Bought 8 links on the Monday, traffic down 30% by the Friday. Talk me out of the obvious conclusion

Timeline first, conclusions after, because I'd rather be argued out of this than agreed with.

Monday 30th June: paid for eight links from a broker I've used twice before. Guest posts, four different sites, staggered placement over the following ten days as usual.

Wednesday 2nd: three of them went live. Anchors were a mix, one exact match, rest brand and naked.

Friday 4th: organic sessions down about 12% against the previous Friday.

Following Friday 11th: down 31% against the same baseline. Same pages, no drop in impressions on brand terms, the loss is almost entirely on four commercial pages that were sitting positions 4 to 8.

Nothing else changed. No deploy, no content, no redirects, robots.txt untouched, no crawl errors, server logs clean, no 5xx spike. I've checked the boring stuff because that's my job.

The obvious conclusion is that I bought eight links and got hit for it inside a week. IME the obvious conclusion is wrong about two thirds of the time, so: what's the other explanation?

Core update. Started the 30th of June, which is your Monday. It's still rolling as of this morning.

I'd stop looking at the links entirely until it finishes. Every single time one of these lands, this board fills up with people who changed something that week and are certain the thing they changed caused it. In 2019 it was HTTPS migrations, in 2021 it was page speed work, this year it'll be links because links are what everyone's spending money on.

The tell is your own data: brand impressions flat, four commercial pages at positions 4 to 8 losing ground. That's not what a link problem looks like. A link problem doesn't select for commercial intent pages and leave your brand queries alone. A core update does that constantly.

Check what's ranking above you on those four terms now. If it's the same four sites reshuffled, it's a query-level re-rank and you have a content problem. If it's sites you've never seen, it's the update.

eight links don't do anything in four days. not up, not down

Dates line up too neatly to ignore. What's the actual gap between the first link going live and the first drop?

Two days. Which is exactly the problem - it's too fast for links and about right for an update.

The two-day gap is the strongest piece of evidence you have and it points away from the links, for a reason worth spelling out.

A bought link has to be crawled, indexed, and then have the target page re-evaluated before it can affect anything. Even in the best case - a site that gets crawled hourly and a target page already in the index - you are looking at days for the link to register and typically weeks before it moves a ranking. That is the mechanism working in your favour. There is no faster path for it to work against you either, because the negative case runs through the same pipeline.

The exception, and it is a real one, is if the links were not new. If your broker placed you on pages that already existed and already had a batch of other buyers' links on them, and something happened to those pages, you can inherit an effect much faster. Worth asking whether the four sites are ones he's placed on before and how many outbound commercial links are on the specific pages you're on.

But the pattern you describe - commercial pages losing, brand flat, four days - is the classic shape of a core update, and there is one rolling right now. I would give it until it finishes, hold everything, and change nothing. The worst outcome here is that you disavow eight perfectly good links because of a coincidence and then spend six months wondering why nothing recovered.

down about 20 here since the 1st and i havent bought anything

hexforge's exception is worth taking seriously and it's the thing I'd check first, because I see it from the supply side.

A lot of brokers work from a fixed inventory of maybe forty sites and place everybody on the same ones. If you're on a page with nine other outbound links to commercial pages in unrelated niches, the page isn't a guest post, it's a link page with an article on top. Those get devalued in bulk and when they go, everyone on them goes at the same time.

Depends what you're running, but the check is quick: pull up the four live placements and count outbound links to external commercial sites on each page. Under three, you're probably fine. Nine or ten and you've been sold a slot on a farm, whatever it looked like in the pitch.

That still wouldn't explain a four-day effect. It would explain why this broker's links stopped working generally, which is a different and slower problem.

Counted. Two of the four pages have 2 external commercial links, one has 4, one has 11.

The 11 is a problem for a different thread. It's not this week's problem.

Update finished rolling last Thursday. What do your four pages look like now?

I'll say in advance what I expect: partial recovery on two, permanent loss on one, and one that never moved and you'll decide was the link. That's the distribution I've seen after every one of these since 2018 and I'd like to know if it holds.

Closing this out since I asked for the argument and I got it.

Two weeks after the update finished: two pages back to roughly where they were, one recovered about half, one is still down and hasn't moved. Sessions overall are down 9% against June rather than 31%. old_growth_seo called the distribution almost exactly, which is either experience or a self-fulfilling prophecy and I'm not going to be able to tell the difference.

The SERP for the page that didn't recover has two sites in it I'd never seen before. That's an update result, not a link result.

So: coincidence, and an expensive one in terms of the fortnight I spent looking at server logs. The genuinely useful thing I got out of this is byte_broker's outbound-link count - the page with 11 on it is one I've bought from twice and I'd never have thought to check.

FWIW the rule I'm writing down for myself: never buy links in the same fortnight as a confirmed update, purely so I don't do this to myself again.