SEO agency red flags, ranked by how badly they end

How to rank a red flag

Most red-flag lists are a flat pile of grievances — some catastrophic, some cosmetic, all presented at the same volume. That's not useful when you're on a call trying to decide whether to walk.

The test that sorts them: does this thing predict a failed engagement on its own, or is it just evidence somebody has bad taste? An agency with a cluttered website and a founder who says "bandwidth" too often can still do excellent work. An agency that won't put Search Console in your name cannot, structurally, ever prove it did.

Here are twelve, ranked. The ones at the top are close to deterministic. The ones at the bottom you can safely note and move past.

Tier 1 — the four that should end the meeting

These four share a structure: each one removes something you'd need in order to hold the agency accountable, or introduces a risk you can't see and can't undo. None of them is a matter of taste.

• Guaranteed rankings. Nobody controls Google's index, and Google's own guidance on choosing an SEO says as much in plain language. There are only two ways to promise a position: promise it for a keyword nobody searches, or buy links. The first is theatre, the second is a liability transferred to your domain. A guarantee against *your own baseline* is a different thing entirely and is worth asking for — see whether SEO agencies guarantee rankings.
• They keep the accounts. Analytics, Search Console, the Google Business Profile, sometimes the CMS. It's always framed as convenience. Its function is that when the relationship ends, so does your history — and while it lasts, you only ever see the version of the data they choose to export.
• Links from a rate card. If there's a per-link price, a turnaround time and a DA filter, those links were bought. You inherit the risk, not them, and the consequences arrive long after the invoices are settled. Ask them to name three placements from last quarter. Real digital PR people answer instantly and enjoy it.
• Nobody will name the team. "An account manager will be assigned during onboarding" means the senior people in the pitch are the sales function. This is the most common complaint founders bring us about a previous agency, and it's entirely preventable by asking one question before you sign.

Tier 2 — fixable, but only in writing, and only before you sign

These four aren't disqualifying. They're the terms of the deal, and they're negotiable exactly once. After signature, every one of them becomes a request you'll be making from a position of no leverage.

1. No baseline. Insist on one number, frozen, before work starts: trailing 90-day qualified leads from organic is the honest choice. Without it, month nine becomes a conversation about vibes, and vibes always favour the party sending the invoice.
2. Reports that stop at rankings. A report leading with average position and sessions can look excellent through a completely flat year — particularly now that impressions and clicks have decoupled. Require leads and revenue on the first page. Also fix the reporting keyword list at the start, because a list that quietly changes each month is the most common way an SEO report is made to look good.
3. Long lock-in with no exit. Twelve months committed with 90 days' notice is a fifteen-month decision made on a forty-minute call. Ask for a short committed period then month-to-month with 30 days' notice. Agencies confident in their work concede this quickly; the reaction is more informative than the term.
4. Scope as a shopping list. "8 articles, 15 links, 2 hours of technical" tells you what you're buying and nothing about why. Ask what happens if the audit reveals the real problem is a broken faceted navigation eating your crawl budget — does the scope bend, or do you get eight articles anyway?

Tier 3 — the four that smell bad and predict almost nothing

These get repeated a lot, mostly by agencies about other agencies. Treat them as weak evidence, not verdicts.

Cold outreach about errors on your site. Lazy and mass-mailed, yes. It also occasionally comes from a competent shop with a thin pipeline. Judge the reply, not the opener.

A free audit that's a tool export. This tells you their sales process is automated. It does not tell you their delivery is. Ask what the top three fixes would be and why — if the answer is specific to your site, the automated PDF was just the doorbell. Why free SEO audits are sales pitches covers the pattern honestly, including the bit where ours is one too.

Their own site ranks for nothing. The most over-read flag on the list. Agency sites are chronically neglected because client work always wins the sprint, and a genuinely busy shop often has a two-year-old homepage. Ask what they'd fix on their own site and listen to whether the diagnosis is sharp.

Trademarked methodology and dense jargon. A proprietary framework with a symbol after it is a marketing decision. It correlates with a certain kind of agency, but plenty of good operators have terrible brand instincts. Ask them to explain it without the diagram.

What actually predicts a good engagement

Inverting the list is more useful than memorising it, because you can check all six of these on a single call.

They can tell you within ten seconds how their hours split across technical, content, links and analysis — because they've planned it. They ask you what you need before they tell you what they do. They name a type of client they're wrong for. They volunteer a baseline before you ask for one. They can list three link placements from last quarter by publication name. And they answer "what happens if this doesn't work" with something written down rather than with "SEO takes time".

Our own answer to that last one: we freeze your trailing-90-day qualified leads from organic search on day one, and if we haven't beaten it in 90 days we keep working free until we do. Never a specific ranking position for a specific keyword, because nobody controls Google's index. It's also why we cap intake at three clients a month — that risk doesn't survive volume. The full set of questions worth asking pairs each one with the answer that should end the meeting.

You're already eight months in and reading this. Now what?

Switching costs more than people expect — new onboarding, a fresh learning curve, and usually a month of nothing. Run this sequence before you decide, in this order.

1. Take back access first, while the relationship is still cordial. Analytics, Search Console, Business Profile, CMS, domain registrar. Do this before any difficult conversation, not after.
2. Export everything. Search Console data expires on a rolling window. Pull the last sixteen months of query and page data to a spreadsheet you own.
3. Reconstruct a baseline retroactively. Not ideal, but Search Console and your CRM will let you build a trailing-90-day organic lead figure for the month you started. Now you have the number nobody wrote down.
4. Pull the backlink profile and look at where links came from. Patterned anchors, unrelated foreign-language sites and a sudden spike in referring domains all point at bought links, which changes the exit plan considerably.
5. Set one 90-day test, in writing. One number, one date, one review meeting. Give them the chance you'd want. Most agencies raise their game measurably when the conversation stops being about deliverables.
6. Then decide. Whether and how to fire your SEO agency covers the mechanics, including what you're entitled to take with you.