I run ImageFormatConverter.in — a small free site where people compress photos for SSC forms, NEET uploads, job portals, and random everyday tasks. Nothing fancy. Just tools that need to work on a phone with bad network.
For a few months, Google Search was actually kind to me.
Impressions kept climbing. At the peak, I was seeing around 4,000 impressions per day in Search Console. Clicks were modest — maybe 30–35 on a good day — but the trend was up. That was enough to keep me going.
Then it fell off a cliff.
My Search Console chart over 3 months: growth through mid-June, then the cliff.
The morning I knew something broke
I have a habit of opening Search Console before breakfast. Stupid habit, but it kept me motivated when numbers went up.
Around late June, the chart stopped looking like growth and started looking like a mistake.
June 16–23 was still normal. Impressions hovered between 3,000 and 5,000 per day. Clicks were there. Average position was decent — around 8–11 depending on the week.
Then June 24 happened.
By June 26, one day showed 3 impressions and 0 clicks. I stared at that number longer than I want to admit.
Zoomed 28-day view — you can see the drop on June 26: 3 impressions, 0 clicks.
There was a weird spike again around July 2 — impressions jumped toward 6,000 for a day — and then nothing. Flat line. The last 28-day view in my account shows about 39,600 impressions and 240 clicks, but most of that is from the good weeks before the drop. Lately it feels like the site barely exists in search.
Google Analytics tells the same story in a different language. Active users down 62%. Views down 64%. Same site. Same tools. Very different traffic.
GA4 told the same story — active users down 62.8%, views down 64.1% compared to the previous week.
I did not get a manual penalty email. No scary red banner in Search Console. Just silence.
What I changed right before the drop (honest list)
I tried to remember the order of things, not just the list. Timing matters.
In a short window around mid to late June, I did a lot at once:
- Published many blog posts in a hurry, several with AI drafts I edited lightly.
- Changed metadata, titles, and page layouts across tools.
- Added internal links everywhere — including links to a sister site I also run.
- Put AdSense on the site because I thought traffic was strong enough to monetize.
- Deployed multiple versions quickly through Vercel without always committing the same changes to git.
- Rewrote older posts instead of leaving stable URLs alone.
I told myself: more SEO work equals more traffic.
That belief cost me.
Looking back, the sister-site links and AdSense on core tool pages were especially dumb moves. People came for a 20KB compressor or 100KB photo tool, and I was sending them elsewhere or loading ad scripts on pages that had finally started ranking. I reversed those changes later. Traffic did not bounce back the next day. That part hurt more than I expected.
What I checked when I panicked
I went through the usual checklist because that's what every SEO article says to do.
Manual actions? None.
robots.txt blocking the site? No. Still allows crawling.
Sitemap missing pages? Sitemap was updating. New posts were in the XML.
Broken 404 waves? Some broken internal links showed up in audits later, especially around PDF paths and old aliases. I fixed those.
Canonical tags pointing wrong? Found duplicate tool URLs — same compression feature living at two paths. That kind of thing eats impressions quietly.
Vercel hosting multiple sites on one account? I worried about this. Research says hosting itself is not a ranking factor. Fair. But messy deploys on my side definitely were.
So it wasn't one smoking gun. It was a pile of changes landing while Google was still figuring out what the site was.
The blog problem I did not want to admit
The hardest part was opening my own content folder.
I had posts for 20KB, 50KB, 100KB, 200KB, WhatsApp compression, exam forms, KYC, PNG vs JPG — you name it.
Individually, each article looked fine.
Put three of them side by side and the skeleton was identical:
intro → quick answer → why it matters → steps → FAQs → conclusion.
Swap the keyword. Ship it. Repeat.
I used AI to speed up drafting. I still do sometimes. But Google does not care whether a paragraph came from a keyboard or a model. It cares whether the page teaches something this specific person needed and could not get from the last four posts on the same site.
I was building volume. Not memory. Not trust.
There's a difference between a guide that helps someone pass a 20KB signature upload on a government portal and a page that merely mentions "compress image to 20KB" twelve times. I had too much of the second kind.
What I stopped doing
I made a rule for myself in July: slow down.
- Fewer new posts. Better ones.
- One technical fix at a time, then wait and watch Search Console.
- No sister-site links on money pages. Full stop.
- No AdSense until traffic is stable for weeks — I removed it.
- Update existing guides instead of spawning near-duplicates.
- Keep canonical URLs clean for core tools like Image Compressor to 20KB and Compress Image to 100KB.
I also rolled the codebase back to an earlier stable deploy at one point because I wanted a known-good baseline. That alone did not restore impressions. Recovery is not a undo button. It's crawl trust rebuilt slowly.
What I am doing now (recovery plan)
This is the part still in progress. I am writing it in public so I cannot lie to myself later.
Technical cleanup first. Redirects consolidated. Broken blog links fixed. Sitemap checked. IndexNow submitted after new posts. GSC URL inspection for important pages — yes, manually, because Google ignores IndexNow.
Content second. I am writing fewer articles like this one — actual experience, real numbers, things I messed up. If I publish a tool guide, it should read like someone used the tool at 11 PM before a form deadline.
Patience third. Search Console updates lag. A new blog can be technically perfect and still sit unindexed for days when the site's crawl rate is low. That is where I am now. Not blocked. Just queued behind my own earlier mistakes.
If you are building image tools for India, the pages that still matter most on my site are the ones users actually need:
- Image Compressor to 20KB for signatures and tiny uploads
- Compress Image to 50KB for many exam photos
- Compress Image to 100KB for job portals and KYC
- Custom Image Compressor when the form shows a random KB limit
I am focusing repair energy there instead of chasing every long-tail variant with a duplicate page.
If your impressions just collapsed
I do not have a magic fix. I have mistakes you can skip.
Before you push another "SEO improvement" live, ask:
- Am I changing ten things at once without a way to measure which one hurt me?
- Did I publish five articles that are really one article in different KB sizes?
- Am I sending users away from the page that was starting to rank?
- Would I bookmark this post, or just bounce after ten seconds?
Those questions slowed me down. That slowdown might be the most useful SEO decision I have made since June.
This post will change
I am not writing a success story yet. I am writing a mid-crisis diary.
If impressions recover — even to 500 a day first, then 1,000 — I will update this page with dates and what actually moved the needle. If something else fails, I will add that too.
The goal is not to game Google back to 4,000 impressions per day with the same thin playbook that broke the site.
The goal is to earn it back with tools and writing that a real person in India would recommend to a friend stuck on a form upload.
If you are in the same situation, I hope this saves you a week of random changes. Check your redirects. Read your own blogs out loud. Fix one thing. Wait. Then fix the next.
That is where I am tonight.
— Rajat