Six months with genuinely no movement means something is wrong. In order of likelihood: the wrong queries were targeted, the corroboration work never really happened, a technical blocker was never found, or the measurement is missing the movement. All four are diagnosable, and none is a reason to accept vagueness.
The short version
- Six months flat is a real signal. Do not accept “it takes time” as the answer.
- Most likely cause: the target queries were too competitive or too broad to win.
- Second: off-site corroboration was promised and never actually delivered.
- Third: a technical blocker nobody found, often crawler access.
- Fourth: it did move and your measurement cannot see it.
Work through them in this order
1. The wrong queries. The most common cause by a distance. If the programme aimed at broad category terms owned by large publishers and established competitors, six months was never going to be enough and probably never will be. The fix is not more effort. It is narrowing to queries where you have a genuine structural advantage: your specialism, your sector, your region, the specific problem you solve better than anyone.
Ask: what was the target query set, and who else is cited on it? If the competitors are national brands and you are a specialist firm, the strategy was wrong from month one.
2. The corroboration never happened. Off-site authority is the slow, expensive, unglamorous half, and it is the first thing that quietly disappears when an agency is stretched. Content keeps being published because it is visible and easy to report. Third-party placement stops because it is hard.
Ask: list every third-party mention, placement or citation earned in six months. If the answer is a number you could count on one hand, you funded content and not authority.
3. A technical blocker. Less common but occasionally the whole explanation. A WAF returning 403 to AI crawlers, main content requiring JavaScript, a robots rule nobody re-read, a migration that broke something. These do not surface in any content report.
Ask: show me AI crawler fetches from the server logs for the last 30 days. Not robots.txt.
The logs.
4. The measurement is wrong. Genuinely possible and worth ruling out before concluding failure. If reporting is CTR-led, an Overview answering in place looks like nothing. If AI referrals land as unassigned traffic, the growth is invisible. If no baseline was captured at the start, there is nothing to compare against.
Ask: what was the day-one baseline, and what does citation share look like now on the same query set? If nobody recorded a baseline, that is its own finding.
How to check it yourself
Run your ten most commercially important questions across ChatGPT, Perplexity, Gemini, Claude and Google AI Overviews. Record who is named. Then compare against the query set your agency has been reporting on. If those two lists are not substantially the same, you have found the problem yourself in twenty minutes.
What to do about it
- Demand the four answers above in writing. A capable agency can produce all four quickly. One that cannot is telling you something.
- Re-scope to winnable queries if cause one applies. This is usually the whole fix.
- Fix the measurement before judging the work, if cause four applies.
- Set a decision point with a specific metric and date. Vague continuation is how another six months disappears.
- Be fair about the starting position. A brand new domain with no corroboration was always a longer job than an established firm with weak structure. That is a scoping failure at the outset rather than a delivery failure now, but it changes what you should expect next.