Discussion
  • Digital PR

How do you attribute ranking gains to a PR campaign when content, technical and brand changes all happened in the same quarter?

Published 1 reply

Discussion starter: a digital PR question, posted with an answer from Niraj Raut to open the thread. If you have dealt with this on a site, reply with what you saw, especially where it differs.

Disclosure: Niraj Raut, who posted this answer, runs the SEO consultancy linked at the end of it.

1 reply

  • admin
    Admin
    Original poster

    With content, technical and brand changes in the same quarter, you usually cannot prove how much of the gain came from PR. What you can do is build a defensible estimate. Each change works through a different mechanism, so each should leave a different fingerprint: on specific pages, specific query types, and at specific times. Compare the places PR should have touched with comparable places it did not, and report a range with a confidence level rather than one attributed number.

    The distinction that matters most is what the PR campaign plausibly did. If it earned links to particular pages, look for gains concentrated on those pages and their topical neighbours. If it mainly raised brand awareness, look at branded demand first. Whether branded demand then lifts non-brand rankings is widely argued and not documented by Google, so I would not claim that path without direct evidence.

    Why a quarter-on-quarter comparison cannot answer this

    A sitewide before-and-after chart mixes at least four internal changes with several external ones:

    • Google updates. The March 2026 and May 2026 core updates both completed this year, and Google has documented that smaller core updates happen in addition to the announced ones.
    • Reporting changes. After Google stopped supporting the &num=100 parameter in September 2025, many sites saw desktop impressions fall and average position improve, while clicks were affected far less, according to Brodie Clark’s analysis. Year-on-year impressions and positions are distorted across that date.
    • Seasonality and demand. A market-wide rise in searches lifts everyone.
    • Competitors. Your gain may be a competitor’s loss in an update.

    It is like a patient who started a new medication, changed their diet and began exercising in the same month. The drop in blood pressure is real, but one reading cannot say which change did it. A doctor would look at markers each intervention affects differently, and at when each one moved. Attribution works the same way: you are not reading causation off one line, you are checking whether the pattern across pages, queries and dates fits one explanation better than the others.

    Map each change to the fingerprint it should leave

    Change Where the effect should show When What it should not affect
    PR links Linked pages, and pages they link to internally, on non-brand queries After the linking pages are published, indexed and crawled, often weeks after coverage Unrelated sections with no new links
    PR exposure (brand) Branded impressions and clicks, direct traffic, brand interest in Google Trends Close to coverage peaks, decaying afterwards Non-brand rankings, at least in the short term
    Content changes Rewritten or new pages After recrawl of those pages Pages nobody edited
    Technical changes Whole templates or sitewide: crawl stats, indexing, rendering After deployment and recrawl Usually no preference for linked pages
    Brand changes (name, design, positioning) Branded queries, SERP click-through rate, direct traffic From launch Non-brand rankings, unless the site changed too

    A step-by-step method

    1. Build a dated change log. Record every deployment, content release, coverage date and brand launch. Record the date each PR link was first live, not the campaign launch date. Add the key dates as Search Console custom annotations so the whole team reads the charts the same way.
    2. Split brand from non-brand. Search Console’s branded queries filter does this with Google’s own classification. Spot-check what it counts as branded, especially if the brand change introduced a new name.
    3. Build a page matrix. Classify each page by whether it received PR links and whether its content changed. The cells are what make the comparison possible.
    4. Compare trajectories, not totals. For each cell, compare non-brand clicks and impressions in the pre-period and post-period. The difference between linked-but-unchanged pages and unlinked-and-unchanged pages is the closest estimate of the link effect, because sitewide factors like the technical release and the core updates hit both groups.
    5. Check timing against link dates. Gains that start before the links existed are not link gains. Also check whether the coverage links are followed, whether syndicated copies carry them, and whether the linking pages are indexed at all.
    6. Model the counterfactual if you have the data. Tools such as Google’s CausalImpact package estimate what would have happened without an intervention by using control series. Its documentation stresses that controls must be “themselves not affected by the intervention” and assumes the pre-period relationship between controls and the treated series stays stable. A sitewide technical release can break that assumption, so choose controls carefully and say so in the report.

    Linked, content unchanged

    The cleanest read on the PR link effect, plus sitewide factors.

    Unlinked, content unchanged

    Your baseline: technical changes, updates and seasonality only.

    Linked, content changed

    Links and content combined. Cannot be separated on its own.

    Unlinked, content changed

    The content effect without links, plus sitewide factors.

    Hypothetical example:

    A retailer’s quarter includes a digital PR campaign earning links to 15 guide pages, a rewrite of 40 category pages, a faster template and a logo refresh. Non-brand clicks rise 18% sitewide. In the matrix, linked-but-unchanged guides rose 30%, unlinked-and-unchanged guides rose 12%, and rewritten categories without links rose 20%. A reasonable reading is that roughly the gap between 30% and 12% relates to the links on those guides, around 12% reflects sitewide factors, and the content rewrite added something beyond that on categories. That is an estimate, not a proof. If the 15 guides were chosen because they were already rising, the gap overstates the PR effect.

    Watch for this:

    Selection bias is the biggest threat. PR teams often point links at pages that are already strong or already trending, and content teams often rewrite the same pages. Check the pre-period trend of each cell. If the linked pages were outgrowing the rest before the campaign, the post-period gap is inflated.

    Evidence that strengthens or weakens the PR case

    Evidence Strengthens PR attribution Weakens it
    Where gains sit Concentrated on linked pages and their internal-link neighbours Spread evenly across the site
    Timing Starts after links were live and indexed Starts before coverage, or on a core update date
    Persistence Non-brand gains persist after coverage fades Gains track the coverage spike and decay with it (more likely brand demand)
    Competitors Competitors’ visibility stable in the same period A competitor lost heavily in the same update
    Branded demand Branded impressions rise in step with coverage volume Branded demand flat despite large coverage

    How to report it

    I would present three things. First, what is well supported: for example, branded impressions rose during coverage weeks, and linked pages outperformed matched unlinked pages. Second, a range for the PR contribution to non-brand growth, with the method behind it. Third, what could not be separated, usually the pages that received links and a rewrite. Stakeholders tend to accept “PR likely contributed between X and Y, and here is why we cannot narrow it further” more readily than a single figure that falls apart under one question.

    For the next campaign, design for measurement. Agree target pages and metrics before launch, hold back a set of comparable pages from both outreach and rewrites, and avoid shipping a sitewide technical release in the same month as a major PR push. A few weeks of separation between changes is worth more than any attribution model applied afterwards.

    If this were my site, I would stop trying to split the whole quarter’s growth into four neat slices. I would build the page matrix, estimate the link effect from the cleanest cell, report brand lift separately from rankings, and label everything I could not separate as shared. That is less satisfying than a single number, but it survives scrutiny.

    Need help with this on your own site?

    Niraj Raut works with businesses in Nepal, Australia, the UK, Europe and the US on technical SEO, ecommerce SEO, local SEO and AI search.

    See SEO services

Add a reply

Replies are open to approved members. Log in or apply to join.