Your Monday morning starts with coffee, then you check Google Search Console. Your stomach drops. Traffic is down 40%. Rankings that took months to build have vanished overnight. Pages that ranked #3 now sit on page four, or worse, they’re gone entirely.
The panic reaction is to start fixing. It is also the most expensive work we get called in to undo, because fixes made in the first 48 hours are usually aimed at a cause nobody has verified yet, and they add new variables on top of the original problem.
This page walks through the diagnostic order Rank Nashville uses when a client’s traffic falls off a cliff: what we look at, in what sequence, what we refuse to touch without evidence, and where the line sits between something you can fix this afternoon and something that needs us.
First: Establish What Actually Dropped
Before diagnosing a ranking drop, confirm that rankings are what you lost. Positions, impressions, clicks, conversions, indexed pages, and search demand are six different things. They look nearly identical on an analytics graph and they lead to six different investigations.
Start by putting Search Console clicks next to your analytics organic sessions. If Search Console holds steady while analytics falls, the problem is probably measurement, not search: a tag removed in a deploy, a consent banner reconfigured, a property or stream swapped, organic traffic landing in the wrong channel. That’s a real problem worth fixing, but it isn’t a ranking problem and no amount of SEO work touches it.
This isn’t pedantry, it’s the cheapest hour of the whole process. A page can hold position one and still lose a substantial share of its clicks. Search demand can fall while your visibility is untouched. A reporting change can move the numbers while nothing happens to your site at all: around the time Google stopped supporting a parameter many rank trackers relied on in September 2025, properties across the industry reported sharp impression declines that had little visible effect on clicks. Anyone who diagnoses an algorithm problem before checking which metric actually moved is guessing, and the guess costs weeks.
Once you know what moved, we work four lanes. More than one can be live at the same time, and often is: a core update landing the same week as a migration, or falling demand hiding a competitor gaining ground.
Google changed something. Recent confirmed core updates ran March 13-27, June 30-July 17, and December 11-29 in 2025, then March 27-April 8 and May 21-June 2 in 2026. Recent confirmed spam updates ran August 26 to September 22, 2025, March 24-25, 2026, and June 24-26, 2026.
The two aren’t the same thing. Google describes core updates as broad changes meant to improve the relevance of results, and says explicitly that a drop after one doesn’t mean anything is wrong with your pages. Spam updates are notable improvements to the automated systems that detect spam; Google’s own example of such a system is SpamBrain. Lining up with one is a reason to review policy compliance, hacked content, scaled publishing, and link practices. It is not proof that you violated a policy, and Google didn’t publish a target category for the June 2026 update at all.
One exception is worth knowing before you spend money chasing it. Google states that after a link spam update, making changes might not produce an improvement, because once its systems remove the effect of spammy links, any ranking benefit those links were generating is gone and cannot be regained. If that’s what happened, there is nothing to recover. The traffic wasn’t taken from you; it was never yours on the terms you thought. Recognizing that early is the difference between rebuilding and paying someone to chase a number that isn’t coming back.
What Google does not publish is what any given update was aimed at. That’s why so much recovery advice assigns intentions to updates that Google never stated, then prescribes fixes based on those invented intentions. We check the current Search Status Dashboard against the exact drop date, because what hit you might be a core update, a spam update, or a separate Search incident.
Something technical broke. Server errors, a stray robots.txt rule, noindex tags that escaped from staging. These often produce sharp drops, and they tend to recover faster than quality problems because the correction is clearer. Google still has to recrawl and reprocess the affected URLs. Recrawling can take a few days or several weeks; canonical consolidation, redirect processing, and any ranking effect can take longer than that, and fixing the technical cause doesn’t guarantee the traffic returns.
Your competitive position changed. Your page may be unchanged, but unchanged is not the same as equally competitive. Someone may have published a better answer, the query may have evolved, your information may have aged, or the authority supporting your page may have shifted. Google’s own guidance on core updates makes the underlying point directly: a ranking drop doesn’t necessarily mean anything is wrong with your page. It also doesn’t mean the page is holding its ground.
The SERP, demand, or the reporting changed around you. AI Overviews answer the query inside the results page: your position holds, your clicks don’t. Seasonality and falling search demand produce a similar-looking graph without anything changing on your site or in the results. And sometimes the numbers move while the reality doesn’t, which is why the first section of this page exists. Coalition Technologies’ analysis of the June 2025 core update found AI Overviews affecting CTR even where rankings were unmoved, and SimilarWeb data cited in that analysis put zero-click results on news queries at nearly 69%, up from 56%, taking publisher traffic from over 2.3 billion to under 1.7 billion monthly visits.
These can look similar on an aggregate traffic graph, especially before the loss is separated by metric, page, query, device, and country. They are not remotely the same problem.
The first five minutes
Before anything else, and before you change a single thing:
- Put Search Console clicks, impressions, and average position side by side for the affected window.
- Confirm whether positions actually moved, at the query and page level rather than in aggregate.
- Check what shipped: deployments, migrations, redirects, robots.txt, noindex.
- Compare the drop date against Google’s Search Status Dashboard.
- Change nothing. No rewrites, no disavows, no redirects, no deletions until one cause has evidence behind it.
The fifth item is the one that gets ignored, and it’s the one that decides how expensive the next month is.
Step 1: Use the Shape of the Drop to Decide Where to Look First
The first thing we pull is the Performance report in Search Console, comparing the last 28 days against the prior 28. The shape of the line does not tell you what happened. It tells you what to check first, which is a different and more modest claim than most recovery guides make. If the interface itself is unfamiliar, our walkthrough on reading Google Search Console covers the reports this guide leans on.
A vertical cliff, most of your traffic gone in a day or two. This makes technical failure, a manual action, or a major site change the urgent suspects. Check the Page indexing report, the Manual Actions report, whether your server was up, whether anything shipped that day, and whether robots.txt is blocking something it shouldn’t. A misplaced robots.txt rule during a migration can take a whole directory out of the index within days; the fix takes minutes and the recovery waits on Google’s re-crawl. Don’t rule out an algorithmic cause on shape alone. Google’s March 2026 spam update completed in about a day, and a core update can reach a specific site faster than the rollout window suggests.
A steady slope over two to four weeks. Update timing, competitor movement, falling demand, accumulated site changes, and gradual indexing decay all produce this shape. Line the timeline up against the confirmed rollout dates in Google’s Search Status Dashboard rather than assuming every stretch of volatility was an update, because unconfirmed volatility is common and misattributing it sends you chasing an algorithm change that never happened. The graph narrows the list. It doesn’t pick from it.
Impressions flat, clicks down. Check position at the query and page level before concluding anything: stable total impressions can hide a shift in which queries you’re found for, or one set of terms falling while another climbs. If the affected positions genuinely held, you lost the click rather than the ranking. Search those queries yourself and look at what sits above you. An AI Overview is one candidate. So are a changed title or snippet, a featured result, and a competitor with a sharper snippet.
Regular peaks and troughs. Probably seasonality, and it’s the one shape that’s usually cheap to confirm. Check the 16-month view before diagnosing anything, and compare against the same period last year rather than last month.
Step 2: Ask What Changed Right Before the Graph Did
The fastest causal test available to you is also the one people skip because it feels too simple: what shipped immediately before the drop started?
Pull the deployment log, the CMS history, and whatever record exists of who touched what. Look for theme or plugin updates, a migration or URL change, a redirect batch, robots.txt or noindex edits, navigation and internal link changes, CDN or firewall rules, a JavaScript deploy, consent platform changes, content pruning, and any move between domains, protocols, or hostnames.
If something shipped within a few days of the drop, it moves to the front of the queue. It isn’t automatically the cause, but a change you made yourself is cheaper to test and cheaper to reverse than an algorithm theory. This is also the step where the answer most often turns out to be embarrassing and free.
Step 3: Narrow It to Specific Pages and Queries
Site-wide traffic numbers hide the answer. In the Performance report, switch to the Pages tab, compare periods, and sort by click difference. Then cross-check the top of that list against conversions, lead quality, and revenue before deciding what to chase. A blog post that lost 700 clicks and a service page that lost 25 are not the same problem, and the second one is often the expensive one. High-volume losses and high-value losses are frequently different pages. Defer the rest until the first-pass pattern is clear.
For each one, open the query-level data and find which searches lost position. Then run those searches yourself and look at what replaced you. This is the step most people skip, and it’s the one that usually produces the answer, because a competitor’s page shows you what Google is currently surfacing, and hands you hypotheses no audit score will.
What we’re looking for is a pattern across the top results. If the pages that beat you repeatedly have something yours lacks, that’s a hypothesis worth testing, not a diagnosis. Three competitors might all have video and all be winning for a reason that has nothing to do with video: better intent match, stronger brand demand, a better link profile, fresher pricing. Before scaling a change across a site, we check whether the gap holds across the whole affected query set, whether closing it serves a real reader need, and whether fixing it on one or two pages actually moves anything. If no content pattern appears at all, widen the investigation rather than concluding content isn’t involved.
Step 4: Rule Technical Causes In or Out
Technical causes are worth checking early because many can be ruled in or out quickly and some have obvious fixes. Rendering, migration damage, and canonicalization at scale are a different category, and the “When It’s Past DIY” section covers them. Our rundown of the technical fixes that matter most for Nashville business sites covers this ground in more depth.
Indexing. The Page indexing report, formerly called Coverage, shows what Google has and hasn’t indexed and the reason for each. A sudden increase in important URLs listed as not indexed, unexpected noindex tags, or a spike in redirect and 4xx errors will show up here. Some URLs going unindexed is expected, so read the reasons rather than the total.
Crawl health. Settings, then Crawl stats. If Googlebot’s request volume falls sharply, there are two directions to look. Something may be blocking it: server errors, a robots.txt change, an aggressive CDN or security rule. Or Google may simply want the pages less. Crawling is governed by capacity and demand both, and Google crawls URLs less when they look unchanged, duplicated, or not worth refreshing. Those are different problems with different fixes, and the graph looks the same either way. There’s no universal number that means “bad” here. What matters is the change against your own baseline.
Mobile. Search Console’s Mobile Usability report and the Mobile-Friendly Test were retired on December 1, 2023. Any checklist that still sends you to that report is out of date. Use URL Inspection to see how Googlebot renders the page, Lighthouse or PageSpeed Insights for diagnostics, and an actual phone for the things tools miss. Mobile-first indexing means the mobile version is the version that counts, so an intrusive popup covering the screen on load isn’t a mobile problem. It’s the problem.
Core Web Vitals. Google’s published thresholds are LCP of 2.5 seconds or less, INP of 200ms or less (INP replaced FID in March 2024), and CLS of 0.1 or less. The detail most guides skip: Google assesses these at the 75th percentile of real user page loads, not on your machine and not in a simulator. Lighthouse is a diagnostic tool; the field data in the Core Web Vitals report is the assessment. These are real page-experience targets worth hitting, but failing them does not by itself explain a dramatic drop.
Canonicals. Run URL Inspection on your worst-hit pages. If Google selected a different canonical than you did, you may have found why that specific URL stopped appearing. Check whether the URL Google chose is actually the wrong one, and whether the change lines up with the queries you lost, before treating it as the cause of the whole drop.
If the drop started right after a migration, start with the redirect map: old URLs returning 404 because they were missed, chains that should be direct hops, or temporary redirects standing in for permanent moves. Don’t stop there. Check for noindex rules that survived staging, robots blocks, canonicals still pointing at the old site, internal links nobody updated, content that didn’t come across intact, sitemaps listing dead URLs, and hostname or protocol mismatches. Our Nashville migration checklist for SEO stability covers the pre-launch version, which is the one you want if a migration is still ahead of you.
Lost link support. Check whether the affected pages lost links around the same time the graph moved, in both directions. A referring page may have been deleted, redesigned, or pointed somewhere else. A redirect or canonical change may have quietly moved where an external link lands. Internal links matter here too: a navigation change or a removed hub page can strip support from exactly the pages that dropped. Treat this as a timeline hypothesis rather than proof, and confirm the lost links actually supported the pages and queries that fell.
If all of this comes back clean, technical failure becomes less likely, which is a genuinely useful result rather than a dead end. It doesn’t leave one answer standing. Update timing, content and competition, SERP presentation, demand, and reporting are all still on the table.
Step 5: Assess the Content Honestly
This is where most ranking-drop advice goes wrong, so it’s worth being precise about what we’re doing here.
We use E-E-A-T (Experience, Expertise, Authoritativeness, Trust) as a review framework, not as an algorithm setting. Google’s quality rater guidelines describe what Google wants its systems to reward. The raters themselves don’t move rankings, and Google has never announced that a given core update turned up an E-E-A-T dial. Anyone telling you otherwise is describing a mechanism that hasn’t been documented. The framework is still worth using, because it asks the right questions:
- Experience. Does the page show that whoever wrote it has actually done the thing? Specific examples, real photographs, prices someone actually paid.
- Expertise. Is there a named author with relevant credentials, and does the content demonstrate knowledge past the surface?
- Authoritativeness. Is the site a recognized source on this subject, or is it a generalist reaching into a topic it has no standing in?
- Trust. Contact details, sourced claims, HTTPS, real reviews. This one underpins the rest.
Against the pages that outranked you, compare depth rather than word count. If competitors cover six subtopics and you cover two, length isn’t the issue, completeness is. Check whether their information is current and yours is three years old. Check whether they have a named author and you have “Marketing Team.”
What we won’t tell you: that a page under 1,000 words needs a rewrite, or that losing ten positions means rewrite while losing three means update. Those rules are made up. The decision depends on whether the page has a foundation worth building on, and that’s a judgment call about the specific page, not a threshold.
Step 6: Sequence the Fixes
The ordering below is the one our agency works to, and the logic is worth stating because most people invert it.
First, anything that stops Google from seeing or serving the page at all. Indexing errors, server errors, security problems, a mobile experience that’s unusable. No amount of content work matters while a page is blocked or broken.
Second, quality signals on pages that are already visible. Author credibility, sourcing, first-hand evidence, the subtopics you’re missing, page speed.
Third, the slow compounding work. Topical depth across related content, internal linking, earned links.
Site owners tend to start with the content, because rewriting feels productive in a way that checking a robots.txt file doesn’t. Meanwhile an indexing error quietly cancels out every hour of it.
On timing. Google’s guidance is that effects from changes may appear within days or may take several months, and that some site-level improvements may not be fully reflected until a subsequent core update. That’s not evasion, it’s how the system works: Google has to re-crawl, re-process, and re-evaluate, and the competitors you’re measured against are moving too. We don’t put a recovery date on a project until we know the failure mode and the implementation scope. Anyone who gives you one before that is guessing with your money.
What we do instead is decide, before the fix ships, what would prove it worked. Write down the pages and queries you expect to move and the metric you expect to move on: impressions if it was an indexing problem, clicks if it was a snippet or SERP problem, positions if it was content or competition. Then leave it alone long enough to read. Google asks for at least a full week after a core update completes before you analyze anything, and technical corrections still wait on a re-crawl. If your named metric hasn’t moved on your named pages once that window has passed, what you know is that the hypothesis didn’t validate. That isn’t the same as knowing the fix was wrong. It may have been necessary but not sufficient, you may have picked the wrong metric to watch, or a second cause may still be sitting on top of the result. Either way the honest move is back to the evidence, not a second fix stacked on the first. That’s the whole discipline: the same evidence standard applies after you act as before.
Step 7: What to Do About AI Overviews
Your page can hold position one and still lose a substantial share of its clicks. That’s the defining feature of this problem: it doesn’t look like a ranking loss in your rank tracker, and it doesn’t respond to ranking fixes. We’ve written separately on how AI Overviews are reshaping search intent if you want the strategic view.
You don’t need a special schema, an AI text file, or a chunking format. That isn’t our opinion. Google’s current guide to optimizing for generative AI features has a section called Mythbusting that names those exact tactics, along with AI-specific rewrites and chasing inauthentic mentions, as things you can ignore. The same guide states that optimizing for generative AI search is optimizing for the search experience, and thus still SEO, because these features run on the same index and the same ranking systems as everything else. Many tactics sold as AI optimization are ordinary SEO relabeled. The rest aren’t supported by Google’s current documentation.
What actually applies:
- Be indexable and be worth citing. Clear structure, direct answers, original information a summary can’t replace. Google’s term for this is non-commodity content: material carrying experience, expertise, or a point of view that goes past what’s already available everywhere. Google’s own framing is that content like this will likely influence your presence in generative AI search over the long run more than anything else in its guide.
- Check the SERP before blaming it. Stable impressions with falling clicks is consistent with an AI Overview, but competing snippets, other search features, and a changed query mix produce the same pattern. Look at the actual results for your actual queries. It’s often said that commercial searches are safe from AI Overviews; Google’s stated rule is different, that an Overview shows when its systems judge it adds something beyond classic Search, and often doesn’t trigger at all. Google launched a Generative AI performance report in Search Console on June 3, 2026 that separates AI feature impressions from ordinary results, though it’s still limited to a subset of properties and reports impressions rather than clicks. If you have it, use it. If you don’t, watch the SERP itself, but treat it as observation rather than measurement. One search is shaped by your location, device, and history. Check the affected queries more than once, across the devices and places your customers actually use, and read those observations against your query-level Search Console data.
- Build recognition. When your name appears in a citation, that’s worth something only if the reader knows the name.
If clicks are what you lost, conventional ranking work alone may not bring them back. What tends to help is harder: better conversion from the traffic you still get, and a shift toward queries where people have an actual reason to click through.
Google is testing a Search Console control that lets eligible site owners include or exclude their content from supported generative AI features. Worth knowing in both directions: Google’s guide lists being included in these features in Search Console as a requirement for appearing in them at all, alongside the ordinary technical requirements. Inclusion is the default, so this is a thing to verify rather than configure. Excluding isn’t used as a ranking signal for ordinary results, but it does mean zero impressions and zero traffic from those features, which solves the click problem by removing the visibility along with it. For most local service businesses, giving up both is unlikely to be the first thing we’d recommend.
Two Things Not to Blame First
Structured data. Missing schema rarely explains a dramatic ranking loss, so don’t open by adding it. It helps Google understand a page and it can make a page eligible for supported rich results, but it doesn’t establish expertise, prove authority, or function as an E-E-A-T signal. That claim is everywhere and it’s wrong. It also can’t win you a feature Google has retired: HowTo rich results went off desktop in September 2023, and FAQ rich results were restricted to government and health sites in August 2023 before being removed entirely on May 7, 2026. Any guide still presenting HowTo or FAQ markup as a route to those rich results is pointing at retired Search features. Existing FAQ markup that accurately matches visible content doesn’t need to be torn out just because Google retired the rich result, but it no longer creates any FAQ rich-result eligibility. Markup that’s irrelevant or misleading is a different matter, and Google can issue a manual action for structured data that doesn’t match the page. Structured data is a description and an eligibility mechanism, not a ranking shortcut.
Your local profile, if the loss is organic. Google Business Profile and the local pack run on different rules, and a drop in one doesn’t imply a drop in the other. Diagnose them separately. Our complete guide to Nashville local search and Google Maps covers the local side properly.
One local warning belongs here anyway, because it appears in SEO checklists that should know better. When you ask for reviews, ask everyone eligible through a neutral process, not only the customers you expect to say something nice. Selecting for expected sentiment is review gating. Google wants reviews to reflect genuine, unbiased experience, and building your solicitation around predicted ratings puts the profile itself at risk.
A Warning About the Disavow Tool
This is the one section where the advice is to do nothing.
When traffic drops and you find spam links pointing at you, disavowing feels obvious. It usually isn’t, and it can turn a recoverable drop into a worse one. Site owners under pressure cast the net too wide, sweep in legitimate links alongside the junk, and lose the equity those links were carrying. The drop deepens, the cause is now self-inflicted, and it’s slow to undo.
Google already ignores most low-quality links automatically. The tool is aimed at sites with a manual action for unnatural links, or a well-supported risk of one because of links the business itself built, bought, or controlled, after reasonable attempts to get them removed. There is no threshold number of spam links that triggers it. A sudden batch of spam links, on its own, is not evidence that your rankings were harmed or that disavow is warranted.
If none of those conditions is met, the correct conclusion isn’t “your link profile is fine.” It’s “there’s no evidence for disavow.” Those are different statements, and the difference is why we don’t touch it.
When It’s Past DIY
Some situations need someone who does this full-time:
- A large loss with no verified cause once you’ve worked through the diagnostic pass, or one that persists after the confirmed cause was corrected and Google has had a fair chance to reprocess the affected pages
- A confirmed manual action in Search Console
- Rendering problems, migration damage, or canonicalization failures at scale
- Hacked content or a genuine negative SEO situation
- A site large enough that crawl budget is the actual constraint
If you’re in one of those, this is where Rank Nashville comes in.
Our agency built the diagnostic order on this page, and we run it on client sites. If you’ve worked through the steps above and the graph is still flat, or you recognized your situation in that list, bring us your Search Console access and your drop timeline. We start where this page leaves off.
We won’t promise you a ranking. Anyone who does is telling you something they cannot know, and you should treat that promise as the disqualifying signal it is. What we will tell you is what actually dropped, which cause or combination of causes is driving it, what addressing it would involve, and whether it’s worth addressing at all. Sometimes the answer is that the traffic isn’t coming back in the form you remember, and the money is better spent elsewhere. We’d rather say that in the first conversation than bill you for six months of finding out.
Send us three things and we can start: the approximate date the drop began, the site, and whether you can grant Search Console access. The audit is free, and what comes back is the root cause, not a proposal wrapped around one.
Getting to it takes three passes, and we don’t skip to the end: a deep analysis of the content, the backlink profile, and the technical state of the site. Those are the same three the steps above walk through, which is the point. If you can reach the data, isolate your changes, and measure what happens, run them yourself. What you’re paying us for is doing it on a site where one of those has broken down. You get it as a PDF, within 72 hours at the outside. Not instant, because three passes take a person reading rather than a tool scoring. Whatever we name in that report is what the three passes support, and nothing more.
Ongoing work is month to month, and we don’t use long-term contracts. Where a specialized engagement needs a defined initial scope, you hear about it before anything is signed, not after.
Get in touch with our team, or read more about how Rank Nashville works.
If you’re comparing us against other agencies, test all of us on this:
- Ask what they’d check first and why. If the answer isn’t a diagnostic order, they don’t have one.
- Ask what they’d refuse to do without evidence. A real answer here is rarer than it should be.
- Be skeptical of guaranteed rankings, guaranteed recovery dates, and anyone who diagnoses your site before looking at it. Delivery dates are fair game to ask for, since those are inside an agency’s control. Results dates are not inside anyone’s.
Apply all three to us.
Frequently Asked Questions
How long does recovery take?
It depends on the cause, on how much Google has to re-crawl and re-process, on your site’s scale, on what competitors do next, and on whether the diagnosis was right. Google’s published guidance is that effects may appear within days or may take several months, and that some site-level improvements may not fully register until a later core update. Technical fixes tend to be at the fast end because they’re mechanical. Anything involving content quality or competitive position is slower, because Google is re-evaluating a judgment rather than reprocessing a file. We don’t assign a date until we know the failure mode.
Can I recover traffic lost to AI Overviews?
Partly, but usually not through ranking improvement alone. Confirm the cause first: if the affected query positions held while clicks fell, you’re looking at click substitution rather than a ranking loss. Find which queries are genuinely losing clicks to an AI Overview instead of assuming commercial searches are safe. Then ask a harder question about those queries: are they still worth the effort? Some are. Where they are, the work is earning the citation, giving people a reason to click your name specifically, and publishing what a summary can’t replace: original data, real experience, judgment. Where they aren’t, the honest answer is to move the effort to queries where people still have a reason to visit. Either way, change what you measure, because clicks alone stopped describing your visibility.
Should I rewrite the page or update it?
Depends on whether there’s a foundation worth keeping. If the page has no real author, no first-hand knowledge, and misses most of what the query needs, editing it is renovation on a bad structure. If it’s solid but stale, update the data, add what’s missing, make the authorship clear. Ignore anyone who gives you a word count or a position-drop number as the deciding rule. Those thresholds are invented, and they’ll have you rewriting pages that needed a paragraph.
Is the drop temporary or permanent?
Watch the affected pages and queries for at least several days, and compare what you see against confirmed updates, changes you shipped, seasonal demand, and competitor movement. Continued swinging tells you the result is unstable. It doesn’t tell you why, and it isn’t evidence that Google is running a test, which is a mechanism nobody outside Google can observe. A lower position that holds tells you this isn’t a one-day fluctuation, which is useful and is also not a diagnosis. Either way the cause has to be worked out separately. If a lot of competitors moved at the same time, a broader update, a demand shift, or a SERP change all become more plausible. If you dropped while everyone else held, start with your own changes and your own technical state.
Why did my impressions drop in September 2025 while clicks held?
Around mid-September 2025, Google stopped supporting the num=100 parameter that many rank trackers used to pull 100 results per query. Properties across the industry saw impressions fall at the same time, and the leading explanation is that automated deep-result requests had been contributing impressions that then vanished from reporting. Google didn’t publish a rule tying every September impression decline to that change, so treat this as a strong hypothesis rather than a confirmed mechanism. The test is everything else: if clicks, conversions, positions on your leading queries, and analytics traffic all held steady, a reporting change explains what you’re seeing better than a ranking loss does.
Do I need an agency for this?
Often not. Most drops are diagnosable and fixable by someone willing to work through it methodically. We publish the framework because you should be able to judge our diagnostic method before you grant anyone access to your data or pay for implementation. Work the steps yourself when you can reach the data, isolate your changes, deploy fixes safely, and measure what happens. Bring in help when any one of those breaks down. The “When It’s Past DIY” section lists the conditions where it’s genuinely past that point, and explains what we do about them.
About the Author: Nick Rizkalla is the founder and SEO lead at Rank Nashville, with over 14 years of experience in search strategy. He leads campaigns across the medical, legal, and e-commerce sectors, and can be reached on LinkedIn.