The Cost of
Not Knowing.

In 1998, Google's PageRank algorithm decided that links were votes. The businesses that understood this first built moats that lasted a decade. Today, LLMs are building the next foundational graph: the citation feedback loop. Models retrain on the answers they previously produced, and the citation graph they build today shapes the citation graph they build next year. Brands cited as authoritative answers now become more likely to be cited later. Brands absent from today's citation pool become harder to surface in future cycles, regardless of how good their product is.

This is not a forecast. It is documented retrieval behaviour in production LLMs; the first compounding curve in commercial digital visibility since the original PageRank, and almost no business owner has been shown a number against it.

We have. It has a name: the Cost of Inaction, or COI. Across more than fifty Web Intelligence engagements over the last twelve months, the figure for a typical UK SMB sits between £18,000 and £90,000 a year, with mid-market businesses frequently breaching £150,000. None of those businesses had been shown that figure by their existing agency, consultancy, or platform vendor before commissioning the work.

Every agency pitch a founder sits through is selling a future. 3x growth. Double the pipeline. The slides change; the maths doesn't. Every promise is a multiplier on a future the buyer has not yet earned, sold against the stranger they will be twelve months from now if everything works.

Almost no-one runs the other calculation. The one that asks: what is your current website costing you, right now, this month, while you decide whether to act? Not the upside of fixing it. The downside of leaving it alone.

The two questions produce different numbers and different decisions. The upside calculation tells you what good looks like. The downside calculation tells you what doing nothing looks like. One is a sales pitch. The other is a diagnosis.

There is a structural reason agencies don't run it, and a deeper psychological one underneath.

Behavioural economics has known the deeper part since 1979. Kahneman and Tversky's Prospect Theory showed that the pain of a loss is roughly twice as powerful, psychologically, as the pleasure of an equivalent gain. Agencies pitching upside are fighting the buyer's neurology. Advisors quantifying downside are using it.

Above the psychology is the mechanics. Upside has unlimited ceiling. The pitch is open-ended: a 3x promise can mean anything between a respectable lift and a transformational year. The buyer fills in the gap with their own optimism. The seller is rarely held to the ceiling, only to the floor. Downside has maths. A specific number, denominated in pounds, with a defensible methodology underneath. The buyer cannot fill in the gap with optimism; they can only argue with the figure or accept it.

There is a second reason, and it is the more important one. Downside numbers force a conversation about responsibility. If your current website is leaking £45,000 a year in lost search visibility, somebody approved the decisions that produced that leak. The previous web build was signed off by someone. The robots.txt that blocks AI crawlers was deployed by someone. The hidden pricing page that loses you the AI vendor bake-off was a positioning decision made years ago. A quantified COI puts a number against those decisions in a way an upside pitch never does.

Most agencies have learned not to do this. The buyer who hears their current website is leaking £45,000 a year does not become more likely to buy. They become defensive. They take the conversation internal. Three months later, the agency hears nothing back, because the buyer was protecting an internal stakeholder rather than buying a solution.

So agencies stopped doing the maths. They sell the next thing instead. This produces the absurdity at the centre of most digital marketing decisions: the buyer is asked to commission a £20,000 engagement to fix a problem they have never been shown the cost of, and the seller is asked to win that business without ever proving the cost is real. Both sides operate on faith. Faith is a poor substitute for arithmetic.

The Cost of Inaction methodology is not novel statistics. It is conventional inputs assembled with discipline, then expressed as a range rather than a point estimate. Four numbers, multiplied. The discipline lies in being honest about each input.

1. Search demand. Total monthly search volume across your category's primary commercial queries. Pulled from Ahrefs or Semrush, cross-checked against Google Search Console for branded reality, stripped of irrelevant intent.

2. Visibility gap. The difference between a fair share of click-through rate at your category's competitive density, and your actual click-through rate today. A business ranking position eight on a query with 14,000 monthly searches is leaving roughly 1,200 sessions a month on the table compared to a position three ranking, sessions a competitor is currently capturing.

3. Conversion rate floor. The realistic rate at which your site converts a visitor into a lead, booking, or sale. Not the rate from a hand-picked landing page. The rate across the journey a real user actually takes. Most B2B services sit between 1.5% and 3%. Most local hospitality between 0.8% and 2%. Most e-commerce between 0.5% and 2%.

4. Average deal value. Contribution margin per converted enquiry, not headline price. This is where most COI calculations die. Businesses overstate it because they're thinking of best-case clients rather than the long tail of small jobs that pad the average.

Multiply, sense-check, then express as a range using a sensitivity envelope of roughly ×0.7 to ×1.4. A point estimate is theatre.

The third-party indexing tools (Ahrefs, Semrush, Google Search Console itself) are imperfect barometers. Search volumes are estimates, click-through rates are model-derived, conversion rates drift. The sensitivity envelope is precisely what stops the methodology from treating any of those numbers as gospel. The discipline is in being honest about the inputs, not in pretending the inputs are clean.

A range that wide tells the buyer something an agency pitch never does: even at the conservative end of our model, the cost of doing nothing is greater than the cost of finding out. That is the threshold the methodology is designed to test.

If the COI is greater than the cost of the report, the report pays for itself before anything else happens. Every fix delivered after that compounds the return.

If the COI is lower than the cost of the report, congratulations. Your visibility is in better shape than the market average. You don't need a report. Keep doing what you're doing. That is the only honest answer most businesses have never been given.

Until 2024, COI calculations were stable. Search traffic moved slowly. Conversion rates moved slowly. Average deal values moved slowly. The methodology produced a number that held up for a year or more between recalculations.

That stability is gone.

AI Overviews now intercept the highest-intent commercial queries before any organic listing appears. Ahrefs' 2025 study of 300,000 keywords measured a 34.5% drop in click-through rates on top-ranking pages when an AI Overview appears alongside them. Zero-click queries, searches that complete without a single visit to any website, now represent a substantial and rising share of UK search behaviour, with Similarweb reporting around two-thirds of AI-Overview-triggered searches ending without any onward click.

Each of these compounds the visibility gap. A business that was leaking £30,000 a year on the old click-through model is leaking somewhere closer to £45,000 on the post-AI-Overview reality. The COI is rising while the cost of the fix is falling. Most of the moves required to recover ground are content and schema work, not platform replacement.

But the visibility gap is the smaller half of the story. The compounding effect underneath it is the part that makes this curve different from anything that came before.

AI citation positions reinforce themselves. Models retrain on the answers they previously produced. Brands cited as authoritative today become more likely to be cited tomorrow. Brands absent from today's citation pool become harder to surface in future cycles. This is documented retrieval behaviour in production LLMs, not speculation. The citation graph is a feedback loop, and feedback loops have memory.

Which means a business that is invisible in AI answers in Q2 2026 is not just losing this quarter's enquiries. It is making itself harder to recover in Q4 2026, harder still in Q2 2027. The longer the absence, the deeper the hole. Inaction is no longer the same shape it was two years ago. It compounds.

The COI methodology, applied seriously, is no longer a one-off diagnostic. It is a quarterly measurement against a moving floor. The methodology hasn't changed. The shape of the curve underneath it has.