Your own analyst — an AI solution for metrics and reports
Reads Stripe, Amplitude, Meta Ads, Git and your spreadsheets, watches the metrics, calculates cohorts and LTV and sends reports with charts — in Slack or Telegram.
Your own analyst from sviy.ai is an AI solution that watches a subscription product's metrics, calculates cohorts, LTV and plan-vs-actual, evaluates ads and answers data questions for founders. It works in Slack or Telegram with Stripe, Amplitude, Meta Ads and spreadsheet data, goes live in 2 days, and we send a priced offer by email within one business day.
offer within a day
live in 2 days
read-only data access
A
Your own analystmetrics · cohorts · reportsworking in Slack, Telegram and Google Sheets
Build the weekly report: MRR, churn, cohorts and ads — in the thread before the 12:00 meeting09:41
Preparing it in the agreed format: Stripe, Amplitude and Meta data for 29 Sep – 5 Oct. Should I put the full version in Google Sheets?09:42
Sound familiar?
Three situations that repeat every month
Every Friday I spend half a day building a report by hand from five spreadsheets.
Churn jumped at the start of the month, and we found out two weeks later — from the report.
Ad spend keeps growing, and nobody can say for sure which campaigns pay off.
The solution
What does your own analyst take over?
Your own analyst reads Stripe, Amplitude, Meta Ads, Git and your spreadsheets — read-only — and returns metrics, cohorts, reports and insights in Slack or Telegram.
Metric monitoring and alerts
Checks the key subscription metrics on a schedule and writes to the chat when a figure drops below its norm or plan — with the numbers and a likely cause.
Cohorts, LTV and plan-vs-actual
Calculates cohorts and LTV from Stripe data, compares the actuals with the plan from your spreadsheets and shows clearly where they diverge.
Dashboards in Amplitude
Creates and updates product analytics dashboards in Amplitude, so the team has standing panels instead of one-off exports.
Ad campaign performance
Matches Meta Ads spend and conversions with Stripe revenue and highlights the campaigns where the spend doesn't pay off.
Answers and insights on request
Answers one-off data questions, reads attached files, finds patterns and proposes growth hypotheses with a confidence level.
Recurring reports with charts
Builds the report on a schedule: a short summary, tables and charts as images in the chat, with the full version in Google Sheets.
what the team sees
Three artefacts that arrive on their own
A deviation alert, cohorts with LTV and an ad review — in the chat as text and charts, with the full version in Google Sheets.
Your own analyst · Slack · #metrics08:00
Alert: trial→paid conversion 8.6% against a 12% norm
MRR$47.1K+1.7%
Churn3.1%−0.2 pp
Trial→paid8.6%−3.2 pp
Trial→paid over 14 days, %
Trial→paid over 14 days, %
24 Sep
12.1
26 Sep
11.9
28 Sep
12.3
30 Sep
11.8
2 Oct
11.6
4 Oct
10.9
6 Oct
9.4
7 Oct
8.6
Likely cause: since 4 October the share of trials from the "Lookalike 2%" campaign has doubled. Confidence: high.
the norm is agreed at onboarding
Metric monitoring
You learn about a drop the same morning, not from a report two weeks later
Compares the key metrics with their norm and plan on a schedule. If a figure drops, it writes to the thread: by how much, since when and what has likely changed.
Cohorts · retention and LTVStripe · 90 days
Cohort
Customers
M0
M1
M2
M3
LTVplan $120
April
214
100%
71%
58%
49%
$118
May
236
100%
74%
60%
52%
$124
June
198
100%
66%
52%
44%
$121
July
251
100%
77%
63%
55%
$131
August
289
100%
81%
68%
59%
$142
August cohort: LTV $142 against a $120 plan. April is $2 below plan — half of it came from a promo code.
the plan comes from your spreadsheet
Cohorts, LTV and plan-vs-actual
What each cohort really brings in, and where the actuals diverge from the plan
Calculates retention and LTV per cohort from Stripe data and compares them with the plan from your spreadsheets. The table updates in Google Sheets; the summary lands in the chat.
Meta campaigns · 30 daysMeta Ads + Stripe
CampaignSpendROAS
Retargeting$1,9003.4
Brand video$2,3002.8
Lookalike 1%$3,1002.1
Interests$2,8001.9
Lookalike 2%$4,2001.1
pays off
below target
target 2.0
Two of five are below target. Suggestion: check whether "Lookalike 2%" brings trials with no intent to pay.
you set the ROAS target
Ad campaign performance
Which campaigns pay off, and where the spend goes nowhere
Matches Meta Ads spend and conversions with Stripe revenue for every campaign. Highlights where ROAS is below target and suggests what to check. It never changes budgets.
use cases
Five jobs your own analyst takes on from the first week
Each job on its own page: what hurts, what it does step by step, which sources it needs and what the team gets.
At work
What does it look like in Slack or Telegram?
Sample Slack or Telegram dialogues with your own analyst: deviation alerts, reports with charts, cohorts, ads and answers — each with numbers and a source.
Build the weekly report: MRR, churn, cohorts and ads — in the thread before the 12:00 meeting
Preparing it in the agreed format: Stripe, Amplitude and Meta data for 29 Sep – 5 Oct. Should I put the full version in Google Sheets?
Yes, and the highlights here
MRR, $K · 8 weeks
47.1
W33W34W35W36W37W38W39W40
W33: 41.2
W34: 42
W35: 42.8
W36: 43.1
W37: 44.6
W38: 45
W39: 46.3
W40: 47.1
This week's highlights: MRR +1.7% to $47.1K, churn 3.1% (norm up to 3.5%), 84 new paying customers. Meta ads: ROAS 2.4 against a 2.0 target.
📄Weekly_report_W40.xlsxFull version in Google Sheets: six months of cohorts, plan-vs-actual and campaigns row by row.
Why are there fewer new paying customers than last week?
Trial sign-ups are flat, but trial→paid conversion fell from 12.4% to 10.9%. It coincides with release 2.14 in Git on 1 October — a new onboarding step. Confidence: medium, checking the Amplitude events.
🔔 08:00Alert: trial→paid conversion for the day is 8.6% against a 12% norm. Last week: 11.8%.
Trial→paid, % · 14 days
8.6%
24 Sep26 Sep28 Sep30 Sep2 Oct4 Oct6 Oct7 Oct
24 Sep: 12.1
26 Sep: 11.9
28 Sep: 12.3
30 Sep: 11.8
2 Oct: 11.6
4 Oct: 10.9
6 Oct: 9.4
7 Oct: 8.6
Likely cause: since 4 October the share of trials from the "Lookalike 2%" campaign has doubled, and its conversion to paid is 4%. Other sources are within the norm. Confidence: high.
Keep watching it and tell me if it isn't back to normal tomorrow
Done. I'll check at 08:00 and post in this thread ✓
Which cohort has the best LTV, and are we on plan?
LTV by cohort, $ · 90 days
$142
AprMayJunJulAugSep
Apr: 118
May: 124
Jun: 121
Jul: 131
Aug: 142
Sep: 139
The August cohort: LTV $142 at day 90 against a $120 plan. The weakest is April at $118: half of it came from a −50% promo code.
Put the monthly plan-vs-actual into the spreadsheet
📄Plan_vs_actual_LTV_2026.xlsxUpdated the "Plan vs actual" sheet: plan, actual and the difference for every cohort. April is below plan, the rest above.
Which Meta campaigns aren't paying off right now?
ROAS by campaign · 30 days
1.1
Retarg.BrandLAL 1%InterestsLAL 2%
Retarg.: 3.4
Brand: 2.8
LAL 1%: 2.1
Interests: 1.9
LAL 2%: 1.1
Two of five are below the 2.0 target: "Interests" at 1.9 and "Lookalike 2%" at 1.1. The second took $4,200 this month, and Stripe revenue from its customers is $4,600.
What do you suggest?
A suggestion, not a decision: test the hypothesis that "Lookalike 2%" brings trials with no intent to pay. The check: compare trial→paid conversion with "Lookalike 1%" over two weeks. I don't touch budgets.
🎤▁▃▅▇▆▄▂▅▇▃▁0:12
Transcribed: "how much did we make in Poland in September, and how does it compare with August".
Poland in September: $6,840 from Stripe, $5,910 in August, so +15.7%. 112 paying customers, 19 of them new.
Send me the breakdown by plan as a file
📄Poland_September_by_plan.xlsxBuilt it: Basic 71, Pro 38, Team 3 — with the revenue for each.
📄Customer_survey_Q3.pdf
Read it: 24 pages, 312 responses. Key points: 41% mention price as a reason for doubt, 27% the missing CRM integration.
Cross-check with churn: do the ones complaining about price actually leave?
Checked against Stripe: among the 128 respondents who named price, 60-day churn is 9%, versus 4% for the rest. The difference is significant. Logged it as an insight in the "Insights" thread ✓
Brainstorm: how do we lift trial→paid conversion to 14%?
Three options from the data. 1) A reminder on day 5 of the trial: people who open the product three times in the first week pay 2.6× more often. 2) Shorter onboarding: since release 2.14 only 58% pass the "import" step. 3) A discounted annual plan: 22% chose it in September.
Which one should we test first?
The first: the cheapest test, with the effect visible in two weeks. Expected +1.5 pp of conversion, confidence medium. I'll prepare an A/B test plan for the team.
Build an activation dashboard for the product team
Setting it up in Amplitude: the sign-up → import → first report funnel, DAU/WAU, retention by cohort. Do you need a breakdown by plan and platform?
Yes, and give the whole group access
The "Activation" dashboard is created, shared with the Product group ✓ It updates itself; I didn't change the event tracking setup.
The result
Before → after
BeforeAfter
Before: Half of every Friday spent building the report by hand
After: A scheduled report with charts in Slack or Telegram, the full version in Google Sheets
Before: We learn about a churn spike two weeks later
After: An alert in the chat the same day: what dropped, by how much and the likely cause
Before: Nobody can say for sure whether the ads pay off
After: A Meta campaign review with Stripe revenue and the areas to improve
Quick contact
Write to us
Two fields, and a priced offer lands in your inbox within one business day.
Live in 2 days. A priced offer by email within one business day.
What's included
Stripe, Meta Ads Manager, Amplitude, Git and Google Sheets — read-only access
Scheduled monitoring of the key metrics and alerts on deviations
Cohorts, LTV and plan-vs-actual in Google Sheets, dashboards in Amplitude
Recurring reports with charts and an ad campaign review in Slack or Telegram
Answers on request, file breakdowns, insights and brainstorms in threads
OAuth access, no passwordsConfirmation before sending or publishingEU company, GDPR
Questions
Frequently asked questions about your own analyst
Your own analyst from sviy.ai is an AI solution for subscription businesses that watches the key metrics, calculates cohorts, LTV and plan-vs-actual, evaluates ad campaigns and answers data questions. It's often searched for as an AI assistant for analysts. It works in a dedicated group in Slack or Telegram, sends reports with charts on a schedule and alerts you to deviations from the norm.
Your own analyst reads Stripe, Meta Ads Manager, Amplitude, a Git repository and Google Sheets. Every access is read-only: it never changes data in any source and never manages the ads. Git is there to explain metric movements with releases and product changes. Other sources are connected by separate agreement after the demo, and it creates Amplitude dashboards only when given that permission.
No. Your own analyst alerts, calculates and proposes, and a person on your team decides. It never changes product or ad settings and never passes an assumption off as a fact: every insight carries a confidence level and a source in the data. If the data for an answer isn't there, it says so instead of inventing numbers. It runs on Claude models (Anthropic) via the commercial API.
Wherever your team already works. We create a dedicated group in Slack or Telegram, and your own analyst splits the work into threads: alerts, reports, requests, brainstorms. The first wave connects up to 10 users with equal rights; per-user permissions are built into the core and switched on later without rework. Full report versions live in Google Sheets.
Your own analyst goes live in 2 days after the demo. We send a priced offer by email within one business day: the price depends on the number of data sources and users. On day one we create the group in Slack or Telegram, connect the sources and load the project context. On day two we agree the metric norms and the report format, and it sends the first one.
Yes, each product or company gets its own group, data sources, metric norms and report format. Your own analyst never mixes data between them and asks when a request doesn't make the product clear. Whether you get a summary per product or one combined report, we discuss at the demo together with the frequency.
Ready to hand your metrics over to your own analyst?
Tick the functions you need in the builder — we email a priced offer within a business day, live in 2 days.