The standard monthly SEO report has not changed in eight years. Keywords ranked. Keywords improved. Estimated monthly traffic. A small chart of organic sessions. A list of top-performing pages. A nod toward backlinks. Roll forward to next month.
The buyer reading the report has changed completely. They are not asking how many keywords moved. They are asking why their phone is not ringing. The keyword-rank metric stopped correlating tightly with revenue around the same time the AI overviews started eating the top of the SERP, and nobody on the agency side updated the scoreboard.
The buyer’s research path has shifted measurably. Pew Research reports that 34% of US adults have used ChatGPT, double the share two years earlier, with 58% of adults under 30 already in the user base. Meanwhile BrightLocal’s 2026 Local Consumer Review Survey places consumer trust in AI platforms for business recommendations at 45%. The keyword report is grading the channel the buyer is using less and ignoring the channel they are using more.
We rebuilt our monthly report from scratch two years ago. The keyword chart is gone. The “estimated traffic” line is gone. The “domain authority” pseudo-score is gone. What we hand back instead is one page, five rows, no charts the buyer needs a manual to read.
What we report instead.
For every client, we hand back a one-page monthly that has five rows. None of them are keyword counts.
Citations across the AI platforms. We run a fixed prompt set every Monday across ChatGPT search, Perplexity, Claude, and Gemini. The prompts are the queries the buyer would actually type. For each prompt, we log whether the client’s domain was cited, which page was cited, and what position it held in the answer. The monthly is the rollup. A client moves from “cited in two of forty prompts” to “cited in twenty-three of forty prompts” and the conversation about ROI becomes very different.
Reviews and review velocity. Total Google reviews, total reviews on the second-most-relevant platform for the vertical (Yelp, Avvo, Healthgrades, Houzz, whichever), and the rate at which new reviews are coming in. New reviews per month is the number that moves rank in local pack and that moves citation rate in the AI platforms. Total review count is a vanity stat above a certain threshold. Velocity is the metric. The recency requirement matters more every year. BrightLocal’s 2026 survey reports that 74% of consumers only care about reviews written in the last three months. A business with 800 reviews from 2019 and zero from the trailing quarter is invisible to the buyer and to the AI platforms reading the same review feed.
Branded-search volume. How many people typed the client’s actual business name into Google last month. This is the only number on the report that captures the cumulative effect of every channel running together: paid, organic, AI, social, referral. If branded search is climbing, the marketing is working. If it isn’t, something is wrong upstream even if the keyword-rank chart looks good.
Inbound by channel, attributed. Phone calls, form fills, and SMS conversations, broken down by source. We instrument the site so every inbound is tagged with the channel that brought it. The client gets a row for organic, a row for AI-platform referrals, a row for paid, a row for direct, and a row for referral. The agency that won’t show you this either doesn’t have the instrumentation or is hiding what the numbers actually say.
One forward-looking commitment. What we are shipping next month. Specific page rebuilds. Specific schema deploys. Specific content. Specific automations. Not “continue to optimize.” Not “monitor performance.” A list of dated commitments with names against them.
How the prompt set is built.
The citation row is only as honest as the prompt set behind it. Most agencies that have started reporting on AI citations are using a single hand-picked prompt that the client happens to be cited on, and calling it good. We do not do that.
For every retainer client, we build a forty-prompt set during onboarding. The prompts come from four sources.
Buyer-search transcripts. What did your last fifty inbound callers say when we asked them how they found you. Real language, not marketing language. “I was looking for someone who could do a panel upgrade on an older house” produces a different prompt than “best electrician in [city].”
Competitor coverage gaps. Prompts your top three competitors are cited on that you are not. We pull these by running each competitor’s domain against our prompt library and recording where they show up. The delta is your immediate citation target.
Service-page coverage. One prompt per major service you offer, phrased the way a homeowner or small business owner would actually ask. Not “HVAC services” but “how much does it cost to replace an old furnace.”
Long-tail buyer questions. Specific high-intent questions like “do I need a permit to replace a water heater in [state].” These rarely have high search volume on Google but they have very high conversion when cited.
The set is locked at the start of the engagement and only updated quarterly with the client’s sign-off. The discipline matters because moving the prompt set mid-engagement makes the numbers unreadable. The forty-prompt benchmark stays fixed so the month-over-month delta means something. The fixed prompt-set discipline is documented in detail under our editorial content practice where the prompt-to-page mapping work lives.
What good monthly movement looks like.
Citation rate moves in a predictable pattern when the underlying work is real.
Months one and two: roughly flat. The platforms have not re-crawled the changes yet. The dashboard shows movement of zero to two new citations.
Months three and four: the first wave. The fastest-crawling platforms (Perplexity, ChatGPT search) start citing new source pages. Typical movement: three to eight new citations across the forty-prompt set.
Months five through eight: the broader lift. Claude and Gemini catch up. Citation rate moves into the twenty-to-thirty range on a forty-prompt set for clients running a serious AI-optimized website rebuild alongside the content work.
Months nine and beyond: durability. The platforms refresh continuously, so citation rate can fluctuate week to week as competing pages get published or refreshed. The monthly average smooths it out. Clients on a continuous retainer typically hold thirty-plus citations out of forty by month twelve.
If the citation rate is not moving by month four, something is wrong. Most often it is that the schema is missing or that the technical foundation is too slow for the AI crawlers, which is why we run the Lighthouse Local audit at month three as a diagnostic check.
Why this matters.
The reason the keyword-rank report has stayed alive is that it is hard to argue with. Keywords go up, the agency points at the chart. Keywords go down, the agency points at an algorithm update. Either way, the client has no way of telling whether the work is paying back.
Citations are different. Citations come from a fixed prompt set the client can read. Reviews are different. Reviews are visible to anyone with a browser. Branded search is different. The client can verify it inside Search Console in eleven seconds. Inbound is different. The phone either rings or it doesn’t.
The five-row format does not just describe what is working. It also makes it impossible to hide what is not. There is no chart to obscure the inbound number. There is no “but our domain rating climbed” footnote to bury a revenue miss. If the numbers were soft, the email reads as soft. The conversation that month is honest, fast, and forward-looking.
What goes wrong when the scoreboard slips.
Three patterns show up in the report before the rest of the engagement starts breaking.
Citation rate flatlines while branded search is climbing. The marketing is working but the content is not built to be cited. The fix is a source-page rebuild on the highest-value service pages. The work usually pairs with our content practice.
Inbound by channel concentrates on paid while organic and AI inbound are flat. The retainer is being floated by media spend. When the paid budget pauses, the operation goes dark. The fix is to harden the organic and AI side, which is exactly the work our national SEO program is built around.
Review velocity drops below the trailing three-month average. Usually a process problem at the client end, not a marketing failure. We flag it in the report and route a review-system check to make sure the post-visit prompts are firing through the existing CRM. Standard work across every service-area industry we serve.
The point of the scoreboard is to surface these patterns in month three or four, while there is still time to course-correct. The keyword report does not show any of them until month twelve, when the contract is up and the damage is already done.
Frequently Asked Questions.
How is the citation row counted, exactly?
Same prompt set, same four platforms, same Monday, every week. We log raw citations and the page that was cited. Month-end is the average across the four weeks. We do not cherry-pick prompts to make the number look good.
What if a client is cited on a prompt that is not in the set?
It still counts toward the trend, but only the locked prompt set drives the headline metric. The discipline matters because rotating prompts makes deltas unreadable.
Why not just count Google rankings too?
We report them as a sixth row when the client cares, but we lead with the five. Most clients stop asking about keyword rank by month four because the citation row gives them a cleaner answer.
Can an agency white-label this scoreboard?
Yes. The agency partnership program covers the same reporting format under the partner’s brand. The dashboard runs inside Lighthouse Local with the partner’s logo and domain.
Where does the email template live?
Documented in detail in our piece on the five-row monthly scoreboard, with examples of how each row reads in practice.
The takeaway.
Run any agency’s monthly report against the five rows above. The ones that pay back will start every meeting at that same five-row scoreboard. The ones that don’t will keep handing you keyword charts.
