This is a repost. Originally published on the Arvio blog: https://arvio.a.xyz/blog/shopify-virtual-assistant. The canonical URL points back there.
Key takeaways
- Page one for this search sells people, not software. Of the ten results we read on 25 August 2026, seven are staffing or outsourcing companies, two are Shopify's own blog, and one is an app vendor's blog. Not one is a tool offering to do the work.
-
Two of those page-one vendors publish their rate on the page. BruntWork's page title reads
$4-$8/hr Skilled Shopify Virtual Assistant; Zenius quotesEntry-level $6.5 - $8 per hourandExperienced $7.75 - $9.25 per hour. So the published band for this job is roughly $4 to $9.25 an hour. - We stopwatched the catalogue half of that job in a live admin. Typing — not writing, just typing — one 147-character product description took 19.6 seconds. Changing one price cell took 5.64 seconds. Both at a machine-steady pace, which is faster than most people type.
- The unit of work is the variant, not the product. Across 43 live storefronts we read separately, the median catalogue runs 2.59 variants per product; our own test store came to 1.76, at the low end of that range. Either way a 500-product repricing is nearer 900 rows than 500, and about 1 hour 23 minutes of pure typing.
- That arithmetic is the opposite of a sales argument. An hour and a half of typing, at those rates, is under $13. At $4 an hour you are buying a second pair of hands. Catching a bad price takes a second pair of eyes, which is a different purchase.
Published 25 August 2026 by Adot Technologies Inc, the team behind Arvio. The timings below are stopwatched inside screen recordings made in a live Shopify admin in August 2026, not estimated. We did not hire a virtual assistant and did not time one — see the method section for exactly what that means for every comparison here.
The quick version
- It is mostly catalogue work. The staffing pages themselves describe it that way: product uploads and edits, collection and tagging logic, discount setup, order status checks, weekly reporting.
- Half of it is keystrokes. Typing descriptions, filling price cells, re-tagging, copying SEO fields. We put a stopwatch on this half and the numbers are below.
- Half of it is judgement, and it does not compress. Deciding what a description should say, whether a discount cannibalises a bestseller, what to tell a customer whose parcel is stuck. Hire for that half. The other half you can automate, or pay someone to type, as long as you know which one you are paying for.
Page one for this query is a staffing market
We read the first ten results for shopify virtual assistant on DuckDuckGo, US locale, on 25 August 2026:
| # | Result | What it sells |
|---|---|---|
| 1 | shopify.com/blog — how to hire virtual assistant services | Shopify's own guide |
| 2 | bruntwork.co | Staffing |
| 3 | upwork.com | Marketplace |
| 4 | shopify.com/blog — how to become a virtual assistant | Shopify's own guide |
| 5 | ecomposer.io | App vendor's blog |
| 6 | atidiv.com | Outsourcing |
| 7 | evirtualassistants.com | Staffing |
| 8 | seodiscovery.com | Agency |
| 9 | ringly.io | Vendor blog |
| 10 | virtualassistanttalent.com | Staffing |
Seven of ten sell you a person. So the answer to "what does a Shopify VA do" is currently being written almost entirely by companies who would like to sell you one. We are not a neutral source either: we make software that does part of this job. Read both kinds of page with that in mind.
What to hand over first, whether or not it is to a person
The split we would use, having now run both halves of the job with a stopwatch going:
| Work | Where it goes | Why |
|---|---|---|
| Filling empty fields across many products | Automate | Pure keystrokes, and the count is always larger than you think |
| Applying one rule to a whole catalogue | Automate, with a preview | The risk is a consistent mistake, not a typo |
| Weekly rollups (bestsellers, out-of-stocks, broken links) | Automate the gathering, keep the reading | Gathering is mechanical; noticing is not |
| Writing what a product is for | Person | It is a judgement about the customer |
| Orders, refunds, exceptions | Person | Policy has edges, and edges are where people are |
| Supplier and 3PL chasing | Person | No fixed shape, lots of interruption |
The job, in the words of the people selling it
Atidiv's guide, the sixth result, publishes a day-to-day list: product uploads and edits (titles, descriptions, tags, variants, images, SEO fields), collection rules and merchandising, inventory and SKU hygiene, discount setup under templates, landing-page publishing support, order operations under policy, apps coordination, and weekly reporting rollups. Quoted from that page as it read on 25 August 2026.
That is a fair list. It is also two different jobs stapled together: one of them is keystrokes, the other is decisions. Nothing on the page separates them. A stopwatch does.
The catalogue half, timed
Every number in this table came off a screen recording in a live Shopify admin, with the stopwatch inside the frame. The description run is published in full: https://www.youtube.com/watch?v=ZMqlmA96aDs — the typing starts around the four-minute mark, and you can count it yourself.
| Task | Unit | Measured | When |
|---|---|---|---|
| Type one product description (147 characters) | one product | 19.6 s | 20 Aug 2026 |
| Change one price in the bulk edit grid | one variant row | 5.64 s | 17 Aug 2026 |
| Variants behind 50 products | — | 88 (1.76 per product) | 17 Aug 2026 |
Extrapolating the first row at the same pace: 50 descriptions is about 16 minutes of typing, 500 is about 2 hours 42 minutes. Hold the 1.76 ratio and a 500-product repricing becomes 880 rows, or about 1 hour 23 minutes.
That ratio is the shakiest number here, so we went and checked it against real stores. 1.76 comes from a catalogue we assembled ourselves, which is exactly the kind of number that quietly does not generalise. Reading the public catalogues of 43 live Shopify storefronts gave a median of 2.59 variants per product — with a range from 1.00 to 30.57, which is the real point. Our own 1.76 sits below that median, so planning from it would have understated the work; the multiplier is nothing like either figure for a store selling in five sizes and four colours. Count your own variants before you plan an afternoon around any of these figures; it is one number in your product list header.
These are typing times, not working times. The 19.6 seconds is the time to type a description that already exists somewhere. It does not include deciding what the product's description should say, which is the larger half and the reason anyone hires a person in the first place.
The pace cuts the other way. The recording types at a constant rate: no pauses, no re-reads, no getting up for coffee. A real person editing a real catalogue is slower than this, not faster. Treat these numbers as a floor.
The count that makes the job look smaller than it is
Select fifty products, hit Bulk edit, and the header will tell you it is editing 10. Scroll, and it fills in to 50. Nothing was dropped — the URL carried all fifty ids the whole time; the grid just loads rows as you reach them.
That is the kind of thing that ends up in a status update as "I did the first ten" when the task was fifty. Catalogue work is full of counts that are not what they look like, and the variant multiplier is the same trap in a different place. You brief 500 products; the person doing it faces 880 cells.
The half that does not compress
Everything below is on that Atidiv list, and none of it is a typing problem:
- Order operations under policy. "Status checks, address updates, cancellations under policy" — the policy is the work. Someone has to decide whether this cancellation is inside it.
- Anything customer-facing that carries tone. The apology for the delayed parcel is not a template.
- Supplier and 3PL chasing. It is a series of interruptions with no fixed shape.
- Judgement about what the copy should say. Which of the two things this product is actually for, whether the size guide is wrong, whether the photography is misleading.
- Deciding the discount, not entering it. Entering it is a form. Deciding whether it cannibalises the bestseller you are already selling out of is not.
If most of what you need is in that list, hire a person, and don't let anyone — including us — tell you a tool replaces it.
What the hourly arithmetic actually proves
Two page-one vendors publish rates, read on 25 August 2026:
- BruntWork's page title:
$4-$8/hr Skilled Shopify Virtual Assistant | Top Shopify VA - Zenius:
Entry-level $6.5 - $8 per hour,Experienced $7.75 - $9.25 per hour
Take our slowest measured job — the 500-product repricing, 1 hour 23 minutes of typing — and put it against that band. It comes to somewhere between about $5.50 and $12.80 of billable time.
So no, we are not going to argue that software is cheaper than typing at those rates. At $4 an hour it usually isn't, and if the only thing standing between you and a finished catalogue is keystrokes, hire the keystrokes.
What the stopwatch does establish is narrower and more useful: this job has a compressible half and an incompressible one, and almost every page on this query prices them as though they were the same thing. An hourly rate buys you both at the same price. If you are one person with no budget, the question is not "what does an assistant cost" but "which half am I actually stuck on" — and the table above is the honest answer to that, whichever way you resolve it. What that comparison leaves out, on both sides, is where the money actually goes.
The typing time is the smallest part of the cost. Writing the brief, agreeing the naming convention, checking the first twenty rows, finding the two that went in wrong three weeks later. None of that is billed at $4 an hour, and all of it is your time.
Cost is not on the screen unless someone puts it there. Cost per item is not one of the columns the bulk editor opens with, and it was never on screen in anything we timed. So a person working through 880 cells at 5.64 seconds each is not weighing margin. They are not being careless. They are looking at a view that does not carry the number.
And at $4 to $9.25 an hour, what you are buying is execution, which is what those rates are for. The mistakes that hurt in a catalogue are rarely typos anyway. They are one decision, applied consistently, to 880 rows.
The same two jobs, run by an agent, on the same store
We ran both tasks a second time on the same store, as prompts, and stopwatched those too.
Descriptions. The store had 69 products, of which 23 had no description at all. Asked to draft one for each and show them before writing anything, it produced all 23 drafts in 322.69 seconds — 5 minutes 23 seconds. We independently re-read the store afterwards: still 23 empty descriptions, so nothing had been written.
Typing 23 descriptions that someone else has already written takes about 7 minutes 30 seconds at our measured pace. Drafting 23 from scratch took 5 minutes 23 seconds. On wall clock alone that is close to a wash, and anyone selling you the drafting speed is selling you the wrong number. The difference is which 5 minutes: one has you at the keyboard producing nothing but keystrokes, the other has 23 pieces of copy waiting for you to approve or reject.
Prices. Same store, Increase the price of every product by 10%. End to end: 5 minutes 46 seconds, of which 2 minutes 17 seconds was spent producing a preview to approve before anything was written. By hand, the same 88 rows is about 8 minutes 20 seconds of continuous typing.
Those 2 minutes 17 seconds are the preview, and on this run it earned them. It flagged that three variants — three sizes of one basket — would rise by more than the 10% asked for, because those prices were already out of order — a flat 10% preserves an inversion rather than fixing it, and would have left the bigger basket cheaper than the smaller one. It also reported margin verified on 65 of 153 variants, and 88 variants with no cost data on file at all. That is the same blind spot the grid has, except stated out loud instead of silently absent.
We have written both of those runs up in full, with what we could not verify in each: bulk editing descriptions and bulk price changes.
Method
What we measured. Two tasks on one live Shopify store: typing a product description in the bulk edit grid, and changing prices in the same grid. Both are screen recordings at 1920×1080, 30 fps, with the stopwatch running inside the frame. Description timing 20 August 2026; price timing 17 August 2026; agent runs 20 and 22 August 2026.
Exclusion rule. The typing timings exclude everything except keystrokes — no time to decide the content, no navigation, no saving. We did this deliberately, because it produces the most conservative possible number for the manual route. Adding the excluded time makes the manual side larger, not smaller.
Sample. One store, one catalogue, single runs. Fifty products and 88 variants for the price grid, 69 products for the description task. This is not a benchmark, and we are not claiming it generalises to a catalogue with three option dimensions or per-market price lists — we have not tested one.
The SERP. Ten results, DuckDuckGo HTML endpoint, us-en locale, 25 August 2026, no personalisation and no logged-in session. Rankings differ by engine, country and day. Upwork returned 403 to our fetch, so its listing page is classified from the search result alone, not from reading it.
Rates. Read from the two page-one vendor pages that publish them, on 25 August 2026, as they appeared to an anonymous visitor. Quoted verbatim. We did not enquire, negotiate, or hire, and published rates are not quotes.
Limits, in one place
- We never timed an actual virtual assistant. Every manual number here is us doing the work at a machine-steady pace. It is a floor for a human, not a measurement of one. If you have real timings from a real VA on a real catalogue, they are better evidence than ours and we would rather see them.
- The agent's own count of 69 products and 153 variants. Our independent read covered 50 products and 61 variants — the product list pages at 50 and the grid is virtualised, so a scripted read does not see the rest. Consistent with the store; not verified.
-
"Margin verified 65/153" and "88 variants with no cost data".
Cost per itemwas never on screen, so we never read cost data at all. We are reporting what the preview said, not confirming it. - Whether the rates on those two pages are what anyone actually pays. We read published prices, nothing more. Five of the ten page-one results publish no rate at all.
- Anything about quality. Not the quality of a VA's descriptions, and not the quality of the drafted ones. We timed both. We did not judge either.
Where the crossover is. On 23 descriptions the manual and drafted routes are within about two minutes of each other, and the difference is who is at the keyboard rather than how long it takes. Nobody has run the same measurement at 200 or 2,000 products, us included, and the two curves almost certainly do not stay parallel. Until someone does, treat any confident claim about "10x faster" — ours or anyone's — as a projection.
FAQ
What does a Shopify virtual assistant do?
Mostly catalogue and order operations: product uploads and edits, tagging and collection rules, discount setup, order status checks and address changes, weekly reporting. The staffing pages selling the service describe it the same way; one of their published lists is quoted in full above.
How much does a Shopify virtual assistant cost?
Two of the page-one vendors publish rates. BruntWork's page title says $4-$8/hr; Zenius lists Entry-level $6.5 - $8 per hour and Experienced $7.75 - $9.25 per hour. Both read 25 August 2026. The other page-one vendors do not publish a rate, and published rates are not quotes.
Is a virtual assistant or an app better for bulk product edits?
For pure field-filling across many products, the honest answer is that either works and the app is faster per unit. The reason to prefer one over the other is not speed, it is review: Cost per item is not one of the columns the bulk grid opens with, so unless someone puts it there, neither the person typing nor you checking their work is looking at margin.
How long does it take to update 500 product prices by hand?
About 1 hour 23 minutes of pure typing on our measurement, assuming 1.76 variants per product, at a pace faster than most people type. Deciding what the new prices should be is not included and is the larger half.
What can a Shopify virtual assistant not do?
Anything where the work is the decision rather than the entry: whether a cancellation is inside policy, what tone the apology takes, whether a discount cannibalises a product you are already selling out of. Those do not get faster with more hands, and they do not get automated either.
How many hours a week should I start with?
Size it from the catalogue rather than a round number. Count your variants, not your products — ours ran 1.76 variants per product — and multiply by the per-cell timings above to get the keystroke floor. Then decide how much judgement work sits on top, because that is the part that actually fills a week.
Can I give a virtual assistant access to my Shopify admin?
Yes, through a staff account with scoped permissions rather than your own login. What to scope it to depends on the task list you hired for; the principle is that catalogue edits and financial settings are different permissions and should stay that way.
Does hiring a virtual assistant help SEO?
It gets the fields filled, which is a prerequisite, not a result. We are not going to claim a ranking outcome for either a person or a tool — filling 400 empty meta descriptions is work that has to happen before anything else can, and that is all it is.
Related: How to bulk edit product descriptions in Shopify — the description run above, in full, including the two CSV import errors nothing documents.
Related: How to change prices in bulk in Shopify — the price run above, in full, including the three variants the preview caught.
Written by Adot Technologies Inc, the team behind Arvio: AI Store Operator — describe the catalogue job in plain language, review every edit before it goes live, and undo anything.
Originally published at https://arvio.a.xyz/blog/shopify-virtual-assistant. More Shopify bulk-editing writeups are on the Arvio blog.
Top comments (0)