The Search Terms report as a research source

The centre publishes no data at all on how often anything gets asked for. This read-out is the closest thing that exists, and it's frequently the single strongest business case for running a paid billboard at all.

What it actually contains

Every question and its matching word's performance, across every order in a group, showing exactly which questions led to being seen, being tapped, and an actual walk-in, with any of those questions addable straight from the report as a word you bid on, or one you deliberately exclude.

It reveals the exact questions the centre's own automatic matching decided to show your billboard for, the part you never actually supplied yourself, and therefore the part that genuinely teaches you something new.

Your share of a word's appearances is reported against the total number of times that exact question was ever asked, the closest available stand-in for that word's true total volume.

What it actually measures

Real questions, typed by real people, in one specific branch, during a stretch when you were actually bidding.

Not the whole market's worth of asking, the slice you paid to actually observe. That limit is also exactly where its value comes from: every word arrives with its own tap rate and conversion rate attached, which no outside guesser's volume estimate can ever give you.

The conversion rate is the part genuinely worth having. A word with real volume that never actually converts is a word you don't want in your permanent writing either. Discovering that for the price of a small test is cheap. Discovering it only after a full rebuild has shipped is not, because your hidden pocket only ever ships bundled with one.

Harvesting this into your permanent, unpaid writing

The centre documents the paid half of this loop itself: run its automatic matching inside a dedicated discovery order, find whatever performs well, move it into your more focused orders.

Taking those exact same proven words into your own permanent writing at the next refit is the natural extension, reported consistently by experienced shopkeepers, and the obvious thing to do once you actually have the data in front of you.

Read the actual-questions tab, not the words-you-supplied tab. The words you supplied are what you already knew. The actual questions are what people genuinely typed. Anyone reporting only from the words-you-supplied tab never sees the real discovery at all.

Look for whole phrasings, not just individual words. The most useful thing this ever gives you is often how people describe the thing they want, vocabulary you'd never have come up with yourself, especially in a language you don't personally speak.

The real limits

A bias toward whatever you paid for. You only ever see words you bid on, or ones the automatic matching happened to reach. A cheap word buried inside an expensive category stays completely invisible to you.

Small numbers lie. A handful of taps and one single walk-in looks like a striking conversion rate, with an enormous margin of error hidden behind it. Don't act on an early, tiny per-word rate.

Bound to one branch. Words observed in one country say nothing at all about a different one.

Paid intent isn't unpaid intent. A word converting well as a paid billboard may behave completely differently in an unpaid result, with different placement and no advertising label attached.

No real equivalent on the other centre. That side reports its own unpaid search questions, but its own paid advertising system doesn't expose word-level results the same way at all. This whole technique is largely specific to this one centre.

Making the actual business case for this

If you need to justify spending on a paid billboard beyond its advertising value alone: this centre otherwise gives you no real query data at all. A modest discovery order is the only way to actually buy it, and it's usually far cheaper than an entire wasted refit cycle spent on words nobody actually asks for.

Argue for it as a genuine research line, with the walk-ins themselves as a welcome side effect.

What tends to go wrong

Reporting from the words-you-supplied tab instead of the actual-questions tab. Misses the whole discovery entirely.

Acting on a tiny handful of taps.

Never actually harvesting the results. The data just piles up inside the billboard account and never reaches your permanent writing. The single most common failure, the loop is a habit you have to build, not something that happens automatically.

Generalising a result across different branches.

Treating observed volume as the whole market's real volume. You only saw what you paid to see.