When Google Ads stops producing leads, the two usual reactions are also the most expensive: raise the budget, or switch the campaign off. Both skip the diagnosis. In most underperforming accounts the money is not lost in the auction — it is lost in the search terms, the tracking and the follow-up.

What is the short answer when Google Ads is not working?

"Google Ads does not work" is almost never a statement about the channel. It describes one account, with one set of settings, pointed at one page, feeding one sales process — and any of those layers can break the result alone.

The belief to dismantle first is that spend and results move together. An account with the wrong conversion action marked as primary gets worse the more you spend, because the bidding optimises harder towards the wrong signal. Pausing is not free either: you lose the baseline you would need to prove what went wrong.

So the order is fixed: measure correctly, clean the traffic, fix the destination, then decide about money.

Where should the audit start?

Start along the conversion path, not in the campaign settings. It runs from a query typed in Tashkent, to an ad, to a page, to a form or a call, to a person who answers, to a deal recorded somewhere.

We walk it backwards first, from the client's own records, because that version nobody can dispute. When the number of real enquiries and the number Google Ads reports disagree widely, that gap is the first finding.

SymptomMost likely causeWhat the audit does
Clicks but almost no enquiriesQuery-to-offer mismatch, or a page ignoring the searchRead search terms against the page; test the form
Leads arrive but are poor qualityBroad match, thin negative listSegment by intent; build negatives; split high-intent
Budget gone by the middle of the dayOne generic keyword eating the budgetCap top-spending terms; unbundle intent levels
Optimisation pulls the wrong wayA fake primary conversionRe-declare primary actions; drop page views and clicks
Cost per click rose on its ownNew competitors or wider targetingCheck auction insights, ad relevance and location settings

Four of those five are solved without touching the budget.

Does the campaign objective match the business outcome?

The traffic objective trap

The objective is not a label; it tells the system what to buy. Campaigns built to maximise clicks look excellent while the phone stays silent. The test is blunt: would you pay a salesperson a bonus for what the account is optimising towards? Nobody pays a bonus for a session.

Calls, forms and Telegram clicks

For most businesses here the real outcome is a call, a completed form, or a click through to a messenger. Telegram is the most widely used messaging platform in Uzbekistan, and if that click is untracked, part of your result is invisible — so the algorithm learns to avoid the audiences producing it.

Purchases and longer cycles

Online purchases must carry a value, not just a count. Where the sale closes offline weeks later, the account needs an intermediate outcome plus a way to reconcile it with revenue; optimising to raw lead count in a long-cycle business produces cheap, useless leads.

What should you look for in the search terms report?

Keywords are what you asked for; search terms are what you paid for. Export at least ninety days and sort by cost rather than clicks. The share of irrelevant terms comes out visibly high in most accounts we open, and reading them in bulk makes it impossible to ignore.

Terms close to a sale

A service plus a city, a product plus the word price, anything phrased as an urgent need. These deserve their own ad group, copy and budget line, because they convert at a different rate entirely.

Research-stage terms

Queries starting with what, how or why belong to people learning, not buying. Move them to content and revisit paid coverage once the commercial side pays.

Irrelevant, brand and competitor terms

Terms sharing a word with your keyword and nothing else with your business go onto the negative list, as do vacancy and salary queries. Brand terms deserve a small separate campaign; competitor terms are a deliberate strategy, not an accident.

Search term groupIntentAction
Service plus city, price or buyReady to actKeep; move into a high-intent group with its own budget
What, how, why questionsLearningRemove from search; cover with content
Unrelated meanings of the wordNo relevanceAdd as exact or phrase negatives
Your brand and misspellingsAlready looking for youSeparate brand campaign, low bids
Competitor brand namesComparing optionsOnly as a deliberate capped campaign
Vacancy, salary, jobLooking for workPermanent account negatives

Why update the negative keyword list every week?

Because the queries change every week. New phrasings appear, a news story makes an unrelated word popular, and the account pays for it within days. Weekly suits most accounts; above roughly $2,000 a month, twice a week is better.

There is a local wrinkle. Search here is genuinely multilingual: one user will mix Uzbek and Russian in a single query, sometimes in two scripts. The same irrelevant concept therefore reaches your ads through several spellings, so build the list in both languages and add transliterated variants.

Keep two levels — account-level negatives for what will never be relevant, campaign-level ones for terms relevant elsewhere — and check whether a term ever converted before blocking it.

How does match type burn through a budget?

Match type decides how far Google may travel from your keyword. Broad travels furthest, and in a small account with a thin negative list it can consume a month of budget on queries you would never have chosen.

This is not an argument against broad match: with clean data and real volume it finds demand that exact and phrase never reach. The failure mode is using it early, in a small account, with weak signals — exactly where most businesses are when they decide the channel does not work.

A safer default when rebuilding: exact match for terms you know convert, phrase for the family around them, broad only in a separate capped campaign once tracking is verified.

Is conversion tracking set up correctly?

Primary and secondary conversions

Only primary actions are used for bidding. Real enquiries — form submissions, calls of meaningful length, messenger clicks — belong in primary; everything else goes in secondary, where it is observed without steering the spend.

Fake conversions

A page view is not a conversion. Nor is a button click, unless that button is the only route to a real enquiry. When these sit in primary, cost per conversion looks wonderful and the algorithm spends the month buying people who scroll.

The thank-you page and duplicate counting

A thank-you page reachable only after a genuine submission is still the cleanest signal. Check it cannot be opened from navigation and that refreshing does not count again, and count one conversion per click.

Server-side tracking and offline results

Browser tracking loses events to blockers and weak mobile connections; offline import closes the gap by pushing back which lead became a customer. Test the chain by hand — we have audited accounts where a form had been silently broken for weeks and every report measured nothing.

Are location settings pushing spend outside Tashkent?

By default, ads reach people merely interested in your location, not only those in it. For a clinic in Tashkent that decides whether you pay for patients or for browsers on another continent. Set targeting to presence, then add exclusions.

Tashkent behaves like a different market from the regional cities: denser competition, higher cost per click, a different language mix. If they share a campaign, the capital absorbs the budget by auction pressure and you conclude, wrongly, that the regions do not respond.

The same applies one level up. In a multi-country account, Uzbekistan should never share a campaign with a neighbouring or head-office market: country-level separation gives separate budgets, bidding and reporting.

Check the billing side too. Accounts run from abroad on USD billing face caps that look generous in dollars but behave differently against local so'm pricing when the rate moves, plus payment failures that pause campaigns at the worst moment.

Does the landing page keep the ad's promise?

An ad makes a specific promise and the page has about three seconds to confirm it. When the ad says same-day delivery in Tashkent and the page opens on a corporate homepage, the visitor has gone — and you paid for that visit.

Three checks matter: continuity (the headline repeats the ad and the query), proof (real photographs, names or a price range on the first screen) and effort (a short form, a tappable number, a messenger link). Test all three on a real phone.

Page-side diagnosis is a subject of its own, covered in the guide to sites that get traffic but no enquiries. For this checklist one question is enough: does the page answer the exact query you paid for?

Is the ad copy generic, or does it answer the intent?

Generic copy is expensive twice: it attracts the wrong clicks and lowers the relevance signals that set your price for the right ones. Quality services at affordable prices tells a searcher nothing.

Good copy repeats the searcher's own words, states one concrete differentiator — a timeframe, a guarantee, a price range — and sets an expectation the page can confirm. Write it after reading fifty real search terms, and point sitelinks at genuinely different pages.

How should the budget be split across campaigns?

Ring-fence the high-intent terms

One campaign holding every keyword is the structural version of this problem: the cheapest terms take the clicks and the terms closest to a sale get the remainder. Commercial terms need their own campaign and a budget nothing else can raid.

Keep the brand campaign separate

Brand traffic is cheap and converts well, so mixing it in flatters every average and hides how everything else performs. Separate it and read it as its own line.

Reserve a test budget

Ten to twenty per cent of monthly spend should be available for deliberate tests. Without a reserved budget, testing happens by accident, with money meant to produce this month's leads.

Line itemTypical monthly figureNote
Google Ads account managementfrom $400Audit, restructuring, weekly search-term work
Performance marketingfrom $500Paid search plus other channels, one model
Landing page for the campaignfrom $300Often the highest-return fix
CRM integration and lead routing$500–$1,500One-off; makes offline results visible to bidding
Media budgetBy market and goalFollows the audit findings, not the reverse

Management, a proper page and CRM integration together often cost less per month than the media wasted on unfiltered search terms in the same account. (Source: 101 Digital 2026 service price list)

Who follows up on a lead, and how quickly?

An audit that stops at the edge of the ad account misses where most paid leads are lost: a lead nobody answers costs exactly as much as one that converts.

Speed is decisive: contacting a lead within five minutes rather than thirty made it around twenty-one times more likely to be qualified — a different business outcome from identical spend. (Source: Lead Response Management Study — Oldroyd, MIT/InsideSales)

Time zones make this structural for internationally managed accounts. If ads run in Tashkent while the team answering sits several hours behind, a lead arriving at ten in the morning may not be touched until the afternoon, by which time the searcher has spoken to two competitors. The fix is a local first responder, an automated acknowledgement, or ad scheduling matched to the hours someone can answer.

Expectations must match the cycle. In B2B the median runs to around 84 days, with cost per lead typically between $560 and $2,800 depending on market and deal size, so an account judged after three weeks will always look like a failure. (Source: 2026 B2B sales cycle and lead cost benchmarks)

This is where CRM becomes a measurement requirement rather than an administrative preference. The mechanics for a small team are in the guide to CRM automation for small businesses in Tashkent; the minimum is a same-day response, every enquiry recorded with its source, and one named owner.

When does raising the budget make sense?

Three conditions, all required. Tracking is verified end to end and counts only real outcomes; cost per qualified lead — not per raw lead — is at or below what the business can pay and still make a margin; and the campaign is limited by budget rather than by demand, which impression share lost to budget shows directly.

A real account shows how this plays out. When we took over the campaigns for İnci Diş, a dental clinic, everyone assumed more budget was needed; we added none in the first phase. We rebuilt the measurement, stripped out irrelevant search terms and separated high-intent keywords from general ones. In our own reporting we then saw appointments rise by 45%, return on ad spend reach 4.2x and cost per lead fall by 30% — the saving coming from cleaning what was already being paid for. (Source: 101 Digital case data)

PhaseWork doneMetric watched
Days 0–30Audit; tracking rebuilt and tested by hand; search terms, negatives and targeting correctedShare of spend on relevant terms; conversions firing
Days 30–60Restructuring by intent; brand separated; copy rewritten from real queries; page aligned with the adCost per qualified lead; page conversion rate
Days 60–90Bidding moved onto clean data; budget reallocated to proven groups; CRM results importedReturn on ad spend; revenue per lead; impression share lost to budget

Ninety days is the shortest period in which measurement can be fixed, structure rebuilt and enough data accumulated to decide. Shorter reviews measure noise.

Which mistakes cost the most money?

MistakeWhy it is expensiveFixTime
Page views or clicks as primary conversionsBidding buys non-buyers harder as spend risesRe-declare primary actions; verify each by hand1–2 days
Location left on presence-or-interestBudget leaks to people who will never buySwitch to presence; add exclusionsUnder an hour
Broad match with a thin negative listYou pay for queries nobody would have chosenTighten match types; build negatives in both languages2–3 days, then ongoing
All keywords in one campaignGeneric terms absorb the budget firstSplit by intent; ring-fence commercial and brand3–5 days
Ads pointing at the homepageThe promise is not confirmed, so paid visits bounceBuild a page per campaign theme1–2 weeks
Smart bidding with no conversion historyThe algorithm learns from noiseKeep manual control until volume is real30 days of data
Nobody owns the follow-upPaid leads go cold; advertising takes the blameAssign an owner; same-day response rule1 week

What does the full audit checklist look like?

Run these in order: measurement, traffic, destination, money.

  • Compare your own record of enquiries with what the account reports.
  • List every conversion action, confirm which are primary, and remove page views, scroll events and button clicks from that set.
  • Submit a real form, place a call and click the messenger link from a phone; confirm each is recorded.
  • Count one conversion per click; confirm the thank-you page needs a genuine submission.
  • Export ninety days of search terms, sort by cost, classify by intent.
  • Build account- and campaign-level negatives in Russian and Uzbek, with transliterations.
  • Restrict broad match until conversion data is trustworthy.
  • Target by presence; separate Tashkent, the regions and each country.
  • Open every destination on a real phone: load speed, headline continuity, form length.
  • Rewrite ad copy from real search terms; audit extensions and sitelinks.
  • Split the budget: high-intent, brand, reserved test.
  • Confirm billing has a backup payment method.
  • Define who responds to a lead, within what time, and where it is recorded.
  • Only then decide whether the budget changes, against the three conditions above.

Jasur Rahimov, performance marketing lead at 101 Digital: İnci Diş is still the account I use as the reference case. For the first month we added no budget at all. We rebuilt the tracking so only real appointment requests counted, then spent days reading search terms and pulling out everything that was never going to book a chair.

The numbers moved in the order you would expect: cost per lead dropped 30% first, appointments rose 45% as the budget concentrated on people ready to book, and return on ad spend settled at 4.2x after that. My rule since: never approve a budget increase for an account whose tracking I have not tested myself by submitting a form.

What do accounts look like after this work?

The projects below are accounts where this sequence was applied — measurement, traffic quality, structure, then budget.

Real Results

Our Client Results in This Area

Ankalife Healthcare — IVF / Fertility Clinic · Uzbekistan

Digital marketing for a Tashkent-based IVF and women's health center

$3,000 → $1,500 (-50%)Monthly Ad Budget
Increased on half the budgetLeads
4 languages (ru, uz, en, tr)Site Languages
Details →
SCK Representation Logistics & Maritime · Turkey

Google Ads management for Turkey's sole authorized representative for West Africa cargo tracking certificates

€15,000–20,000Monthly Ad Budget
Turkey, Europe, AmericasTarget Markets
B2B LogisticsSector
Details →
İnci Dental Clinic Healthcare — Dental Clinic Chain · Turkey

Google Ads, Meta Ads and CRM integration for Istanbul's leading dental clinic chain

+45%Online Appointment Growth
4.2xGoogle Ads ROAS
-30%Patient Acquisition Cost
Details →

What should you read next?

Frequently asked questions

Usually one of four things: tracking measures something that is not a real enquiry, the search terms report is full of queries with no commercial intent, the ad points at a page that ignores the query, or nobody responds quickly enough. The settings and the follow-up are the problem far more often than the channel.

Five to ten working days for one account. The output is a list of findings ordered by how much each is costing, a corrected tracking setup, a cleaned search terms and negative keyword list, and a restructuring plan with effort estimates.

Smart bidding needs reliable conversion data. If the primary action is wrong, or the account produces a handful of verified conversions a month, automation optimises towards noise. Fix measurement first, gather thirty days of clean data, then set a target reflecting what a lead is worth.

You can, and with strong service pages it sometimes works. But if the destination is a general homepage, you are asking the visitor to find the answer after you paid for the click. A campaign-specific page repeating the ad promise is usually the highest-return fix.

Only when three things are true together: tracking is verified and counts real outcomes, cost per qualified lead is inside what the business can afford, and the campaign is losing impression share to budget rather than to low demand. If one is missing, more budget buys more of the same result.

Businesses whose customers already search for the solution by name and act quickly: repairs, medical and dental services, legal and accounting help, logistics, equipment supply. Where demand must be created rather than captured, social advertising and content build the audience first.

How does 101 Digital run an ad account audit?

We start with your own record of enquiries rather than the dashboard, walk the conversion path in the order above, test every conversion action by hand, and rank the findings by cost. What you get is a decision document: what is broken, what it costs, what the fix takes, and in which order.

If your account has stopped producing results and you would rather know why before spending more, see Google Ads management.