SEO Baseline Before You Start: What to Record and Why

Before changing a website, create a focused SEO baseline covering priority pages, search visibility, conversions, technical conditions and business context.

Starting SEO without an agreed record of what was already happening creates a familiar problem: later, everyone has a different explanation for the results.

Organic traffic may rise because a brand campaign increased demand. Leads may fall because tracking changed. A technical fix may improve an important template, but the affected pages were never included in the original reporting. Months later, the business is left reconstructing the starting point from memory, screenshots and whichever dashboard happens to be open.

A useful SEO baseline reduces that uncertainty. It is a small, agreed record of the pages, search visibility, organic outcomes, technical conditions and external factors that matter before implementation begins.

It is not a complete export of every available metric. The aim is to create a decision-ready starting point: enough evidence to compare later observations fairly, without burying the useful signal under a mountain of charts.

What an SEO baseline is, and what it is not

A baseline is the pre-work reference point for an SEO programme. It records the current position before changes are made, including the scope and limitations of each measure.

That last part matters. A number without its definition is difficult to interpret later. “Organic conversions” could mean completed purchases, form submissions, phone calls, qualified opportunities or an analytics event that somebody named “Lead” three years ago and never revisited.

For each important measure, record:

• Date range: the period covered, including the reporting time zone where relevant.
• Source: such as Search Console, analytics, a CRM, a crawler or internal finance data.
• Segmentation: for example, organic search, landing-page group, country, device, brand status or conversion type.
• Definition: what counts as a page, query group, conversion, revenue figure or technical issue.
• Confidence: high, medium or low can be sufficient, provided the labels are used consistently. These are practical judgements, not formal statistical confidence intervals.
• Known limitations: missing data, tracking changes, incomplete query visibility, attribution rules or other reasons the measure may not tell the whole story.

This small amount of documentation is more valuable than it looks. It stops a later comparison from quietly changing the rules halfway through.

Start with the decisions the baseline needs to support

Before choosing metrics, write down the decisions the SEO work is expected to inform. For example:

• Did priority category pages gain relevant search visibility?
• Did an information architecture change improve discovery of commercially important sections?
• Did organic traffic to a set of service pages lead to more enquiries?
• Did a technical change improve access to important pages without creating new problems?
• Did non-branded visibility change, or did the apparent growth come mainly from existing brand demand?

These questions point towards a manageable baseline. They also make it easier to reject data that is interesting but not useful.

A dashboard can contain hundreds of metrics and still fail to answer any of those questions. That is how measurement becomes busywork: every platform export is treated as essential, while nobody agrees which pages, searches or outcomes actually matter.

1. Record the priority landing pages

Begin with the pages where search visibility could make a meaningful difference to the business. These might include:

• important ecommerce categories;
• high-value product or service pages;
• lead-generation pages;
• locations or market pages;
• key editorial or educational pages that support commercial journeys;
• new sections that SEO work is expected to grow.

Do not choose pages only because they currently receive the most traffic. A high-traffic article may be strategically useful, but a low-traffic service page may matter more because it represents a valuable commercial opportunity or is about to receive significant investment.

Page selection should combine several signals:

• commercial value, such as revenue, lead quality or strategic importance;
• current search visibility and organic visits;
• planned SEO or development work;
• the opportunity suggested by relevant demand;
• the page template or site section affected by the work;
• the risk of losing existing visibility if the page changes.

For each selected page or page group, record the URL or page definition, page type, site section, target market and business role. If the work concerns a template, include a representative set of URLs as well as the template itself. Otherwise, a sitewide change can affect thousands of pages without appearing in the starting view.

There is also a technical reason to define pages carefully. Search Console may associate performance with a canonical URL selected by Google rather than exactly the URL a site owner expected. That does not make page-level analysis useless, but it does mean the baseline should record how URLs were selected and where page data may not align neatly between systems.

2. Group relevant queries and visibility

Next, record how the priority pages and topics currently appear in search. Search Console measures activity associated with Google Search, including impressions, clicks, click-through rate and average position. Those measures describe search visibility and visits from search; they do not, by themselves, show whether the business gained qualified leads, sales or revenue.

Use query groups where they make the later decision clearer. A group might represent:

• a commercial category, such as “standing desks”;
• a service and its close variations, such as “corporate tax advice” and related searches;
• a problem-led topic that supports a product or service;
• a location and service combination;
• a product family or strategically important market segment.

Write down the group definitions and keep them stable. If one person includes “best standing desk” in a category group while another later excludes it as an editorial query, the apparent change may reflect classification rather than SEO performance.

For each group, consider recording:

• impressions and clicks;
• click-through rate and average position, treated as visibility indicators rather than success measures;
• the pages receiving visibility or clicks;
• country and device where those distinctions affect the work;
• the branded or non-branded classification;
• the date range and any comparable historical periods.

Query data is not a complete transcript of everything people searched. Some queries may be anonymised or omitted, and reporting availability can vary by property, date range and data volume. Record that limitation rather than presenting the visible query set as the entire market.

For larger sites, it may be more practical to combine a defined query taxonomy with representative samples and aggregated groups. A longer keyword list is not automatically more useful than a smaller set of stable groups linked to the pages and decisions that matter.

Use branded and non-branded demand as context

Separate branded and non-branded visibility where the distinction is meaningful. Branded searches often reflect existing awareness, while non-branded searches can provide useful context for discovery beyond people who already know the organisation.

That split helps explain a later result. If organic clicks rise, but most of the increase comes from branded searches during a major advertising campaign, the story is different from a rise in relevant non-branded visibility for priority categories.

It is still only context. Branded and non-branded classification can be imperfect when a brand name is also an ordinary word, when product names are ambiguous or when a business has recently rebranded. Search platforms may also classify queries incorrectly. Keep the classification rules and caveats with the baseline, and do not turn the split into the main definition of SEO success.

3. Record organic conversions and business outcomes

Search visibility tells you what happened in search. Analytics, CRM and finance data help describe what happened after the visit. The baseline needs both, because neither category answers the whole business question alone.

Choose the outcomes that are reliable enough to use. Depending on the business, that may include:

• completed purchases and revenue;
• enquiries or bookings;
• calls, where call tracking is dependable;
• account registrations or applications;
• qualified leads or opportunities recorded in a CRM;
• pipeline value, if the relationship between leads and revenue is understood.

Record the conversion definition, source, attribution setting and known tracking issues. A form-submission event may be useful, but it is not necessarily a qualified lead. A reported sale may be more robust, but it may not connect cleanly to the original organic visit.

Attribution assigns credit under a set of rules. It does not prove that SEO caused incremental conversions. Paid campaigns, direct visits, email, brand activity, pricing, stock and sales activity can all influence the outcome. The baseline should preserve the data needed for interpretation without promising a level of causality that the measurement setup cannot support.

Low-volume sites need particular care. A monthly conversion figure of two, three or four can move dramatically without representing a meaningful change in underlying performance. In that situation, a longer historical window, qualified lead stages or pipeline value may be more useful than a precise-looking monthly chart.

4. Capture technical health as a starting condition

Technical data should describe the conditions that may affect important pages, not become a league table of crawler warnings.

Record the technical issues that are relevant to the planned work, such as:

• whether priority pages can be reached and rendered;
• indexation and canonical patterns for important page groups;
• status codes, redirects and broken internal pathways;
• internal links into priority pages;
• template-level duplication or metadata problems;
• structured data where it supports the relevant page type;
• Core Web Vitals or other performance conditions where user experience is part of the work;
• known migration, release or JavaScript dependencies.

Google’s technical requirements describe eligibility conditions such as crawl accessibility, successful page responses and indexable content. Meeting those conditions does not guarantee indexing or strong search performance. Likewise, Core Web Vitals provide performance and user-experience context; they are not a complete measure of technical SEO or commercial success.

Prioritise technical findings by their likely effect on important pages and business opportunity. A crawler reporting many warnings may be less urgent than one issue affecting the template behind the company’s most valuable category pages. Counts help describe scale, but they do not decide priority on their own.

Save the crawl date, tool configuration, scope and representative URLs. Technical tools can change their rules and scores, so a future score may not be directly comparable with the starting score even if the website has not changed.

5. Add the context that can change the result

The baseline should include important events outside the SEO work. At a minimum, note:

• seasonal demand patterns;
• brand, paid media, PR or offline campaigns;
• promotions, pricing changes and stock constraints;
• site migrations, redesigns and major releases;
• changes to analytics, consent, call tracking or CRM processes;
• competitor activity or major market disruption;
• search-system changes that overlap the measurement period.

Search-demand indicators can help with this context. Google Trends, for example, provides normalised relative interest rather than absolute search volume. It can help identify seasonality or a broad change in interest, but it should not be presented as the exact size of the market.

Imagine that organic traffic to a travel site rises sharply in May. Without context, the increase may look like evidence that an SEO change worked. With the baseline, you can see that the destination was featured in a national campaign and that wider demand rose at the same time. The SEO change may still have helped, but the evidence supports a more careful conclusion.

Record a trend, not just one starting number

A single month can be a useful snapshot, but it is a fragile basis for comparison. Where data quality allows, preserve several observations before implementation so you can see the prior level, seasonal pattern and direction of travel.

Longer history is not automatically better. A migration, tracking change, new business model or major market disruption can make older data incomparable. Choose a period that is long enough to provide context but still represents the current site and measurement setup.

Where practical, define comparison page groups. For example, if SEO work will change a set of category templates, an unaffected group of similar pages may provide useful context. It is not a perfect control: sitewide internal linking, brand demand, technical releases and search-system changes can affect both groups. Comparison groups improve interpretation only when their pre-work behaviour and limitations are understood.

This is one reason the timing of the baseline matters. For a useful discussion of how evidence emerges after implementation, see After an SEO Change: When Should You Expect Evidence?.

A practical minimum for a small site

A small site does not need a complicated measurement architecture. A compact, agreed record might include:

• a manageable set of priority landing pages or clearly defined page groups;
• stable query or topic groups linked to those pages;
• organic clicks, impressions and average position for the relevant scope;
• branded and non-branded visibility, if the classification is meaningful;
• reliable organic conversions or enquiries, with definitions;
• revenue or qualified lead data where it can be reconciled;
• a short technical snapshot of the priority pages;
• seasonality, campaigns, releases and tracking changes;
• the date, source, definitions, confidence and limitations for each measure.

Store the raw exports as well as the summary. The summary is easier to use, while the raw data gives you a route back if a definition needs checking later.

What changes on a larger or more complex site?

Scale creates two problems: there are too many pages to inspect individually, and different teams often describe the same measure differently.

A larger baseline may need:

• a page taxonomy covering templates, markets, categories and commercial roles;
• stable query-group rules and versioned classifications;
• representative URL samples for large templates;
• raw exports from Search Console, analytics, CRM and technical tools;
• a change register covering releases, campaigns and tracking updates;
• reconciliation notes explaining why platform totals do not match;
• separate confidence levels for visibility, conversion and revenue measures;
• an agreed process for re-baselining after a migration or major measurement change.

The purpose is not to create an impressive reporting system. It is to make future decisions safer. If a recommendation affects a very large site or a high-value international operation, knowing which pages were included, which were sampled and which data sources disagree becomes part of implementation quality.

The baseline should make later disagreements smaller

Consider three common arguments that a baseline can prevent.

“Traffic is up, so SEO worked.” The starting record shows whether branded demand, paid media or a seasonal event also changed, and whether the priority non-branded groups moved.

“The technical fix made no difference.” The baseline shows which page groups were affected, what their starting conditions were and whether the expected outcome was improved access, visibility, conversions or something else.

“Leads are down because rankings fell.” The record lets you check rankings alongside landing pages, tracking changes, lead definitions, stock, pricing and wider demand. It may not produce a neat answer, but it prevents one metric from being asked to explain the entire business.

That is the real value of the baseline. It does not remove uncertainty or prove causation. It gives the business a shared starting point and makes uncertainty visible instead of allowing it to become an argument later.

Final checks before SEO implementation begins

Before work starts, ask whether the baseline can answer these questions:

• Which pages and page groups matter most to the business?
• Which search topics and query groups are relevant to those pages?
• Are visibility metrics clearly separated from business outcomes?
• Are branded and non-branded definitions documented rather than assumed?
• Do conversion and revenue figures have clear definitions and known limitations?
• Does the technical snapshot focus on important pages and planned changes?
• Have seasonality, campaigns, releases and tracking changes been recorded?
• Is there enough historical context to recognise normal variation?
• Does every measure have a source, scope, confidence level and caveat?

If the answer is yes, you probably have enough to begin. You do not need every metric available in every platform.

A straightforward site may need only a simple agreed snapshot. A large site, unclear analytics setup or high-risk implementation may need specialist help to reconcile data sources, define page and query groups, and establish a comparison set that people can trust. The aim is the same in both cases: record the important starting conditions before they change, so future SEO decisions are based on evidence rather than memory.

Once the baseline exists, ongoing measurement can focus on what changed and what the business needs to decide next. Our guide to what good ongoing SEO looks like covers that later operating rhythm, while SEO forecasting under uncertainty explores why even a well-designed baseline cannot make future outcomes perfectly predictable.