SEO Measurement Standards

The standards

Seven rules. Each is written so that a report breaking it can be identified as breaking it.

1. Like-for-like or nothing

A before-and-after comparison must use the same query set, the same geographic points and the same method at both ends. Changing any of them between scans invalidates the comparison. This is the rule most often broken silently.

2. State the baseline

A percentage is meaningless without the number it grew from. A rise from position 18 to 9 and a rise from 4 to 2 can be reported identically as large percentage gains while representing very different commercial realities.

3. Grid, not point

Local results vary by searcher position, so a single-point check is not a measurement of local visibility. See search grid methodology.

4. Date everything

Results move. A screenshot without a date cannot be checked and should not be relied on.

5. Show declines

Every campaign has losing periods. A report that only ever shows gains is selecting, and the selection is the finding.

6. Separate visibility from outcome

Rankings, impressions and grid gains are leading indicators. Calls, consultations and signed matters are outcomes. Presenting the first as though it were the second is the most common way honest-looking reporting misleads.

7. Disclose who held the instrument

A measurement made by the party being evaluated is first-party evidence. It can still be useful; it cannot be independent, and it should say so.

Applied to this network's own figures

The published case studies are graded against the standards above, honestly:

Three “partly” rows. Publishing that table is more useful than publishing the case studies alone, because a reader can now see exactly where the evidence is thinner than it looks. A standard applied only to competitors is not a standard.

Why a standard is needed at all

Search results are not a fixed object, which is what makes reporting on them so easy to bend. Local results vary by where the searcher stands, they vary between runs on the same query, and they change as competitors work. A number quoted without saying where, when and on what it was measured has not described any of that.

The result is a market where two agencies can report honestly on the same firm and produce figures that disagree, and where an agency reporting dishonestly is hard to catch. A firm without a standard to hold reports against has no way to tell which it is looking at.

The standard set out here is not sophisticated. Measure the same thing, from the same points, on the same terms, on a stated cadence, and publish what was fixed. Most of the value comes from insisting on those conditions rather than from any refinement beyond them.

The chain from visibility to cases

Every measurement in this practice sits somewhere on one chain, and most reporting arguments come from measuring one link and discussing another. The chain runs from visibility, to impressions, to clicks and calls, to consultations, to signed matters.

Only the last is revenue, and it is the one an agency has least ability to report, because it happens inside the firm. That is a real limitation and it should be stated instead of covered by reporting something else more confidently. An agency can demonstrate visibility rigorously and can demonstrate calls where tracking is in place. Whether those calls became clients depends on intake, which is the firm’s side of the arrangement.

The productive division is for the agency to report the links it can measure honestly, and for the firm to hold the last one. A firm that never counts signed matters against enquiries has no way to tell a visibility problem from an intake problem, and firms lose more cases to unanswered phones than most would expect.

Attribution and its limits

Connecting a signed matter back to the work that produced it is harder in criminal defense than in most categories, and pretending otherwise produces confident numbers that are wrong.

The difficulties are specific. Much of the contact is by telephone, frequently from a family member whose name never matches the client’s. The decision often involves more than one person and more than one session, so the last click before the call is a poor account of what persuaded anybody. And people who found a firm through search will often say they were recommended, because a name they read three days ago feels like something they already knew.

What works despite this is coarse and honest. Call tracking on the profile and the site separates the two surfaces. Asking every caller how they found the firm produces unreliable answers that are still directionally useful in volume. And counting enquiries and signed matters month over month against the visibility record gives a firm the correlation, which is weaker than attribution and considerably better than nothing.

An agency claiming precise attribution in this practice area is claiming more than the data supports. This reference makes no such claim about its own published figures, which measure visibility and are labelled as a leading indicator instead of a count of cases.

The baseline is the whole argument

A measurement without a before reading is not a measurement. This sounds obvious and it is the single most common gap in agency engagements, because the baseline has to be taken before any work starts and by then the pressure is to begin producing.

A usable baseline records the grid scan with its centre point, dimensions, spacing and query list; the profile’s state including category, services and review count; the directory records as found; and the technical state of the site. Everything afterwards is measured against that, and a rescan on the identical grid is what makes a comparison like-for-like.

Where a baseline was never taken, the honest position is to say so and start one, treating earlier work as undemonstrable. That is an uncomfortable conversation and it is better than constructing a comparison from figures that were never gathered.

Reporting cadence and what each interval can show

Different work lands on different schedules, and reporting everything monthly implies a uniformity that does not exist.

At thirty days, profile and technical corrections may be visible and content work will not be. A report at this point describes what was done and what has begun to move, and a firm evaluating an agency here is measuring the wrong thing at the wrong time.

At ninety days the profile work has largely landed and early content is indexed and finding its position. This is the first interval where a grid rescan against the baseline is genuinely informative, and it is the window every published case study on this network reports.

At six months content work has had time to compete, and the pattern across the grid is worth more than the average. At twelve months the question becomes whether enquiries and signed matters moved, which is the firm’s number and the one that decides whether any of it was worth doing.

See also: Search grid methodology · Case study methodology · Agency evaluation reference