Facebook Ads in Nepal 2026: Meta Ads Manager Guide

Most Nepali businesses that say they "run Facebook ads" are boosting posts from the Page with a dollar card and hoping. That is fine until someone asks what a customer costs. Answering that question means using Meta Ads Manager properly, and Ads Manager in 2026 works differently from the version most local guides describe — the targeting controls changed, the event setup changed, and Nepal has its own payment and tax rules layered on top.

This is a working guide for a business in Nepal: what to set up, how to structure campaigns, what targeting still does anything, how the Pixel and Conversions API fit together, what to budget in rupees, and how to actually pay Meta from a Nepali bank. It covers the paid side only; the organic Page and Instagram work that sits alongside it is social media management, a different job with a different rhythm.

The numbers worth knowing before you spend anything

Meta's own advertising tools reported a Facebook ad reach of 14.8 million people in Nepal in late 2025 — about 50 percent of the total population and 75.6 percent of adults aged 18 and over. Instagram ad reach was 4.35 million, or 21.7 percent of adults. Kepios put Nepal's internet users at 16.6 million, a 56 percent penetration rate, as of October 2025. Those figures are compiled in DataReportal's Digital 2026: Nepal report.

Two things follow. First, Facebook's addressable audience in Nepal is more than three times Instagram's, so splitting a small budget evenly across the two platforms is a mistake — put both placements in one ad set and let delivery decide. Second, do not size an audience from connection counts. The same report puts Nepal at 32.4 million cellular mobile connections in late 2025, equivalent to 109 percent of the population, because plenty of people carry more than one. Meta's reach estimate is the number that matches what you can actually buy, and it is the one figure in the wider digital marketing picture in Nepal that maps directly onto a media plan.

Setting up Ads Manager in the order that avoids rework

Two ad account settings are effectively permanent, and getting them wrong costs you your history. Do this in sequence.

1. Create a business portfolio (what used to be called Business Manager) rather than advertising from a personal profile. Assets stay with the business when staff leave.
2. Turn on two-factor authentication. Meta may restrict or remove advertising access, shop editing rights, or a payment method if it is off, and Meta also says advertisers may be required to complete two-factor authentication in order to publish ads, a group that includes people working in business portfolios more than 90 days old. Each person sets it up on their personal Facebook account.
3. Claim the Facebook Page and Instagram account into the portfolio before you build anything.
4. Create the ad account and set currency and time zone deliberately. Changing either one later closes the existing ad account and creates a new one. Running ads stop, and the spend history does not move. Set the time zone to Kathmandu at creation rather than accepting whatever the form offers, or every report you read will be shifted against the day your client actually trades.
5. Set up a dataset in Events Manager and install the Pixel before you launch, not after.

One constraint that surprises agencies: a new business portfolio can create only one ad account, and the limit rises to three once a campaign delivers and a payment is made. Plan client accounts around that.

Your ad account cannot be denominated in rupees

The Nepali rupee is not on Meta's list of accepted currencies for ads. In practice Nepali advertisers run their accounts in US dollars. That means every budget you type into Ads Manager is a dollar figure, and your rupee cost depends on the rate your bank applies to the card charge. Budget in NPR internally, convert once, and set the USD figure.

Paying Meta from Nepal

Meta's accepted payment options for Nepal are short: credit cards or co-branded debit cards (Visa, Mastercard, American Express) and PayPal. That is the whole list. Nepal has no local manual payment method with Meta — there is no eSewa, Khalti, Fonepay or connectIPS route, and no prepaid "available funds" balance to top up. Compare that with Vietnam, where Momo and VNPay are listed as local manual payment methods, and you can see the gap is country-specific, not a bug in your account.

What that leaves is a foreign-currency card issued by a Nepali commercial bank. Most banks sell one, physical or virtual, under their own brand name. Annual usage ceilings are set under Nepal Rastra Bank's foreign exchange rules and are not one flat number: they differ for individuals and for registered firms, and there are separate provisions for IT and export-earning businesses. Those ceilings have been revised more than once, and most of the figures circulating in Nepali blog posts are out of date. Confirm your current limit with your bank's card department and against the notices published by NRB's Foreign Exchange Management Department before you plan a year of spend around it.

Practical warnings from running these accounts:

• A declined charge can get the ad account disabled, not merely paused. Keep the card loaded ahead of your billing threshold rather than topping up when Meta asks.
• Thresholds rise as you spend. Meta sets a new account's payment threshold to a small amount — its own worked example starts at USD 25 — and may raise it each time a payment clears, until the account reaches a final threshold. Each charge is therefore liable to be larger than the last, a growing single hit on a prepaid card.
• Keep the card in the business's name where the bank allows it. Personal cards on client accounts create a reconciliation mess.

VAT and your PAN

Since August 2023, Meta ads sold to customers in Nepal who have not declared a business purpose are charged Nepal VAT at the applicable local rate. That rate is 13 percent: Nepal has levied 13 percent VAT on digital services supplied by non-residents to consumers in Nepal with effect from 29 May 2022. If you are buying for business purposes, add the declaration and your Permanent Account Number in payment settings; you are then responsible for VAT under the reverse charge mechanism instead of paying it to Meta. If you do not declare, Meta treats the purchase as non-business and charges VAT. Note that VAT is added on top of your charges, so a single billing can exceed your threshold amount.

What to budget, in rupees

Budget arguments in Nepal usually start from what the owner feels comfortable spending. Start from the delivery system's arithmetic instead.

An ad set is flagged Learning limited when it is unlikely to receive about 50 optimization events in the week after your last significant edit. Meta is explicit that this is not a penalty — it is a warning that the budget is not being spent effectively. Fifty events a week is roughly seven a day. So the floor for any ad set is:

Daily budget ≈ 7 × what one optimization event costs you.

Meta gives a second, similar rule for the cost per result goal bid strategy: the daily budget should be at least five times the cost per result goal. Both push in the same direction.

The table below is arithmetic, not a benchmark. The cost figures are placeholders — substitute your own once you have two weeks of data. Rupee conversions use Nepal Rastra Bank's rate for 29 August 2026, USD 1 = NPR 152.91 selling, published on the NRB foreign exchange rate page. Your card rate will be slightly worse.

Optimization event	Example cost per result	Daily budget floor	Roughly per month
Landing page view	NPR 15	NPR 105 (about USD 0.7)	NPR 3,150
Messaging conversation started	NPR 60	NPR 420 (about USD 2.7)	NPR 12,600
Lead via instant form	NPR 150	NPR 1,050 (about USD 6.9)	NPR 31,500
Purchase on your website	NPR 400	NPR 2,800 (about USD 18)	NPR 84,000

The useful lesson is in the last row. If purchases cost you NPR 400 each, the cheapest ad set that can learn properly runs about NPR 84,000 a month — and that is one ad set. Most small Nepali retailers cannot fund that, which is why they should optimize for a cheaper, more frequent event such as a messaging conversation or an add to cart, and treat purchases as a reporting metric rather than the optimization target. Meta's own advice on learning limited ad sets says the same thing: change your optimization event to one that occurs more often.

The corollary is structural. If your total monthly budget is NPR 30,000, you can afford roughly one properly funded ad set, not six. Splitting NPR 30,000 across six ad sets guarantees that all six stay in learning and none of them produce a stable cost figure you can act on.

Campaign structure

Meta Ads Manager offers six campaign objectives — Awareness, Traffic, Engagement, Leads, App promotion and Sales — and you pick one when you create the campaign. The Sales, Leads and App promotion objectives open the Advantage+ campaign setup, where audience, placement and budget automation are on by default; manual settings remain available underneath.

For most businesses in Bharatpur or Kathmandu, the honest objective is not Sales. Sales assumes a website with a working checkout and enough purchase volume to optimize against. If your orders close over Messenger, WhatsApp or Viber — which is how a large share of Nepali retail actually operates — the Engagement objective with a messaging conversion location, or the Leads objective with an instant form, will give the delivery system an event it can find seven times a day.

A structure that survives a small budget:

• One campaign per objective. Not one per product.
• One or two ad sets inside it, with campaign-level budget so spend moves to whichever is working.
• Three to five genuinely different creatives per ad set — different opening frame, different offer, different language — rather than five crops of the same image.
• No edits for the first week. Every significant edit restarts the learning phase and resets the 50-event clock.

Targeting after the privacy changes

This is where most older advice has quietly stopped being true. Advantage+ audience is applied automatically where it is available, and the critical distinction is between controls and suggestions.

Per Meta's documentation on audience controls and audience suggestions, controls limit who can see your ads: locations, minimum age, languages, and custom audiences to exclude. Suggestions — age, gender, detailed targeting, custom audiences to include — guide delivery but do not constrain it. Meta's own example is blunt: if you suggest the gender "women", your ads may still deliver to men if the system finds them likely to respond.

So an interest like a cuisine, a brand or a hobby is now a hint, not a fence. If a particular input genuinely must be binding, choose further limit your audience and uncheck the use as a suggestion box. Two more constraints worth knowing, both set out in Meta's instructions for creating a campaign using Advantage+ audience: it is not possible to exclude lookalike audiences, because exclusion narrows the audience and cuts delivery opportunities; and the minimum age control will not go below 18, since Meta states that when you use Advantage+ audience you cannot select ages under 18, or older in some countries.

What this means practically in Nepal:

• Geography is still hard targeting. Location is a control, so Kathmandu Valley, Chitwan and a diaspora market can be separated reliably. Split them only when the offer, price or delivery promise genuinely differs — not for the sake of granularity.
• Creative does the segmenting. Since interests are suggestions, the copy and the first three seconds of video decide who self-selects. A Nepali-language hook and an English-language hook will pull different audiences from the same ad set.
• Your own data is the strongest input you have. Customer lists, Page and Instagram engagers, and website visitors are the only signals that describe your actual buyers rather than Meta's inference about them.

The Meta Pixel and the Conversions API

The Pixel runs in the browser and the Conversions API sends the same events from your server. The Pixel loses events to ad blockers, browser settings and dropped connections — a real problem on Nepali mobile networks. The server-side path is not affected by any of that. Run both, which in practice means a website with the Pixel and Conversions API installed properly rather than a hosted page builder you cannot add server-side code to.

A server event is a small JSON payload, and Meta's server event parameters are specific about what has to be in it:

• event_name — required. The standard or custom event name, and one half of the deduplication key.
• event_time — required. A Unix timestamp in seconds, sent in GMT, no more than seven days old. One stale timestamp in a batch and Meta returns an error for the whole request.
• user_data — required. The customer identifiers, covered below.
• action_source — required. Where the conversion happened. website, chat, phone_call, physical_store and business_messaging are the values Nepali advertisers reach for most, and chat or business_messaging is the honest answer when the sale closes on Messenger or WhatsApp.
• event_source_url — required for website events, and it should match your verified domain.
• event_id — documented as optional, but it is the field deduplication runs on, so treat it as mandatory. An order number makes a good one.
• custom_data — optional, and where value, currency and contents live. Leave it out and you cannot optimize on value or report revenue.

Sending the same event twice means you must deduplicate. Meta's rules are specific: the event_name must match and the event_id must be identical across the browser and server event; alternatively, matching can occur on event_name plus external_id or fbp. Where a match is found between events sent within 48 hours of each other, Meta considers only the first one — except when the two arrive within about five minutes of each other, in which case it favours the browser event. The full logic is in the help centre article on deduplication for Pixel and Conversions API events. Get this wrong and your conversion counts inflate, your reported ROAS is fiction, and you allocate budget against numbers that are not real.

Then check event match quality, scored 0 to 10 per event from the last 48 hours of data. Meta ranks the customer information parameters by usefulness: email and click ID are high priority; phone number, country, external ID, browser ID, birthdate and Facebook login ID are medium; first name, last name, city and postal code are low. Nepali businesses frequently collect a phone number and no email at all, so phone plus an external ID plus the fbc/fbp cookies are what you will realistically send. Hash those contact fields as the customer information parameters specify — SHA-256 after normalisation, so em trimmed and lowercased, and ph stripped of symbols and leading zeros with the country code included — and make sure you have lawful grounds and consent before you send anything.

Two setup steps you can stop doing

Meta has updated Aggregated Event Measurement for website conversion campaigns. You no longer prioritise eight conversion events per domain, the Aggregated Event Measurement tab has been removed from Events Manager, value sets are not required to use value optimization, and you do not select a conversion domain when creating a campaign. Domain verification may still be needed for other reasons, but not for event configuration. Any tutorial still walking you through the eight-event ranking screen was written for a version of Events Manager that no longer exists.

If you have no website at all, you are not locked out. Offline and CRM events can be sent to a dataset server-side, so a shop that closes every deal on Viber can still feed real outcomes back into optimization instead of optimizing towards clicks.

A realistic first thirty days

1. Week zero. Portfolio created, two-factor authentication on, ad account in USD with the Kathmandu time zone, PAN and business-purpose declaration added, dollar card funded, dataset installed and confirmed firing in Events Manager.
2. Week one. One campaign, one ad set, five distinct creatives, budget set at roughly seven times your target cost per result. Do not touch it.
3. Week two. Read cost per result, not click-through rate. Do not kill anything until the ad set has produced around 50 optimization events, because the numbers before that are noise.
4. Week three. Add the Conversions API, verify deduplication in the Test Events tool, and work on event match quality until the score stops improving.
5. Week four. Duplicate the winning creative angle rather than the winning ad. Add a retargeting ad set only if the source audience is large enough to spend the daily floor.

Frequently asked questions

Can I pay for Facebook ads in Nepal with eSewa, Khalti or connectIPS?

No. Meta's accepted payment options for Nepal are credit cards and co-branded debit cards (Visa, Mastercard, American Express) and PayPal. Nepal has no local manual payment method with Meta, so digital wallets and domestic bank transfers cannot fund an ad account. A foreign-currency card from a Nepali commercial bank is the standard route.

Can I set my ad account currency to Nepali rupees?

No. NPR is not on Meta's accepted currency list for ads, so Nepali advertisers generally run their accounts in US dollars. Plan in rupees internally, then convert. Also remember that changing an existing ad account's currency closes that account and opens a new one, taking its ad history with it.

Does Meta charge VAT on ads in Nepal?

Yes, since August 2023, for customers in Nepal who have not declared a business purpose. If you provide a business-purpose declaration and your PAN in payment settings, you pay VAT under the reverse charge mechanism instead. VAT is added on top of your ad charges, so a single billing can come out higher than your billing threshold.

What is a realistic monthly budget for a small business in Bharatpur or Kathmandu?

Work backwards from the delivery system rather than from comfort. An ad set needs roughly 50 optimization events a week, about seven a day, to leave the learning phase. Multiply seven by what one result costs you. If a messaging conversation costs NPR 60, that is about NPR 420 a day for one ad set. If the resulting figure is unaffordable, choose a cheaper and more frequent optimization event rather than splitting the same money across more ad sets.

Do I still need to rank eight conversion events and verify my domain?

Not for event configuration. Meta removed the eight-event prioritisation requirement and the Aggregated Event Measurement tab in Events Manager, and you no longer select a conversion domain when building a campaign. Domain verification can still be required for other purposes, so check your own case, but it is no longer a prerequisite for setting up web events.

Where this usually goes wrong

The failures we see are rarely creative failures. They are an ad account in the wrong time zone so the client's daily report never matches Ads Manager; a prepaid card that ran dry over Dashain and took the account down with it; VAT charged for two years because nobody added a PAN; six ad sets on NPR 500 each, none of which ever left learning; and a Pixel firing purchases twice because the Conversions API was bolted on without an event_id. All five are setup problems, all five are cheap to fix before launch, and all five are expensive to fix after three months of bad data. Auditing the account, the dataset and the payment method before the first campaign goes live is the cheapest hour in any digital marketing engagement.