Skip to content
Get Ranked On TopIllustrative case studyContent Engine plan

A fictional business. The figures are a modelled scenario, not a client result.

Gridlark Analytics · Data-pipeline analytics SaaS, 120 staff · Austin

How a data-pipeline SaaS rewired its trend-chasing blog so engineers' questions reach it, not its rivals

+186% organic trial sign-ups, named in 41% of AI answers

18 months · North America · Content Engine plan · Topic clusters · Content pruning · AI-search visibility · Original research

Exchange status · month 1841%of tracked AI answers now name Gridlark, up from 6%

§ 1The trouble on the line

Engineers kept calling. Someone else picked up.

Its blog chased trends with no topic structure, and when engineers asked AI assistants how to monitor pipelines, the answers named competitors.

Gridlark sells pipeline observability to data teams, and its buyers research for weeks before they book a demo. The blog published three times a week, but on whatever was trending: conference recaps, hot takes on the latest AI launch, listicles. Nothing tied one post to the next, so search engines never saw Gridlark as an authority on anything. Then engineers started asking AI assistants instead of searching, and the assistants, quoting competitors' how-to guides, put them straight through to someone else.

The blog, before · 310 posts by type
  • Trend & news takes
  • How-to guides
  • Product updates

§ 2Fault report

Four faults on the exchange.

Dead lines

47%

146 of 310 posts had not earned a single organic click in twelve months. Most were reactions to that week's news, stale within a month.

Fault 1 of 4

Crossed wires

38 pairs

38 pairs of posts competed for the same query. Nine different posts targeted “data pipeline monitoring”, and not one of them reached page one.

Fault 2 of 4

No switchboard

0 hubs

No pillar pages and no internal-linking plan. Posts linked to the homepage and to pricing, almost never to each other, so no topic ever built up authority.

Fault 3 of 4

Calls ringing out

6%

Across 200 prompts engineers really ask, AI assistants named Gridlark in 6% of answers. They cited competitors' step-by-step guides because Gridlark had none to cite.

Fault 4 of 4

§ 3Operating procedure

Four procedures to wire the board.

Built on our Content Engine plan: the content and AI-search engagement for companies whose buyers research long before they talk to sales.

HUB
Hub and spoke · the pipeline-monitoring clusterEach spoke answers one question in full and links back up to the hub; the hub links down to every spoke. Search engines and AI assistants read that structure as depth.
  1. 3.1

    Topic clusters

    Six hub pages, one for each job an engineer needs done: monitoring, data quality, lineage, alerting, warehouse cost and schema changes. Around them, 84 spoke pages that each answer one specific question in full, mapped from 1,900 queries, support tickets and sales-call notes. Every spoke links up to its hub, and every hub links down to all its spokes.

  2. 3.2

    Content pruning

    142 posts were merged into stronger pages or retired with a 301 to the closest spoke. The 31 best were rewritten as spokes, and the nine competing “pipeline monitoring” posts became one page that finally ranked.

  3. 3.3

    AI-search visibility

    Every spoke opens with a two-sentence direct answer, then steps, code and a comparison table: the shape AI assistants quote. Named authors with real bylines, TechArticle and FAQPage schema, and a monthly check of 200 prompts across four assistants.

  4. 3.4

    Original research

    A yearly Pipeline Reliability Report, built from anonymised incident data across 900 pipelines. Its figures are the ones other writers and AI answers now quote, and every citation leads back to a hub.

§ 4The switchboard

Plug each question into the page that answers it.

On the left, twelve calls engineers place every day. On the right, six hub jacks. Choose a call, then the hub that should answer it, and the cord connects it to that hub's spoke page. Every call left unplugged rings out to a competitor.

Incoming calls

Hub jacks

Answered by Gridlark
0 / 12
Ringing out
12 / 12
Clusters complete
0 / 6
AI mention rate (modelled)
6%

What an AI assistant answers

Put a call on the line to hear the answer.

Esc clears the line.

Choose a call to put it on the line.

§ 5Shift log

18 months, line by line.

  1. Months 1–2

    Listen in

    Crawl of all 310 posts, a year of Search Console data, a 200-prompt AI baseline and 1,900 questions pulled from search, support and sales calls.

  2. Months 3–4

    Draw the board

    Six hubs and 84 spokes mapped, cannibalising pairs resolved, the merge-and-redirect plan signed off with the product team.

  3. Months 5–8

    Wire the hubs

    Hubs and the first 40 spokes live, 142 old posts merged or retired, internal links rebuilt from hub to spoke and back.

  4. Months 9–13

    Fill the lines

    The remaining 44 spokes, the first Pipeline Reliability Report, author pages and schema across the cluster.

  5. Months 14–18

    Keep the exchange

    Monthly prompt checks, refreshes for the 20 spokes that slipped, the second report and a webinar series built on the hubs.

§ 6Results

Now most calls are put through to Gridlark.

Organic trial sign-ups / month

+186%

210 → 600 a month

AI-assistant mention rate (200 prompts)

6% → 41%

Share of 200 tracked prompts, across 4 AI assistants

Non-branded organic clicks / month

9,800 → 31,500

Per month, quarter by quarter

Topic clusters wired

6 hubs · 84 spokes

31 rewritten from old posts, 53 new

Modelled scenario for a fictional business — not a client result. Rankings, traffic and AI-assistant mentions are never guaranteed; see our Terms.

Organic trial sign-ups and AI mentions, by monthTrial sign-ups / monthAI mention rate
02004006000%20%40%60041%M1M4M7M10M13M16M18

Tracked prompts: who the AI answer put through

Prompt, as typedHubAI answer · month 1AI answer · month 18Google rank
how to monitor airflow dag failuresPipeline monitoringnot named#1 of 3#24 → #2
data pipeline monitoring toolsPipeline monitoring#4 of 4#1 of 4#17 → #3
data quality tests in dbtData qualitynot named#2 of 3#31 → #4
column level lineage explainedData lineagenot named#1 of 2#22 → #1
reduce alert fatigue data teamAlerting & on-callnot named#1 of 3#40 → #3
why is my warehouse bill so highWarehouse costnot named#2 of 4#28 → #5
handle schema drift in streamingSchema changes#3 of 3#1 of 3#15 → #2
upstream schema change broke my pipelineSchema changesnot named#1 of 2#36 → #4

“#1 of 3” means Gridlark was the first of three tools the answer recommended, checked across four AI assistants. Google rank is the matching spoke page's position for the same query, month 1 → month 18.

Get a plan like this

Is your blog busy but your pipeline quiet? Start with the Content Engine plan, or tell us which questions your buyers ask before they call.

← All case studies