is a stronger story than diversification alone.

It helps to retire the language that makes this sound like disloyalty to the incumbent. This is not “moving away from Google”. It is reducing a single point of failure and adding a second reading on performance. The incumbent stays. The exposure comes down.

Sizing a sensible second engine

Diversification does not mean a 50/50 split, and pretending it does is how the conversation gets killed. The point is to move off a 90%+ concentration to a defensible allocation, not to rebuild the programme from scratch.

A reasonable approach is to size the second engine to the demand that genuinely exists on it rather than to an arbitrary percentage. Microsoft’s UK desktop share gives a rough ceiling on available query volume for many categories; the floor is whatever spend lets you read a clean signal within a sensible window. Between those two, the allocation is an empirical question you answer by running it, not a number you set in advance.

Three principles keep the move disciplined. First, fund it from the risk budget, not by starving the core; the second engine often pays for part of itself through lower CPCs, so the net efficiency hit is smaller than a naive split would suggest. Second, hold the targeting and measurement platform-native rather than copying Google’s structure across wholesale, because the auctions, the audience and the cost base differ and a clone underperforms. Third, judge it on blended outcomes: the question is whether the portfolio is more efficient and less exposed, not whether the second engine beats the first line for line.

Done this way, a second engine is the cheapest insurance in the media plan and frequently a source of incrementality rather than duplicated demand. The teams that resist it are usually not defending a strategy. They are deferring a decision, and concentration is the default that fills the gap. Naming the risk is the first step to managing it, and managing it is what the rest of the business already expects. If you want help sizing the move against your own category, that is the kind of question we exist to answer.