Straight from the systems the work touched
CRM, analytics, ad accounts, search and call recordings, pulled on a schedule. Nobody types a number in, and a figure we cannot see is marked as missing, never estimated.
Outcomes
Know what is working, and why.
Every system your business runs on feeds the Company OS. It measures each piece of work against what it was meant to move, sends you the report, and writes the lesson back into the company brain so the next piece starts better.
Your sources
GA4Search ConsoleGoogle AdsHubSpot CRMLinkedInSemrushCalls and meetingsAI answers
Company OS assesses
Company brainTargets, past results and what worked before
MeasurePulled from each system on a schedule, never typed in
AssessAgainst the target and the last run, with gaps marked
Lessons written back into the brain
Your reports
Click a report to open it. Each one is rebuilt with demo data for a fictional company.
How it works
Measure, assess, feed back.
Outcomes is the layer that checks the work. It does not produce anything new; it tells you what the work did, and teaches the rest of the system from it.
Against the target and the last run
Each result is read against what it was meant to move and the run before it, so you see what moved, what did not and the biggest lever left.
Into the company brain
A winning angle, a new objection, a page that stopped working: the lesson is written back where every agent reads it, so the next piece of work starts from it.
What we measure
The numbers each team actually decides on.
Only the workstreams we run, and every metric tied to the decision it informs.
Sales
- Booked meetings and where they came from
- Connect rate and calls per meeting
- Pipeline by lead type
- Deals opened from outbound
Marketing
- Demo and call requests
- Cost per lead and per demo
- Buyer searches won against the plan
- How often AI assistants recommend you
Operations
- Time from request to first draft
- Work waiting on an approval, and for how long
- Hours handed back to the team
- Errors caught before anything ships
Leadership
- Adoption by team and by workflow
- Spend against what it returns
- Progress against the roadmap
- What the system learned this month
Reporting
Reporting from every source.
Each report reads the systems the work touched, on a schedule, and says what moved, what did not and why. These are the real formats, rebuilt around a made-up company so no client's numbers are on show.
Pulls fromGA4Search ConsoleGoogle AdsHubSpotLinkedInChatGPTClaudeGeminiPerplexitySemrushPageSpeedAutocompleteWebflowGmailDriveNotion
Every report here uses demo data for Halcyra, a fictional company. The sections, checks and layouts are the ones clients receive, in their own brand.
Halcyra weekly: 12 October 2026
Demo data · fictional companyHeadline
MovedDemo requests doubled to 6, four from the new pricing page.
Against 3 last week. Every one came from organic or direct visits, none from paid.
Did not moveThe clinic roster checklist held at 21 downloads.
Steady for a third week. Most readers arrive from the newsletter, not from search.
Biggest leverTeach Google Ads to bid on booked demos.
The account spends AUD 1,180 a week learning from downloads instead of demos.
High-intent leads Asking to talk
6Demo requestsLast week 3. Checked against the sales calendar
2Free trial startsLast week 2
AUD 1,180Google Ads spendLast week AUD 1,240. 142 clicks
AUD 197Cost per demoAll channels. Last week AUD 413
Mid-funnel leads Raising a hand
Checklist downloads by channel
Source: GA4 and HubSpot form submissions by session source.
| Intent signal | This week | Last week |
|---|---|---|
| Rostering software searches Impressions · clicks, Search Console | 2,940 · 61 | 2,610 · 48 |
| AI assistant visits GA4 | 14 | 9 |
| Target clinics visiting HubSpot | 7 | 4 |
Actions In order of impact on demo requests
| # | Action | Owner | Waiting |
|---|---|---|---|
| 1 | Import booked demos into Google Ads Until then the account optimises for downloads. | Skein | 7 days |
| 2 | Add a demo link to the pricing page header Four of six demos started on this page. | Halcyra | 14 days |
| 3 | Fix the trial form on mobile The submit button sits under the cookie bar on small screens. | Skein | New |
| 4 | Approve the two case study quotes Both customers have signed off. Waiting on the client review. | Halcyra | 21 days |
Sources: GA4, Search Console, Google Ads, HubSpot. Pulled 12 October 2026. A figure we cannot see is marked, never estimated.
SDR performance and insights
Demo data · fictional companyTakeaway. Three meetings this week, all with clinics that downloaded the roster checklist first. Since the start, signal leads book at six times the rate of cold lists.
Headline numbers As a funnel
Leads worked4123.1 calls each
Calls1,28418% connect
Connects2319% booked
Meetings2143% opened a deal
Deals opened95 past stage 1
This week
| Measure | This week | Last week | Change |
|---|---|---|---|
| Calls | 96 | 88 | +8 |
| Connects | 19 | 14 | +5 |
| Meetings booked | 3 | 1 | +2 |
| Deals opened | 1 | 1 | 0 |
How Riverbend Vet booked
Read: the checklist did the qualifying. One connected call was enough.
By lead type Cumulative
| Lead type | Contacts | Connect rate | Meetings | Per 100 called |
|---|---|---|---|---|
| High intent | 38 | 31% | 7 | 18.4 |
| Signal | 96 | 22% | 8 | 8.3 |
| Target accounts | 120 | 16% | 4 | 3.3 |
| Cold list | 158 | 11% | 2 | 1.3 |
Source: HubSpot calls, emails, meetings and deals owned by the SDR. Pulled 12 October 2026.
September: demos up, paid search still learning from the wrong thing
Demo data · fictional companyThree things to act on
1 · ActPaid search bids on downloads, not demos.
Changing the conversion goal is the one change most likely to move demo volume in October.
2 · KeepThe pricing page now converts.
Demo requests from it rose from 2 to 9 after the comparison table went live on 9 September.
3 · DecideAI assistants name two competitors first.
Visibility score 40 out of 100. Three comparison pages and an llms.txt are ready for approval.
The numbers that decide things Scorecard
| Metric | Sep | Aug | Jun | Change | Trend |
|---|---|---|---|---|---|
| Demo requests | 19 | 12 | 10 | Up 58% | |
| Free trial starts | 8 | 9 | 6 | Down 1 | |
| Organic clicks | 2,410 | 2,180 | 1,760 | Up 11% | |
| Google Ads spend | AUD 5,020 | AUD 4,860 | AUD 4,300 | Up 3% | |
| Cost per demo | AUD 264 | AUD 405 | AUD 430 | Down 35% | |
| AI visibility score | 40 | 33 | not run | Up 7 |
What happened, by area
Google Ads
Baseline audit complete. 64 negative keywords applied after approval and read back. Display network switched off.
Content
Four articles published from the plan. The roster template guide reached position 7 in its third week.
Eight posts approved and published by the founder. Two passed 10,000 impressions.
Time against estimate
46 of 50 hours
| Waiting on a decision | Waiting |
|---|---|
| Publish the three comparison pages? Unblocks the AI visibility fixes | 18 days |
| Who approves LinkedIn posts in November? The founder is away for two weeks | 6 days |
Sources: GA4, Search Console, Google Ads, HubSpot, LinkedIn. Every metric is tied to the decision it informs.
How often AI assistants recommend Halcyra
Demo data · fictional company40/ 100
Up from 33 in September. Accuracy is not the weak dimension; consistency is.
By assistant
Six dimensions Each scored out of 10
| Dimension | Score | Read | |
|---|---|---|---|
| Presence | 5 / 10 | Named in about half of category answers | |
| Position | 3 / 10 | Usually third or fourth when named | |
| Accuracy | 6 / 10 | Pricing model right, plan names out of date | |
| Sentiment | 5 / 10 | Neutral, described as a newer option | |
| Completeness | 3 / 10 | Integrations rarely mentioned | |
| Consistency | 2 / 10 | Different answers across assistants |
Three findings
Two competitors own the comparison questions
Asked for the best rostering software for a vet clinic, every assistant names them before Halcyra.
Old plan names are still quoted
Gemini and Perplexity cite a 2025 pricing page that now redirects.
The roster template guide is being cited
Perplexity links it in 6 of 12 answers about building a clinic roster.
On-site machine readability Observed, never scored
| llms.txt | Missing | Drafted, waiting on deployment |
| AI crawlers in robots.txt | Allowed | GPTBot, ClaudeBot, PerplexityBot, Google-Extended |
Answers captured 6 to 8 October 2026 from ChatGPT, Claude, Gemini and Perplexity. An unassessed dimension is left blank, never scored zero.
Fixes to complete before sign-off
The new site is a clear improvement. The homepage now shows its main content in 2.1 seconds on mobile, down from 5.9. Two templates still fail Google's layout stability test, and four launch settings are open.
Demo data · fictional companyWhat we measured
| Mobile | Old site | New site | Result |
|---|---|---|---|
| Homepage, main content visible | 5.9s | 2.1s | Passes |
| Pricing page, main content visible | 4.4s | 3.2s | Better, still over 2.5s |
| Feature pages, layout shift | 0.02 | 0.18 | Fails, new problem |
| Every page on desktop | 1.1 to 1.6s | 0.8 to 1.2s | Passes |
Fixes
1 · Feature pages jump when the hero image arrives CLS 0.18
- Where
- All eight feature pages, the hero image wrapper
- Cause
- No reserved height, so the page is laid out without the image, then pushed down.
- Fix
- Give the wrapper a fixed aspect ratio in the designer and set the image to cover.
2 · Nothing links to the pricing page Critical
- Where
- Not in the navigation, the footer or any page body
- Fix
- Add it to the navbar and footer components.
3 · The staging site is public and indexable Open
- Fix
- Turn on subdomain indexing protection, then publish to staging once.
Already fixed, thank you
Old feature URLs redirect in one hop. Every page has a correct canonical. Missing pages return a proper 404.
Google PageSpeed Insights, five runs per page, mobile and desktop. A change counts only where old and new ranges do not overlap.
The learning loop
Every result teaches the system.
Reports are not the end of the work. They are how the next piece of work starts better.
- A new ad performs worse than the last one: it is flagged, rotated out once you approve, and the lesson goes back into the company brain.
- Three prospects raise the same new objection: it is filed in the brain with the answer your best person gave.
- One cold email angle beats the other: the winner becomes where the next campaign starts.
- A figure we cannot see is marked as missing, never estimated.
Start