Where to start if ads are already running?

Do not stop Ads immediately if leads still arrive somehow. In 1–2 weeks connect CRM and deal creation from bot/form, add UTM, then fix the landing to the offer. Otherwise you keep paying for clicks with no tracking. After the seam — review breaks in the table, then think about raising budget.

Do we need Instagram DM, WhatsApp API and site chat at once?

No. Stitch one path to a deal (for example landing → bot → CRM), then add channels. Many entries without one CRM multiply breaks and duplicates. When one path is stable — add the next with the same statuses and UTM.

What budget for a 3–4 week pilot?

Soft media guide for Tashkent — from ~$150–300 on one narrow offer plus one-off stitching (landing + bot + CRM). In very competitive niches media may be higher. This tests the chain, not a year-long strategy and not a payback guarantee.

When should we raise Ads budget?

When a test lead reliably lands in CRM with source, the manager replies within the agreed SLA (often the same day), the conversion is a lead or booking — not a button click — and there are no mass private chats around the system. If an item is red — fix the junction.

How does the bot differ from CRM in this scheme?

The bot replies fast and collects qualification fields. CRM holds statuses, tasks, history and “where money comes from”. Without CRM the bot turns sales into private chats again: dialogue exists, controllable tracking does not.

Is a website required if clients write straight to Telegram?

Not always on day one. For a simple offer you can test ads into the bot — but UTM/gclid tags and a CRM card are mandatory. For a complex service a landing is usually needed: prices, trust, match with the ad. See [bot and Google Ads](/en/blog/telegram-bot-i-google-ads-svyazka).

How many CRM statuses at the start?

Usually 4–6: new, qualified, in progress, won, lost (with reason). More than ten statuses on a pilot often confuses the team. Better a short clear funnel people actually use every day.

Marketing20 minJuly 20, 2026

Funnel website → Ads → bot → CRM in Tashkent: build order and break points

Robert MirovUpdated Sep 5, 2026

Owners in Tashkent often buy pieces separately: a “ads landing”, a Google Ads account (paid search and display), a bot “for night replies”, and a CRM (lead and deal tracking system) “like everyone else”. Then clicks exist but sales are thin — not because “ads are bad by themselves”, but because the links are not stitched. Typical picture: someone saw an ad → opened the homepage → wrote a manager’s private chat → a week later nobody remembers the source → Ads is optimized on CTR (click-through rate), not on money. In Uzbekistan clients also prefer writing in Telegram or WhatsApp faster than filling a long form. Below — a practical build order website → Ads → bot → CRM, break-point and Before|After tables, roles of each link, a 3–4 week pilot, a checklist before raising budget, and a soft pilot range. No “ROAS ×10” promises: only what to connect, in what sequence, and how to know the seam holds. Full service loop — sales automation.

01What chain a business in Uzbekistan needs

Clients rarely fill long forms on desktop. More often on a phone: saw an ad → opened the site or went straight to Telegram → wrote → waited → bought or left. If a reply came an hour later in “Aziz from ads” private chat, a deal may still happen — but you will not know which campaign brought it, and you cannot repeat the result. Minimum working chain — four links, each with a clear role:

Link	Role	Without it
Website / landing	Offer, trust, one primary CTA (call to action)	Traffic with no entry — or a “homepage about everything”
Ads	Paid demand + UTM (source tags in the URL)	No controllable flow, only word of mouth
Bot	Fast reply, field capture, routing	Leads in managers’ private chats
CRM	Statuses, tasks, “where money comes from”	Guesswork, no source or history

Before	After
Landing, Ads, bot and CRM bought separately	One chain with a shared status and source vocabulary
“Message us” into a private chat	Bot or form creates a CRM card
Source “from Telegram / from the site”	UTM + campaign on the card from first touch
Argue: ads vs managers	Review leads → qualified → deals by source

A bot without CRM turns sales into private chats again. Ads without an offer landing is expensive traffic nowhere. CRM without source is blind reporting. A site without one CTA leaves people unsure what to tap. The stitch does not replace a real offer and price — it removes blind spots between click and cash.

02Build order: what comes before ad spend

Mistake #1 — run Ads while there is nowhere to store leads. Order beats a pretty creative. Creative without a seam scales a hole: more clicks — more messenger chaos. Recommended sequence: 1. Offer and stages in CRM: 4–6 statuses (new → qualified → in progress → won / lost), 5–7 required fields (service, district/branch, contact, source, next step). 2. Entry point: landing or service page + bot button or form — page copy matches the ad. 3. Bot or form → deal in CRM — manual test, no ads: you walk the path and see the card. 4. Tracking: UTM, Ads goals, test conversion = business event (lead / booking), not a button click. 5. Only then — launch Google Ads on one narrow offer (one service, one district or branch).

Before	After
Ads budget first, CRM “later somehow”	CRM and entry first, then small budget
Ad sends to homepage	Ad → service page / offer landing
Conversion = “Message” click	Conversion = lead or booking in the system
Three offers in one campaign on day one	One offer for a 3–4 week pilot

The landing must match the ad — otherwise bounce rises and trust drops: see landing for ads. Bot + Ads (UTM, gclid, dialogue) — Telegram bot and Google Ads.

03Break-point table: where the chain snaps

Most holes are not in Google’s algorithm — they are at the junctions. The owner sees “ads don’t work”, while the lead never reached a manager or arrived without a source.

Break	Symptom	Fix
Ad → homepage	Offer mismatch, user hunts for a button	Send to service page / landing
Telegram without UTM / start	Empty source in CRM or “from Telegram”	Params in the button / deep-link
Bot without a deal	Reply “in someone’s head”, no statuses	Create the card automatically
Managers’ personal numbers	Parallel chats, lost threads	One official channel + CRM
Conversion = button click	Ads “wins”, sales don’t	Count lead / booking
Watch only Ads spend	No funnel picture	Weekly: leads → qualified → deals
Night form with no reply	Leads go cold until morning	Bot + notification + reply SLA
Two CRMs / Excel sheets	Duplicates and “whose lead?” fights	One tracking loop for the pilot

If messengers are the main contact, see CRM and WhatsApp Business. For pre-manager qualification (short bot questions) — AI in CRM and bot. Counting Telegram leads without a classic form — attribution.

04Link roles: who owns what

Fix ownership so you do not spawn duplicates or blame “ads” or “managers” without data. • Website — offer, trust proof (address, phone, cases without invented %), one primary CTA • Ads — traffic and account learning on the right conversion (lead / booking), not a click • Bot — fast reply, 3–5 qualification fields, routing to the right branch / manager • CRM — statuses, tasks, touch history, “where money comes from” What should not be duplicated: • two “primary” CTAs on the landing (form + three messengers + call with no priority); • managers’ personal WhatsApp next to the official bot; • manual card creation “when someone remembers” while the bot already works; • different names for the same statuses across people. In Tashkent for a salon, clinic, delivery, B2B service or local retail the scheme is the same: one entry → one card → one report. Only qualification fields and acceptable reply speed change by niche.

053–4 week pilot: one offer, one branch

Pilot goal — not “ROI in 48 hours”, but proof that every paid click leaves a trail to a manager. Then bids, creatives and wider geo. Pilot plan (one offer, one district or branch):

Week	Focus	“Done” criterion
1	CRM: stages and fields	Statuses clear to managers, no useless columns
2	Landing/page + bot → CRM	Test lead creates a deal with source
3	Tracking + Ads with small budget	Conversion = business event; phone and desktop test
4	Fix breaks and reply scripts	Source on every card; weekly report exists

Before	After the pilot
“They click, sales are thin”	Visible: click → lead → qualified → deal / lost
Budget cut or raised by feel	Decision by qualified CPL and reply time
Manager does not know lead source	Campaign / channel on the card
Three offers and two branches at once	One offer, one entry — clean stitch test

Do not compare Google and Instagram in week one of the pilot. Stitch one channel first. Channel comparison makes sense when UTM and CRM vocabulary are already the same.

06Checklist before scaling budget

Raise budget only when the chain already clicks. Otherwise ×2 budget = ×2 chaos and ×2 cost of the same hole. Before ×2 budget, check: • a test lead from the ad creates a CRM card with source (UTM / campaign); • the manager sees a notification and updates status the same day (or within the agreed SLA); • Ads conversion is a lead / booking, not only a button click; • no parallel private chats “around” the official channel; • weekly owner review: leads → qualified → deals / losses with reason; • messenger reply speed is known and acceptable for the niche (salon booking — minutes; B2B — hours in a workday); • share of empty bot starts or abandoned forms is understood — you are not optimizing CTR alone. If any item is red — fix the junction, not creatives. Creative will not fix a truncated UTM or a manager chat outside CRM.

07Pilot budget ranges and what to watch

Figures are a soft guide for Tashkent — not a UZNEO price list and not a payback guarantee. They depend on niche, keyword competition and site readiness.

Item	Guide	Note
Stitching (landing + bot + CRM)	one-off project	Depends on site readiness and integrations
Ads media for pilot	from ~$150–300 / 3–4 weeks	One narrow offer; competitive niches may need more
Management / tweaks	by hours	After week 4 — targeted: breaks, copy, bids

Pilot KPIs (review every week): • share of leads with a filled source; • time to first manager reply; • lead → qualified conversion; • CPL (cost per lead) on qualified leads, not every click and not empty `/start`; • with enough volume — deal rate and loss reasons. Do not compare channels until you have one CRM and consistent UTM. Stitch first, scale second. When the seam holds — strengthen ads and improve the website, instead of buying another tool “just in case”.

08Typical mistakes and when the funnel is too early

Most failures are not “funnels don’t work in Tashkent” — they are wrong build order or too wide a start. Typical mistakes: 1. Ads before CRM and bot — you pay for clicks with no tracking. 2. Five entry channels in week one (site, IG DM, WhatsApp, three private chats) without one system of record. 3. Ads conversion = click or bot open without qualification. 4. CRM statuses “as it goes” — offline Ads learning and honest CPL become impossible. 5. Scaling budget on a crooked landing: ad offer does not match the page. 6. Expecting ROAS ×10 in two pilot weeks — the pilot is about the seam, not a year-long strategy. When it is too early to complicate: if leads are a few per week and one person reliably replies in 15 minutes from one chat — first fix offer and reply speed. Full website + Ads + bot + CRM makes sense when paid traffic is already running (or about to), and messenger chaos blocks you from seeing what pays back. Need a turnkey build — sales automation or by block: web, ads, bots, CRM.

Summary

A Tashkent funnel works when website, Ads, bot and CRM are stitched in one order — tracking and entry first, budget second. Look for breaks at junctions: UTM, deal creation, private chats around the system, false conversions, homepage instead of an offer landing. A 3–4 week pilot on one offer proves that a paid click leaves a trail to a manager. Only after a green checklist does ×2 budget make sense. UZNEO builds the loop calmly and practically — without magic ROAS promises. Start: sales automation or blocks web, ads, bots, CRM.