How to Set a Defensible Baseline Before You Spend a Dollar on SEO

Here is a conversation that happens in law firms more often than any agency wants to admit. Eighteen months into an SEO engagement, a managing partner asks a fair question. What did this actually do for us. The marketing lead pulls up a dashboard showing 41,000 organic sessions and 380 form fills for the trailing year. The partner asks what the number was before. Nobody knows. The old analytics property was replaced. The agency that ran the site before this one had its own tracking, and that account went dark when the contract ended. The rank tracker was set up three months into the engagement, so its first data point already includes three months of work.

At that point the engagement cannot be evaluated. It can only be argued about. Everyone involved falls back on impressions and vibes, and the decision to renew or fire gets made on personality rather than evidence. That is a preventable failure, and preventing it costs about one working day.

A defensible baseline is a dated, stored, independently verifiable record of where the firm stood on the day the meter started. Not a report an agency generated. A record the firm owns, in the firm’s own storage, in formats that outlive any platform, any dashboard login, and any vendor relationship. This article is the checklist for building one.

Why This Is an Evidence Problem, Not a Reporting Problem

Law firm partners already understand the concept. You do not evaluate a damages claim by describing how the client feels now. You establish the condition before the event, document it, and measure the delta. SEO deserves the same discipline, and it almost never gets it, because the people setting up the engagement are usually excited about the work rather than about the record of the work.

There is also a self-interested reason for the firm to insist on this. Marketing vendors get swapped. Roughly every agency relationship ends eventually, and the messy ones end with disputed logins, deleted dashboards, and two parties who each believe the numbers favor them. A baseline the firm controls converts that from a shouting match into arithmetic. It also protects a good agency, because a documented starting line is the only way honest work in a slow-moving practice area ever gets credit for what it did.

The uncomfortable part is that the baseline has to be captured before anyone touches anything. The moment a title tag changes, a Google Business Profile category gets fixed, or a stale page is redirected, the pre-state is gone. Some of it is gone permanently.

The Data Windows That Expire While You Wait

This is the section most firms find out about too late. Several critical data sources are on rolling deletion clocks. If you capture them in month four instead of month zero, part of the history simply no longer exists to capture.

• Search Console performance data. Query, page, country, and device data is retained on a sixteen month rolling window. Anything older falls off and does not come back. Google documents this in its Search Console performance report reference. If your firm has never exported this, you currently hold sixteen months of history and you are losing a day of it every day.
• Google Business Profile performance data. The insights view exposes a rolling six month window. Calls, direction requests, and the searches that surfaced the profile are gone after that. For firms whose intake is phone-heavy, this is the single most valuable thing that quietly deletes itself.
• Google Analytics 4 event-level data. Retention defaults to two months on user and event scoped data unless someone changed the setting to fourteen months. Standard reports are aggregated and persist, but the granular exploration data behind them does not. Check the setting today. If it says two months, change it to fourteen before you do anything else in this article.
• Third party visibility estimates. Tools like Ahrefs, Semrush, and Moz keep long historical charts, which is useful, but those are modeled estimates rather than measured truth. Treat them as corroboration, never as the primary record.
• The site itself. Once pages are rewritten and redirects are set, the previous version exists only in the Wayback Machine, and only for URLs it happened to crawl. Your own crawl file is the reliable copy.

The practical rule is simple. Export first, then investigate, then start work. Never the other way around.

The Snapshot, Field by Field

What follows is the actual capture list. Everything here should end up as a file, not as a screenshot of a dashboard and not as a link to a report that requires someone’s login to open.

Search Console

Export the full sixteen month window from the performance report, filtered to web search, at maximum date range. Pull it four ways and save four files. Queries, pages, countries, devices. Then export the same sixteen months split by search appearance if your pages generate rich results, because losing a FAQ or review snippet later is a real ranking-independent traffic event that people misdiagnose for months.

Separately, export the coverage or page indexing report so you know how many URLs were indexed, how many were excluded, and for what stated reason. A firm that starts with 210 indexed pages and ends with 190 has a story to tell that raw traffic will not reveal. If nobody at the firm currently reads this data weekly, that habit is worth building alongside the baseline, and the mechanics are covered in our guide to using Search Console for law firm weekly reports.

One frequent trap. Many firms only have a URL-prefix property covering the https www version. Verify a domain property as well, so subdomains and protocol variants are captured. Do it now, because a domain property starts collecting from the day it is verified, not retroactively.

Google Analytics 4

Record the property ID, the data stream IDs, the retention setting, the list of configured conversion events, and any filters excluding internal traffic. Then export at least thirteen months of monthly figures for sessions by channel group, users, engaged sessions, and every conversion event, with a separate export for organic search only.

Also record what is broken, because that is part of the baseline too. If the contact form thank-you page is not firing an event, if phone taps are untracked, if a chat widget’s conversions are double counted, write it down and date it. Later improvements in measured leads will partly be improvements in measurement, and the only honest way to present that is to have flagged it on day zero.

Google Business Profile and the Local Surface

For every office location, capture the exact business name string, primary category, all secondary categories, the address as displayed, the phone number, the website URL, the service areas, the hours, and the total review count with the average rating to one decimal. Screenshot the profile as displayed, and separately record the numbers from the interface itself rather than reading them off an image.

Export the full six month performance window per location. Then run a map pack visibility grid for your core practice terms, something like a five by five or seven by seven grid across the metro, and save the grid image plus the underlying position values. Local rank shifts geographically before it shifts in aggregate, and a grid from day zero is the only way to prove that later.

Finally, run a citation and NAP audit and save the raw output. Not the summary score. The actual list of directories with what each one currently says.

The Site as It Stands

Run a full crawl with Screaming Frog or Sitebulb and save the crawl file itself, not just the export. Preserve URL, status code, title, meta description, H1, word count, canonical, indexability, and internal link counts for every page. Then save a static archive of key page HTML, at minimum the home page, every practice area page, every location page, and the top twenty organic landing pages by Search Console clicks.

Capture Core Web Vitals from the field data in Search Console, not just a lab score from PageSpeed Insights, and note the date. Record the CMS, the theme, the plugin or module list with versions, the host, and where DNS is managed. A year from now, when something regresses, this file is how anyone reconstructs what changed.

Rankings, Recorded Honestly

Set up rank tracking before the first optimization ships, with the keyword list locked and dated. Two hundred to four hundred terms is a workable range for most firms, covering core practice areas, modifiers, city combinations, and a set of long tail informational terms. Track desktop and mobile separately and set the geographic target to the metro the firm actually serves, not the national default.

Be honest in the memo about what rank tracking is. It is a directional sample of a personalized, geographically variable result set. Two people standing in the same building can see different results. Rankings are a diagnostic input, not the deliverable, and any engagement that reports rankings as the outcome is measuring the wrong thing. Traffic, qualified inquiries, and signed cases are the outcome.

The Intake and Revenue Side

This is where firms get the most value and do the least work. Record, for the trailing twelve months if it exists, the monthly count of new inquiries, the split by channel, the count that became consultations, the count that became signed cases, the practice area mix of those cases, and average case value by practice area. If the firm does not track average case value, an estimate signed off by the partners beats nothing, as long as the memo says it is an estimate.

Also document how a lead is currently attributed. Which call tracking numbers exist, where they are placed, what the form does on submit, whether the CRM has a source field and whether anyone actually fills it in. Without this, organic growth two years out is unprovable at the level partners care about, and the fixes are worth doing before the campaign rather than during it, which is the argument we make in attributing law firm marketing to signed cases.

Storing It So It Survives a Vendor Swap

Storage is where most baselines quietly die. The record lives in a Looker Studio dashboard connected to an agency account, or in a folder owned by a marketing coordinator who leaves, and two years later nobody can open it.

Use these rules. Store everything in a folder in the firm’s own cloud storage, under a partner-owned account, not the agency’s and not a personal one. Name the folder with the date, something like SEO Baseline 2026-09-03. Make it read-only after the capture is complete. Grant the agency view access, never ownership.

Keep every export in an open format. CSV and PDF, never a proprietary dashboard link as the only copy. Any file that requires a login or an active subscription to open is not a record, it is a rental. Screenshots go in as supporting evidence, timestamped, but never as the sole source of a number. Include a short README naming every file, its source, the exact date range it covers, and who pulled it.

Set a calendar reminder to repeat the whole capture every six months. That does two things. It creates continuous history that survives the retention windows, and it means that if the agency changes, the successor inherits a real timeline rather than starting the argument over.

The Baseline Memo Everyone Signs

The files are the evidence. The memo is what makes them usable by people who will never open a CSV. Keep it to two pages and include six things. Where the firm stands today, stated in plain numbers. What is measurably broken right now, listed and dated. What will be worked on and in what order. What each party will consider success at six and twelve months. What is deliberately out of scope. And a clear statement of what cannot be attributed, such as referral volume, paid campaigns running in parallel, or a competitor exiting the market.

Have the managing partner and the agency lead both sign it. Not for legal force, but because signing forces both sides to read numbers they would otherwise skim, and it prevents the twelve month conversation where each party remembers a different promise.

Set the review dates in the memo itself, and set them at realistic intervals. A first meaningful read at month six and a real evaluation at month twelve is defensible for most law firm markets. A ninety day verdict is not, for reasons we lay out in how long law firm SEO actually takes.

What a Baseline Cannot Do

Honesty about limits is part of the value. A baseline does not create a control group. You cannot know what the firm’s traffic would have done with no work at all, because you cannot run the counterfactual. Competitor moves, algorithm updates, AI overviews eating informational clicks, and seasonal case flow all move numbers independently of anything an agency does.

Nor does a baseline make attribution clean. If the firm runs LSAs, paid search, TV, and organic simultaneously, some signed cases genuinely have multiple causes and no honest model resolves them to one channel. What the baseline does is narrow the argument to a smaller and more useful set of unknowns. Instead of debating whether anything happened, you debate how much of what happened was the work. That is a far better conversation, and it is one a partner can actually make a budget decision from.

It also has a cost, and the cost is real. A thorough baseline takes six to ten hours of hands-on work and delays the first optimization by a few days. Some firms will not want to pay for a day of exports before a single page is improved. Our position is that a day spent making the next twenty-four months provable is the highest-return day in the entire engagement.

Take the Snapshot Before the First Invoice

If you are about to start an SEO engagement, or you are three weeks into one and nobody has mentioned a baseline, stop and pull the exports today. Search Console is losing a day of history every day, Google Business Profile is losing a day of a much shorter window, and no amount of budget recovers either. Rubiks runs a full baseline capture as the first deliverable on every law firm engagement, files handed to the firm in the firm’s own storage, and we will do the same capture as a standalone piece of work for firms that have already hired someone else. Talk to Rubiks and we will tell you exactly what is expiring in your accounts right now and what it will take to preserve it.