Reading Your Baseline

You've built a panel and run it. You've audited your brand. What's sitting in front of you now is a pile of saved answers, and a pile of answers is not a baseline. This chapter turns it into one: a set of numbers you can track monthly and a diagnosis that tells you which work to do first.

Score simply

For every question in every engine, record four things. Named: were you mentioned at all, yes or no. Cited: was your website one of the linked sources, yes or no. Position: if the answer is a shortlist, where do you sit in it. Description: one line on how you were characterised, in the answer's own words. That's it. Resist the urge to build an elaborate weighted index on day one. A simple yes-no grid you can maintain beats a sophisticated score you'll abandon, and the trend line matters more than the scale.

From the grid, calculate presence rates: the share of questions where you're named, and the share where you're cited, broken down by question family and by engine. Those two numbers, tracked monthly, are your scoreboard. Expect them to be low. Most established businesses running this exercise for the first time find they're strong on brand questions, patchy on category questions, and absent on cost and problem questions. Low is not bad news. Low is the size of the opportunity, measured.

Diagnose every gap to one of four causes

Here's where the baseline becomes a plan. Every question you're absent from has a cause, and the cause is always one of the four selection pressures from chapter one. Work through them in order for each gap.

First, check retrieval: search the underlying topic in ordinary Google and Bing. If you're nowhere in the results, the engine's shortlist never contained you, and this is a search visibility gap. The fix lives in classic SEO and the technical layer, chapter eleven. Second, if you rank but aren't lifted, open your ranking page and ask whether anything on it answers the question directly, in a passage that stands alone. Usually nothing does. That's a content gap, chapters seven and ten. Third, if your content is used but you're not clearly named, or the engine seems unsure who you are, that's an attribution gap, chapter eight. Fourth, if the engines describe you thinly or hedge about you while competitors get confident treatment, you have a corroboration gap: not enough of the web agrees about you. Chapter nine.

Tally the diagnoses and you have your priority order for Part III. Most businesses discover their gaps cluster: a manufacturer with a thirty-year reputation and a six-page website is almost all content gap; a business with a big site and no reviews footprint is corroboration. Do the biggest cluster first. Spreading effort evenly across all four is how a year passes without any number moving.

A warning about noise

Answer engines are not deterministic. Ask the same engine the same question twice in a day and you can get different names in the answer. This means a single run is weather, not climate. Your baseline should be at least two runs a couple of weeks apart before you treat any number as real, and month-to-month movement of a few points in either direction is noise, not signal. What you're watching for is direction sustained over quarters. Anyone who promises you a precise visibility score from one automated pass is selling you the weather.

One constraint we hold ourselves to and recommend you hold anyone to: scores need human eyes. Automation can run the panel and capture the answers, but whether a mention is positive, whether a description is accurate, whether a citation is really your page, are judgement calls. Baseline in hand, diagnosis done, you now know exactly which of the next five chapters matters most for you. That's Part III: the work.