Your Best Channel Is Probably Harvesting Demand You Had Anyway

On 8 January 2026 the IPA published an argument by Charlie Ebdy, Chief Strategy Officer of Omnicom Media Group UK, that the advertising environment itself has become hostile to effectiveness (distraction, clutter and ad load at new heights). He borrowed a term from depression-era economics for it: secular stagnation, Alvin Hansen’s name for an economy whose performance is capped by structural conditions rather than by the decisions of the people running it.

The claim is worth taking seriously because of what it implies about your account. If the ceiling is environmental, then the thing capping your returns is not sitting in your campaign structure, your creative testing cadence or your bidding strategy. It is sitting in the feed, and no amount of optimization reaches it.

The retreat that follows, and why it feels like progress

When attention gets expensive, budget moves toward wherever attribution is cleanest. That means the bottom of the funnel: brand terms, retargeting pools, audiences already deep in consideration. Reported efficiency improves almost immediately. Tom Goodwin has called the resulting pattern a doom loop, arguing that the better strategy is building pricing power rather than chasing immediate sales, a position that drew a week of argument from effectiveness specialists across the industry (summarised here by PPC Land).

The mechanism is simple enough to describe in a sentence. Spending on people who were already coming makes your dashboard look better and your business no different. The numbers improve because the denominator changed, not because anything was created.

The cleanest evidence for this is a decade old

The strongest test of the idea was not run by an agency. Economists Thomas Blake, Chris Nosko and Steven Tadelis ran a series of large-scale field experiments at eBay, switching paid search off and measuring what happened. Their finding on brand keywords was blunt: ads on the company’s own brand terms produced no measurable short-term benefit. Returns from paid search overall came out as a fraction of what conventional non-experimental estimates suggested. For non-brand keywords they found new and infrequent users genuinely were influenced by ads, while frequent users, whose behavior was not influenced, accounted for most of the spend, producing negative average returns (Blake, Nosko and Tadelis, NBER Working Paper 20171).

Read the caveats honestly, because they matter. This was one very large marketplace with overwhelming brand recognition and top organic placement for its own name. A challenger brand with weak organic visibility and three competitors bidding on its name is in a different position, and for that firm branded search may be defensive spend that genuinely protects revenue. What the study establishes is not that brand terms are always waste. It establishes that the reported return on them is not evidence either way, and that the gap between reported and incremental can be the entire result.

What to do with a budget you have to set on Monday

Three things, none of them expensive.

Split branded and non-branded into separate reporting lines and never blend them again. A single efficiency figure that mixes demand capture with demand creation cannot tell you which one is failing, and it will keep looking healthy right up until the branded volume stops growing.

Then run a holdout on your brand terms. Pause them in a few matched regions for a couple of weeks and watch total conversions rather than campaign conversions. It is the cheapest incrementality test available and it is the one most accounts have never run, usually because nobody wants the answer during a quarter they are being judged on.

Finally, look at your non-branded acquisition cost on its own, with no subsidy from the harvesting side. That number is the real price of growth, and it is the one to plan against. If an agency’s compensation happens to ride on blended efficiency, expect resistance to all three, which is why the metric a performance fee is written against deserves more scrutiny than the percentage attached to it. The same split is the first thing to demand of any agency case study, alongside the other questions that collapse a pitch deck.

Pricing power is the hedge, not better optimization

Here is why the brand argument is a financial one rather than an aesthetic one. If media costs rise and you cannot raise your prices, the increase comes straight out of margin. Brand equity is what lets you decouple the two. It is also, less romantically, what produces the branded search volume your performance account has been harvesting all along, which means the budget that looks least accountable is generating the demand the accountable budget takes credit for.

That relationship also sets a ceiling on how long a pure harvesting strategy can run. You can only intercept the demand that exists. Stop creating it and the excellent branded numbers decay quietly, over quarters, in a way no monthly report is designed to show.

The other deflationary lever is what happens after the click. When traffic gets more expensive, improving the rate at which it converts is the only place left where the economics move in your favor, which makes the path from click to completed action the least glamorous and most reliable investment available in a stagnating environment. Both levers are slower than reallocating spend, and both survive the next algorithm change, which is more than can be said for the alternative. If you want the split done properly across a live account, that is where a paid media program should start rather than finish.

Frequently Asked Questions

Does this mean I should stop bidding on my brand name?

Not on the strength of a study run at a different company. It means the reported return on brand terms is not evidence of their value, and that you should test rather than assume. Firms with weak organic visibility for their own name, or with competitors actively bidding on it, often find brand terms are genuinely defensive. Firms that already own the entire first screen for their name frequently find they are paying for clicks they would have received free.

How do I run a branded search holdout without losing sales?

Geographically, and briefly. Pause brand campaigns in a handful of regions that resemble the rest of your market, keep them running everywhere else, and compare total conversions rather than campaign-attributed ones over the same window. Two weeks is usually enough to see a directional answer at reasonable volume. Doing it region by region caps the downside at a small share of revenue while producing an answer that applies to the whole account.

Does the same problem apply to retargeting?

Yes, and often more severely, because retargeting audiences are built entirely from people who already showed intent. The attribution is clean, the reported returns are strong, and a meaningful share of those conversions were going to happen regardless. The test is the same: hold out a portion of the audience, measure total outcomes, and see how much of the reported return survives.

How should I split budget between brand and performance?

There is no ratio that transfers between businesses, and anyone offering you one is selling something. The more useful approach is to work out your non-branded acquisition cost with no subsidy from harvesting, decide what growth at that price is worth, and treat brand investment as the thing that lowers it over time. Then watch branded search volume as the leading indicator. When it stops growing, the harvesting is running out of things to harvest.